Dr. Lex Fridman: Machines, Creativity & Love

Time: 0.125

[bright music]

Time: 0.958

- Welcome to the "Huberman Lab Podcast,"

Time: 2.33

where we discuss science and science-based tools

Time: 4.93

for everyday life.

Time: 9.35

I'm Andrew Huberman,

Time: 10.46

and I'm a Professor of Neurobiology and Ophthalmology

Time: 13.27

at Stanford School of Medicine.

Time: 15.06

Today I have the pleasure of introducing Dr. Lex Fridman

Time: 17.95

as our guest on the "Huberman Lab Podcast."

Time: 21.15

Dr. Fridman is a researcher at MIT specializing

Time: 23.92

in machine learning,

Time: 25.16

artificial intelligence and human robot interactions.

Time: 29.36

I must say that the conversation with Lex

Time: 31.85

was without question,

Time: 33.83

one of the most fascinating conversations

Time: 35.7

that I've ever had,

Time: 36.64

not just in my career, but in my lifetime.

Time: 39.131

I knew that Lex worked on these topics.

Time: 41.55

And I think many of you are probably familiar with Lex

Time: 43.95

and his interest in these topics

Time: 45.24

from his incredible podcast, the "Lex Fridman Podcast."

Time: 48.29

If you're not already watching that podcast,

Time: 50.28

please subscribe to it.

Time: 51.18

It is absolutely fantastic.

Time: 53.76

But in holding this conversation with Lex,

Time: 55.5

I realized something far more important.

Time: 58.327

He revealed to us a bit of his dream.

Time: 60.95

His dream about humans and robots,

Time: 63.4

about humans and machines,

Time: 65.02

and about how those interactions

Time: 66.63

can change the way that we perceive ourselves

Time: 68.92

and that we interact with the world.

Time: 70.8

We discuss relationships of all kinds,

Time: 73.1

relationships with animals,

Time: 74.75

relationships with friends,

Time: 76.47

relationships with family and romantic relationships.

Time: 80.46

And we discuss relationships with the machines.

Time: 83.23

Machines that move and machines that don't move,

Time: 86.224

and machines that come to understand us in ways

Time: 89.02

that we could never understand for ourselves,

Time: 91.73

and how those machines can educate us about ourselves.

Time: 95.77

Before this conversation,

Time: 96.94

I had no concept of the ways

Time: 99.42

in which machines could inform me

Time: 101.38

or anyone about themselves.

Time: 103.74

By the end,

Time: 104.62

I was absolutely taken with the idea,

Time: 106.82

and I'm still taken with the idea

Time: 108.58

that interactions with machines

Time: 109.98

have a very particular kind,

Time: 111.97

a kind that Lex understands and wants to bring to the world,

Time: 115.21

can not only transform the self,

Time: 116.972

but may very well transform humanity.

Time: 120.14

So whether or not you're familiar

Time: 121.39

with Dr. Lex Fridman or not,

Time: 123.35

I'm certain you're going to learn

Time: 124.42

a tremendous amount from him

Time: 125.89

during the course of our discussion,

Time: 127.55

and that it will transform

Time: 128.73

the way that you think about yourself and about the world.

Time: 132.09

Before we begin,

Time: 133.21

I want to mention that this podcast

Time: 134.69

is separate from my teaching and research roles at Stanford.

Time: 137.58

It is however part of my desire and effort

Time: 139.632

to bring zero cost to consumer information about science

Time: 142.77

and science-related tools to the general public.

Time: 145.6

In keeping with that theme,

Time: 146.75

I'd like to thank the sponsors of today's podcast.

Time: 149.63

Our first sponsor is ROKA.

Time: 151.331

ROKA makes sunglasses and eyeglasses

Time: 153.38

that are of absolutely phenomenal quality.

Time: 155.67

The company was founded

Time: 156.53

by two All-American swimmers from Stanford,

Time: 158.53

and everything about the sunglasses

Time: 160.2

and eyeglasses they've designed had performance in mind.

Time: 163.95

I've spent a career working on the visual system.

Time: 166.26

And one of the fundamental issues

Time: 167.73

that your visual system has to deal with

Time: 169.74

is how to adjust what you see

Time: 171.96

when it gets darker or brighter in your environment.

Time: 174.64

With ROKA Sunglasses and Eyeglasses,

Time: 176.86

whether or not it's dim in the room or outside,

Time: 179.14

or not there's cloud cover,

Time: 180.19

or whether or not you walk into a shadow,

Time: 181.59

you can always see the world with absolute clarity.

Time: 184.32

And that just tells me that they really understand

Time: 186.26

the way that the visual system works.

Time: 187.65

Processes like habituation and attenuation.

Time: 190.18

All these things that work at a real mechanistic level

Time: 192.56

have been built into these glasses.

Time: 194.59

In addition, the glasses are very lightweight.

Time: 196.79

You don't even notice really that they're on your face.

Time: 199.11

And the quality of the lenses is terrific.

Time: 201.8

Now, the glasses were also designed

Time: 203.5

so that you could use them,

Time: 204.41

not just while working or at dinner, et cetera,

Time: 206.86

but while exercising.

Time: 208.59

They don't fall off your face or slip off your face

Time: 210.62

if you're sweating.

Time: 211.81

And as I mentioned,

Time: 212.643

they're extremely lightweight.

Time: 213.476

So you can use them while running,

Time: 214.86

you can use them while cycling and so forth.

Time: 217.07

Also the aesthetic of ROKA glasses is terrific.

Time: 219.69

Unlike a lot of performance glasses out there,

Time: 221.69

which frankly make people look like cyborgs,

Time: 224.286

these glasses look great.

Time: 226.03

You can wear them out to dinner,

Time: 227.21

you can wear them for essentially any occasion.

Time: 230.48

If you'd like to try ROKA glasses,

Time: 232.01

you can go to roka.com.

Time: 233.66

That's R-O-K-A .com and enter the code Huberman

Time: 236.9

to save 20% off your first order.

Time: 239.2

That's ROKA,

Time: 240.033

R-O-K-A .com and enter the code Huberman at checkout.

Time: 243.54

Today's episode is also brought to us by InsideTracker.

Time: 246.899

InsideTracker is a personalized nutrition platform

Time: 249.43

that analyzes data from your blood and DNA

Time: 252.01

to help you better understand your body

Time: 253.61

and help you reach your health goals.

Time: 255.74

I'm a big believer in getting regular blood work done

Time: 258.25

for the simple reason that many of the factors

Time: 260.74

that impact our immediate and long-term health

Time: 263.21

can only be assessed from a quality blood test.

Time: 265.87

And now with the advent of quality DNA tests,

Time: 268.5

we can also get insight

Time: 269.386

into some of our genetic underpinnings

Time: 271.78

of our current and long-term health.

Time: 274.019

The problem with a lot of blood

Time: 275.61

and DNA tests out there, however,

Time: 277.47

is you get the data back and you don't know what to do

Time: 279.41

with those data.

Time: 280.51

You see that certain things are high

Time: 281.81

or certain things are low,

Time: 283.03

but you really don't know what the actionable items are,

Time: 285.24

what to do with all that information.

Time: 287.33

With InsideTracker,

Time: 288.69

they make it very easy to act in the appropriate ways

Time: 291.69

on the information that you get back

Time: 293.29

from those blood and DNA tests.

Time: 295.11

And that's through the use of their online platform.

Time: 297.63

They have a really easy to use dashboard

Time: 299.91

that tells you what sorts of things can bring the numbers

Time: 302.529

for your metabolic factors, endocrine factors, et cetera,

Time: 305.77

into the ranges that you want and need

Time: 308.07

for immediate and long-term health.

Time: 310.19

In fact, I know one individual just by way of example,

Time: 313.13

that was feeling good,

Time: 314.27

but decided to go with an InsideTracker test

Time: 316.5

and discovered that they had high levels

Time: 318

of what's called C-reactive protein.

Time: 319.85

They would have never detected that otherwise.

Time: 321.85

C-reactive protein is associated

Time: 323.39

with a number of deleterious health conditions,

Time: 325.97

some heart issues, eye issues, et cetera.

Time: 328.24

And so they were able to take immediate action

Time: 330.22

to try and resolve those CRP levels.

Time: 333.42

And so with InsideTracker,

Time: 334.66

you get that sort of insight.

Time: 336.02

And as I mentioned before,

Time: 337.06

without a blood or DNA test,

Time: 338.48

there's no way you're going to get that sort of insight

Time: 340.41

until symptoms start to show up.

Time: 342.86

If you'd like to try InsideTracker,

Time: 344.35

you can go to insidetracker.com/huberman

Time: 347.4

to get 25% off any of InsideTracker's plans.

Time: 350.29

You just use the code Huberman at checkout.

Time: 352.72

That's insidetracker.com/huberman

Time: 355.45

to get 25% off any of InsideTracker's plans.

Time: 358.73

Today's podcast is brought to us by Athletic Greens.

Time: 361.57

Athletic Greens is an all-in-one

Time: 363.24

vitamin mineral probiotic drink.

Time: 365.68

I started taking Athletic Greens way back in 2012.

Time: 369.05

And so I'm delighted that they're sponsoring the podcast.

Time: 371.91

The reason I started taking Athletic Greens

Time: 373.76

and the reason I still take Athletic Greens

Time: 375.84

is that it covers all of my vitamin mineral probiotic basis.

Time: 379.47

In fact, when people ask me, what should I take?

Time: 382.05

I always suggest that the first supplement people take

Time: 384.49

is Athletic Greens,

Time: 385.65

for the simple reason,

Time: 386.96

is that the things that contains covers your bases

Time: 389.52

for metabolic health, endocrine health,

Time: 391.92

and all sorts of other systems in the body.

Time: 394.03

And the inclusion of probiotics

Time: 395.73

are essential for a healthy gut microbiome.

Time: 398.617

There are now tons of data showing

Time: 400.57

that we have neurons in our gut,

Time: 402.85

and keeping those neurons healthy requires

Time: 404.8

that they are exposed

Time: 405.66

to what are called the correct microbiota,

Time: 407.98

little microorganisms that live in our gut

Time: 409.93

and keep us healthy.

Time: 411.15

And those neurons in turn help keep our brain healthy.

Time: 413.9

They influence things like mood, our ability to focus,

Time: 416.55

and many, many other factors related to health.

Time: 419.93

With Athletic Greens, it's terrific,

Time: 421.44

because it also tastes really good.

Time: 423.2

I drink it once or twice a day.

Time: 424.74

I mix mine with water and I add a little lemon juice,

Time: 427.08

or sometimes a little bit of lime juice.

Time: 429.43

If you want to try athletic greens,

Time: 431.13

you can go to athleticgreens.com/huberman.

Time: 434.13

And if you do that, you can claim their special offer.

Time: 436.71

They're giving away five free travel packs,

Time: 438.7

little packs that make it easy to mix up Athletic Greens

Time: 441.08

while you're on the road.

Time: 442.228

And they'll give you a year supply

Time: 444.28

of vitamin D3 and K2.

Time: 446.31

Again, go to athleticgreens.com/huberman

Time: 448.97

to claim that special offer.

Time: 450.93

And now, my conversation with Dr. Lex Fridman.

Time: 454.64

- We meet again.

Time: 455.58

- We meet again.

Time: 457.46

Thanks so much for sitting down with me.

Time: 459.65

I have a question that I think is on a lot of people's minds

Time: 463.001

or ought to be on a lot of people's minds,

Time: 466.44

because we hear these terms a lot these days,

Time: 470.09

but I think most people,

Time: 471.36

including most scientists and including me

Time: 475.03

don't know really what is artificial intelligence,

Time: 480.09

and how is it different

Time: 481.377

from things like machine learning and robotics?

Time: 485.29

So, if you would be so kind as to explain to us,

Time: 488.96

what is artificial intelligence,

Time: 491.61

and what is machine learning?

Time: 494.04

- Well, I think that question is as complicated

Time: 497.14

and as fascinating as the question of, what is intelligence?

Time: 501.81

So, I think of artificial intelligence,

Time: 505.65

first, as a big philosophical thing.

Time: 508.78

Pamela McCormick said AI was the ancient wish

Time: 517.21

to forge the gods,

Time: 518.34

or was born as an ancient wish to forge the gods.

Time: 521.76

So I think at the big philosophical level,

Time: 524.29

it's our longing to create other intelligence systems.

Time: 528.34

Perhaps systems more powerful than us.

Time: 531.77

At the more narrow level,

Time: 534.21

I think it's also set of tools

Time: 537.04

that are computational mathematical tools

Time: 539.29

to automate different tasks.

Time: 541.16

And then also it's our attempt to understand our own mind.

Time: 545.66

So, build systems that exhibit some intelligent behavior

Time: 549.96

in order to understand what is intelligence

Time: 552.412

in our own selves.

Time: 554.63

So all of those things are true.

Time: 556.21

Of course, what AI really means as a community,

Time: 559.36

as a set of researchers and engineers,

Time: 561.54

it's a set of tools,

Time: 562.68

a set of computational techniques

Time: 565.33

that allow you to solve various problems.

Time: 568.638

There's a long history that approaches the problem

Time: 573.05

from different perspectives.

Time: 574.25

What's always been throughout one of the threads,

Time: 577.69

one of the communities goes under the flag

Time: 580.31

of machine learning,

Time: 581.25

which is emphasizing in the AI space,

Time: 585.688

the task of learning.

Time: 588.13

How do you make a machine

Time: 589.56

that knows very little in the beginning,

Time: 591.2

follow some kind of process

Time: 593.479

and learns to become better and better at a particular task?

Time: 597.71

What's been most very effective in the recent about 15 years

Time: 603.7

is a set of techniques that fall under the flag

Time: 605.73

of deep learning that utilize neural networks.

Time: 608.7

When your networks are these fascinating things inspired

Time: 612.97

by the structure of the human brain,

Time: 616.32

very loosely, but they have a,

Time: 618.46

it's a network of these little basic computational units

Time: 621.6

called neurons, artificial neurons.

Time: 624.048

And they have,

Time: 625.82

these architectures have an input and output.

Time: 628.76

They know nothing in the beginning,

Time: 630.19

and their task with learning something interesting.

Time: 633.25

What that's something interesting is,

Time: 635.27

usually involves a particular task.

Time: 638.26

There's a lot of ways to talk about this

Time: 641.2

and break this down.

Time: 642.033

Like one of them is how much human supervision

Time: 645.92

is required to teach this thing.

Time: 648.37

So supervised learning is broad category,

Time: 651.89

is the neural network knows nothing in the beginning

Time: 656.21

and then it's given a bunch of examples in computer vision

Time: 660.42

that will be examples of cats, dogs, cars, traffic signs,

Time: 664.057

and then you're given the image

Time: 666.15

and you're given the ground truth of what's in that image.

Time: 669.59

And when you get a large database of such image examples

Time: 673.07

where you know the truth,

Time: 675.302

the neural network is able to learn by example,

Time: 678.66

that's called supervised learning.

Time: 681.114

There's a lot of fascinating questions within that,

Time: 683.88

which is, how do you provide the truth?

Time: 686.07

When you given an image of a cat,

Time: 689.916

how do you provide to the computer

Time: 692.96

that this image contains a cat?

Time: 694.96

Do you just say the entire image is a picture of a cat?

Time: 698.01

Do you do what's very commonly been done,

Time: 700.37

which is a bounding box,

Time: 701.57

you have a very crude box around the cat's face saying,

Time: 705.43

this is a cat?

Time: 706.55

Do you do semantic segmentation?

Time: 708.77

Mind you, this is a 2D image of a cat.

Time: 711.05

So it's not,

Time: 713

the computer knows nothing

Time: 713.98

about our three-dimensional world,

Time: 715.7

is just looking at a set of pixels.

Time: 717.37

So, semantic segmentation is drawing a nice,

Time: 721.18

very crisp outline around the cat and saying, that's a cat.

Time: 724.96

That's really difficult to provide that truth.

Time: 727.11

And one of the fundamental open questions

Time: 729.55

in computer vision is,

Time: 730.83

is that even a good representation of the truth?

Time: 733.84

Now, there's another contrasting set of ideas

Time: 738.6

that our attention they're overlapping is,

Time: 741.58

well, it's used to be called unsupervised learning.

Time: 744.37

What's commonly now called self-supervised learning.

Time: 747.19

Which is trying to get less and less

Time: 749.46

and less human supervision into the task.

Time: 754.01

So self-supervised learning is more,

Time: 758.966

it's been very successful in the domain of language model,

Time: 762.04

natural English processing,

Time: 763.22

and now more and more as being successful

Time: 765

in computer vision task.

Time: 766.51

And the idea there is,

Time: 768.9

let the machine without any ground-truth annotation

Time: 773.18

just look at pictures on the internet,

Time: 775.77

or look at texts on the internet

Time: 777.63

and try to learn something generalizable

Time: 782.36

about the ideas that are at the core of language

Time: 785.84

or at the core of vision.

Time: 787.23

And based on that,

Time: 789.61

we humans at its best like to call that common sense.

Time: 792.94

So with this,

Time: 793.773

we have this giant base of knowledge

Time: 796.03

on top of which we build more sophisticated knowledge.

Time: 798.75

We have this kind of commonsense knowledge.

Time: 801.57

And so the idea with self-supervised learning

Time: 803.42

is to build this commonsense knowledge about,

Time: 807.32

what are the fundamental visual ideas

Time: 810.47

that make up a cat and a dog

Time: 812.05

and all those kinds of things

Time: 813.31

without ever having human supervision?

Time: 815.81

The dream there is the,

Time: 818.471

you just let an AI system that's self supervised

Time: 823.35

run around the internet for awhile,

Time: 824.495

watch YouTube videos for millions and millions of hours,

Time: 827.5

and without any supervision be primed

Time: 831.42

and ready to actually learn with very few examples

Time: 834.68

once the human is able to show up.

Time: 836.72

We think of children in this way,

Time: 839.3

human children,

Time: 840.18

is your parents only give one or two examples

Time: 843.08

to teach a concept.

Time: 844.325

The dream with self-supervised learning

Time: 847.07

is that will be the same with machines.

Time: 850.12

That they would watch millions of hours of YouTube videos,

Time: 854.01

and then come to a human and be able to understand

Time: 856.82

when the human shows them, this is a cat.

Time: 859.4

Like, remember this' a cat.

Time: 860.83

They will understand that a cat

Time: 862.74

is not just the thing with pointy ears,

Time: 864.98

or a cat is a thing that's orange, or is furry,

Time: 868.83

they'll see something more fundamental

Time: 870.85

that we humans might not actually be able

Time: 872.75

to introspect and understand.

Time: 873.94

Like, if I asked you,

Time: 875.09

what makes a cat versus a dog,

Time: 876.83

you wouldn't probably not be able to answer that,

Time: 879.44

but if I showed you, brought to you a cat and a dog,

Time: 882.79

you'll be able to tell the difference.

Time: 884.43

What are the ideas that your brain uses

Time: 887.09

to make that difference?

Time: 888.84

That's the whole dream with self-supervised learning,

Time: 891.31

is it would be able to learn that on its own.

Time: 893.93

That set of commonsense knowledge,

Time: 896.12

that's able to tell the difference.

Time: 897.91

And then there's like a lot of incredible uses

Time: 901.64

of self-supervised learning,

Time: 904.04

very weirdly called self-play mechanism.

Time: 907.34

That's the mechanism behind

Time: 908.376

the reinforcement learning successes

Time: 912.41

of the systems that won at Go,

Time: 915.76

at, AlphaZero that won a chess.

Time: 918.89

- Oh, I see.

Time: 919.723

That play games?

Time: 920.87

- [Lex] That play games.

Time: 921.703

- Got it.

Time: 922.536

- So the idea of self-play is probably,

Time: 924.86

applies to other domains than just games.

Time: 927.92

Is a system that just plays against itself.

Time: 930.9

And this is fascinating in all kinds of domains,

Time: 933.62

but it knows nothing in the beginning.

Time: 936.64

And the whole idea is it creates

Time: 938.6

a bunch of mutations of itself

Time: 941.09

and plays against those versions of itself.

Time: 946.67

And the fascinating thing is when you play against systems

Time: 950.27

that are a little bit better than you,

Time: 951.85

you start to get better yourself.

Time: 953.75

Like learning,

Time: 954.583

that's how learning happens.

Time: 956.15

That's true for martial arts.

Time: 957.33

It's true in a lot of cases.

Time: 959.29

Where you want to be interacting with systems

Time: 962.08

that are just a little better than you.

Time: 963.93

And then through this process of interacting with systems

Time: 966.84

just a little better than you,

Time: 968.17

you start following this process

Time: 969.98

where everybody starts getting better and better and better

Time: 971.98

and better until you are several orders of magnitude

Time: 975.26

better than the world champion in chess, for example.

Time: 977.947

And it's fascinating because it's like a runaway system.

Time: 981.09

One of the most terrifying and exciting things

Time: 983.59

that David Silver, the creator of AlphaGo and AlphaZero,

Time: 987.24

one of the leaders of the team said,

Time: 990.38

to me is a,

Time: 991.98

they haven't found the ceiling for AlphaZero.

Time: 996.67

Meaning it could just arbitrarily keep improving.

Time: 999.4

Now, in the realm of chess,

Time: 1000.87

that doesn't matter to us.

Time: 1001.9

That it's like,

Time: 1003.03

it just ran away with the game of chess.

Time: 1005.02

Like it's like just so much better than humans.

Time: 1007.935

But the question is what,

Time: 1009.47

if you can create that in the realm that does have a bigger,

Time: 1014.33

deeper effect on human beings and societies,

Time: 1017.91

that can be a terrifying process.

Time: 1020

To me, it's an exciting process

Time: 1021.82

if you supervise it correctly,

Time: 1023.75

if you inject, if what's called value alignment,

Time: 1029.83

you make sure that the goals that the AI is optimizing

Time: 1033.53

is aligned with human beings and human societies.

Time: 1037.03

There's a lot of fascinating things to talk about

Time: 1039.26

within the specifics of neural networks

Time: 1043.24

and all the problems that people are working on.

Time: 1045.62

But I would say the really big,

Time: 1047.58

exciting one is self-supervised learning.

Time: 1049.66

We're trying to get less and less human supervision,

Time: 1055.57

less and less human supervision of neural networks.

Time: 1058.78

And also just a comment and I'll shut up.

Time: 1062.01

- No, please keep going.

Time: 1063.13

I'm learning.

Time: 1064.39

I have questions, but I'm learning.

Time: 1065.52

So please keep going.

Time: 1066.51

- So, to me what's exciting is not the theory,

Time: 1069.47

it's always the application.

Time: 1070.818

One of the most exciting applications

Time: 1072.96

of artificial intelligence,

Time: 1075.14

specifically neural networks and machine learning

Time: 1077.51

is Tesla Autopilot.

Time: 1079.17

So these are systems that are working in the real world.

Time: 1081.62

This isn't an academic exercise.

Time: 1083.65

This is human lives at stake.

Time: 1085.25

This is safety-critical.

Time: 1087.25

- These are automated vehicles.

Time: 1088.577

Autonomous vehicles. - Semi-autonomous.

Time: 1090.43

We want to be.

Time: 1091.3

- Okay.

Time: 1092.37

- We've gone through wars on these topics,

Time: 1095

- Semi-autonomous vehicles.

Time: 1096.31

- Semi-autonomous.

Time: 1097.143

So, even though it's called a FSD, Full Self-Driving,

Time: 1102.59

it is currently not fully autonomous,

Time: 1105

meaning human supervision is required.

Time: 1107.78

So, human is tasked with overseeing the systems.

Time: 1110.94

In fact, liability-wise, the human is always responsible.

Time: 1115.25

This is a human factor psychology question,

Time: 1117.99

which is fascinating.

Time: 1119.41

I'm fascinated by the whole space,

Time: 1123.03

which is a whole 'nother space of human robot interaction

Time: 1126.2

when AI systems and humans work together

Time: 1128.8

to accomplish tasks.

Time: 1130.01

That dance to me is one of the smaller communities,

Time: 1135.44

but I think it will be one of the most

Time: 1137.63

important open problems once they're solved,

Time: 1140.37

is how the humans and robots dance together.

Time: 1143.561

To me, semi-autonomous driving is one of those spaces.

Time: 1147.573

So for Elon, for example, he doesn't see it that way,

Time: 1151.72

he sees semi-autonomous driving as a stepping stone

Time: 1156.55

towards fully autonomous driving.

Time: 1158.66

Like, humans and robots can't dance well together.

Time: 1162.76

Like, humans and humans dance and robots and robots dance.

Time: 1165.37

Like, we need to,

Time: 1166.461

this is an engineering problem,

Time: 1168.1

we need to design a perfect robot that solves this problem.

Time: 1171.73

To me forever,

Time: 1172.78

maybe this is not the case with driving,

Time: 1174.17

but the world is going to be full of problems

Time: 1177.17

with always humans and robots have to interact,

Time: 1180.43

because I think robots will always be flawed,

Time: 1183.55

just like humans are going to be flawed, are flawed.

Time: 1187.169

And that's what makes life beautiful,

Time: 1190.04

that they're flawed.

Time: 1191.04

That's where learning happens

Time: 1192.4

at the edge of your capabilities.

Time: 1195.89

So you always have to figure out,

Time: 1197.56

how can flawed robots and flawed humans interact together

Time: 1203.522

such that they, like the sum is bigger than the whole,

Time: 1208.4

as opposed to focusing on just building the perfect robot?

Time: 1211.58

- Mm-hmm.

Time: 1212.59

- So that's one of the most exciting applications

Time: 1215.37

I would say of artificial intelligence to me

Time: 1217.83

is autonomous driving, the semi-autonomous driving.

Time: 1220.69

And that's a really good example of machine learning

Time: 1223.29

because those systems are constantly learning.

Time: 1226.8

And there's a process there that maybe I can comment on,

Time: 1231.28

the, Andrej Karpathy who's the head of autopilot

Time: 1233.957

calls it the data engine.

Time: 1236.098

And this process applies for a lot of machine learning,

Time: 1238.94

which is you build a system that's pretty good

Time: 1240.94

at doing stuff,

Time: 1241.851

you send it out into the real world,

Time: 1245.38

it starts doing the stuff

Time: 1246.87

and then it runs into what are called edge cases,

Time: 1249.23

like failure cases,

Time: 1250.7

where it screws up.

Time: 1252.491

We do this as kids.

Time: 1253.99

That you have-

Time: 1254.997

- You do this as adults.

Time: 1256.38

- We do this as adults.

Time: 1257.358

Exactly.

Time: 1258.62

But we learn really quickly.

Time: 1260.17

But the whole point,

Time: 1261.38

and this is the fascinating thing about driving,

Time: 1263.309

is you realize there's millions of edge cases.

Time: 1267

There's just like weird situations that you did not expect.

Time: 1271.02

And so the data engine process

Time: 1273.12

is you collect those edge cases,

Time: 1275.15

and then you go back to the drawing board

Time: 1277.11

and learn from them.

Time: 1278.047

And so you have to create this data pipeline

Time: 1281.02

where all these cars,

Time: 1282.69

hundreds of thousands of cars are driving around

Time: 1285.38

and something weird happens.

Time: 1287.08

And so whenever this weird detector fires,

Time: 1290.94

it's another important concept,

Time: 1293.43

that piece of data goes back to the mothership

Time: 1297.43

for the training,

Time: 1299.36

for the retraining of the system.

Time: 1300.88

And through this data engine process,

Time: 1302.68

it keeps improving and getting better and better

Time: 1304.423

and better and better.

Time: 1305.91

So basically you send out

Time: 1307.65

a pretty clever AI systems out into the world

Time: 1310.56

and let it find the edge cases,

Time: 1314.46

let it screw up just enough to figure out

Time: 1317.24

where the edge cases are,

Time: 1318.57

and then go back and learn from them,

Time: 1320.84

and then send out that new version

Time: 1322.67

and keep updating that version.

Time: 1324.28

- Is the updating done by humans?

Time: 1326.34

- The annotation is done by humans.

Time: 1328.98

The, so you have to,

Time: 1331.07

the weird examples come back,

Time: 1333.47

the edge cases,

Time: 1334.83

and you have to label what actually happened in there.

Time: 1337.73

There's also some mechanisms for automatically labeling,

Time: 1343.18

but mostly,

Time: 1344.23

I think you always have to rely on humans to improve,

Time: 1346.79

to understand what's happening in the weird cases.

Time: 1350.04

And then there's a lot of debate.

Time: 1351.79

And this, the other thing,

Time: 1352.69

what is artificial intelligence?

Time: 1354.5

Which is a bunch of smart people

Time: 1356.85

having very different opinions about what is intelligence.

Time: 1359.78

So AI is basically a community of people

Time: 1361.92

who don't agree on anything.

Time: 1363.178

- And it seems to be the case.

Time: 1365.97

First of all,

Time: 1366.93

this is a beautiful description of terms

Time: 1368.7

that I've heard many times among my colleagues at Stanford,

Time: 1371.94

at meetings in the outside world.

Time: 1373.8

And there are so many fascinating things.

Time: 1375.99

I have so many questions,

Time: 1376.98

but I do want to ask one question about the culture of AI,

Time: 1380.25

because it does seem to be a community

Time: 1381.76

where at least as an outsider,

Time: 1384

where it seems like there's very little consensus

Time: 1386.18

about what the terms

Time: 1387.33

and the operational definitions even mean.

Time: 1389.39

And there seems to be a lot of splitting happening now

Time: 1392.24

of not just supervised and unsupervised learning,

Time: 1394.93

but these sort of intermediate conditions

Time: 1398.11

where machines are autonomous,

Time: 1400.82

but then go back for more instruction

Time: 1402.15

like kids go home from college during the summer

Time: 1404.05

and get a little,

Time: 1405.26

moms still feeds them

Time: 1406.27

then eventually they leave the nest kind of thing.

Time: 1409.8

Is there something in particular about engineers,

Time: 1412.67

or about people in this realm of engineering

Time: 1415.86

that you think lends itself to disagreement?

Time: 1419.12

- Yeah, I think,

Time: 1421.27

so, first of all, the more specific you get,

Time: 1423.48

the less disagreement there is.

Time: 1424.68

So there's lot of disagreement

Time: 1425.96

about what is artificial intelligence,

Time: 1427.9

but there's less disagreement about what is machine learning

Time: 1430.67

and even less when you talk about active learning

Time: 1432.73

or machine teaching or self-supervised learning.

Time: 1436.63

And then when you get

Time: 1437.55

into like NLP language models or transformers,

Time: 1440.67

when you get into specific neural network architectures,

Time: 1443.82

there's less and less and less disagreement

Time: 1445.67

about those terms.

Time: 1446.71

So you might be hearing the disagreement

Time: 1448.1

from the high-level terms,

Time: 1449.37

and that has to do with the fact that engineering,

Time: 1452.11

especially when you're talking about intelligence systems

Time: 1455.24

is a little bit of an art and a science.

Time: 1460.87

So the art part is the thing that creates disagreements,

Time: 1465.36

because then you start having disagreements

Time: 1468.66

about how easy or difficult the particular problem is.

Time: 1473.91

For example,

Time: 1474.743

a lot of people disagree with Elon

Time: 1477.61

how difficult the problem of autonomous driving is.

Time: 1481.14

And so, but nobody knows.

Time: 1483.16

So there's a lot of disagreement about,

Time: 1484.77

what are the limits of these techniques?

Time: 1487.2

And through that,

Time: 1488.49

the terminology also contains within it the disagreements.

Time: 1494

But overall,

Time: 1494.833

I think it's also a young science

Time: 1496.81

that also has to do with that.

Time: 1498.86

So like it's not just engineering,

Time: 1501.23

it's that artificial intelligence truly

Time: 1503.7

is a large-scale discipline,

Time: 1505.52

where it's thousands, tens of thousands,

Time: 1508.14

hundreds of thousands of people working on it,

Time: 1510.07

huge amounts of money being made as a very recent thing.

Time: 1513.87

So we're trying to figure out those terms.

Time: 1516.49

And, of course,

Time: 1517.323

there's egos and personalities and a lot of fame to be made.

Time: 1522.65

Like the term deep learning, for example,

Time: 1525.77

neural networks have been around

Time: 1527.02

for many, many decades since the '60s,

Time: 1528.657

you can argue since the '40s.

Time: 1530.88

So there was a rebranding of neural networks

Time: 1533.49

into the word, deep learning,

Time: 1535.45

term, deep learning,

Time: 1536.604

that was part of the re-invigoration of the field,

Time: 1540.94

but it's really the same exact thing.

Time: 1542.72

- I didn't know that.

Time: 1543.67

I mean, I grew up in the age of neuroscience

Time: 1546.06

when neural networks were discussed,

Time: 1549.09

computational neuroscience and theoretical neuroscience,

Time: 1551.42

they had their own journals.

Time: 1553.11

It wasn't actually taken terribly seriously

Time: 1555.06

by experimentalists until a few years ago.

Time: 1557.26

I would say about five to seven years ago.

Time: 1560.72

Excellent theoretical neuroscientist like Larry Abbott

Time: 1563.54

and other colleagues,

Time: 1566.35

certainly at Stanford as well

Time: 1567.363

that people started paying attention

Time: 1568.83

to computational methods.

Time: 1570.36

But these terms,

Time: 1571.193

neural networks, computational methods,

Time: 1573.25

I actually didn't know that neural network works

Time: 1575.18

in deep learning where those have now become

Time: 1578.57

kind of synonymous.

Time: 1579.461

- No, they're always the same thing.

Time: 1582.74

- Interesting.

Time: 1583.573

It was, so.

Time: 1584.406

- I'm a neuroscientist and I didn't know that.

Time: 1585.79

- So, well, because neural networks

Time: 1587.73

probably means something else

Time: 1588.97

and neural science not something else,

Time: 1590.23

but a little different flavor depending on the field.

Time: 1592.64

And that's fascinating too,

Time: 1594.12

because neuroscience and AI people

Time: 1597.1

have started working together and dancing a lot more

Time: 1600.392

in the recent,

Time: 1601.6

I would say probably decade.

Time: 1603.06

- Oh, machines are going into the brain.

Time: 1605.992

I have a couple of questions,

Time: 1607.65

but one thing that I'm sort of fixated on

Time: 1609.84

that I find incredibly interesting

Time: 1611.88

is this example you gave of playing a game

Time: 1615.77

with a mutated version of yourself as a competitor.

Time: 1619.46

- Yeah.

Time: 1620.293

- I find that incredibly interesting

Time: 1622.36

as a kind of a parallel or a mirror

Time: 1625.25

for what happens when we try and learn as humans,

Time: 1627.64

which is we generate repetitions

Time: 1630.03

of whatever it is we're trying to learn,

Time: 1631.96

and we make errors.

Time: 1633.28

Occasionally we succeed.

Time: 1635.06

In a simple example, for instance,

Time: 1636.7

of trying to throw bulls eyes on a dartboard.

Time: 1638.8

- Yeah.

Time: 1639.633

- I'm going to have errors, errors, errors.

Time: 1640.51

I'll probably miss the dartboard.

Time: 1641.73

And maybe occasionally, hit a bullseye.

Time: 1643.44

And I don't know exactly what I just did, right?

Time: 1645.946

But then let's say I was playing darts

Time: 1648.76

against a version of myself

Time: 1650.37

where I was wearing a visual prism,

Time: 1652.48

like my visual, I had a visual defect,

Time: 1656.65

you learn certain things in that mode as well.

Time: 1658.98

You're saying that a machine can sort of mutate itself,

Time: 1662.92

does the mutation always cause a deficiency

Time: 1665.13

that it needs to overcome?

Time: 1666.56

Because of mutations in biology

Time: 1668.01

sometimes give us super powers, right?

Time: 1669.63

Occasionally, you'll get somebody

Time: 1671.1

who has better than 2020 vision,

Time: 1672.75

and they can see better than 99.9% of people out there.

Time: 1676.41

So, when you talk about a machine playing a game

Time: 1679.17

against a mutated version of itself,

Time: 1681.37

is the mutation always say what we call a negative mutation,

Time: 1684.266

or an adaptive or a maladaptive mutation?

Time: 1687.68

- No, you don't know until you get,

Time: 1691.01

so, you mutate first

Time: 1692.06

and then figure out and they compete against each other.

Time: 1694.5

- So, you're evolving,

Time: 1695.83

the machine gets to evolve itself in real time.

Time: 1698.43

- Yeah.

Time: 1699.263

And I think of it,

Time: 1700.86

which would be exciting

Time: 1701.86

if you could actually do with humans.

Time: 1703.64

It's not just.

Time: 1705.21

So, usually you freeze a version of the system.

Time: 1710.37

So, really you take on Andrew of yesterday

Time: 1713.96

and you make 10 clones of them.

Time: 1716.64

And then maybe you mutate, maybe not.

Time: 1719.02

And then you do a bunch of competitions

Time: 1721.13

of the Andrew of today, like you fight to the death,

Time: 1724.12

and who wins last.

Time: 1725.69

So, I love that idea

Time: 1726.61

of like creating a bunch of clones of myself

Time: 1729.04

from like from each of the day for the past year,

Time: 1732.43

and just seeing who's going to be better

Time: 1734.58

at like podcasting or science,

Time: 1736.55

or picking up chicks at a bar or I don't know,

Time: 1740.23

or competing in Jujitsu.

Time: 1742.26

That's the one way to do it,

Time: 1743.17

I mean, a lot of Lexes would have to die for that process,

Time: 1746.4

but that's essentially what happens,

Time: 1747.93

is in reinforcement learning

Time: 1749.48

through the self-play mechanisms,

Time: 1751.014

it's a graveyard of systems that didn't do that well.

Time: 1754.37

And the surviving, the good ones survive.

Time: 1759.8

- Do you think that, I mean, Darwin's Theory of Evolution

Time: 1763.74

might have worked in some sense in this way,

Time: 1766.34

but at the population level.

Time: 1767.77

I mean, you get a bunch of birds with different shaped beaks

Time: 1769.6

and some birds have the shaped beak

Time: 1771.01

that allows them to get the seeds.

Time: 1772.32

I mean, is a trivially simple example

Time: 1774.92

of Darwinian in evolution,

Time: 1776.22

but I think it's correct even though it's not exhaustive.

Time: 1780.85

Is what you're referring to?

Time: 1782.19

You essentially that normally this is done

Time: 1784.13

between members of a different species,

Time: 1785.6

lots of different members of species have different traits

Time: 1788.03

and some get selected for,

Time: 1789.005

but you could actually create multiple versions of yourself

Time: 1792.62

with different traits.

Time: 1793.92

- So, with, I should probably have said this,

Time: 1796.5

but perhaps it's implied with machine learning,

Time: 1799.963

with reinforcement learning through these processes.

Time: 1802.52

One of the big requirements,

Time: 1803.885

is to have an objective function, a loss function,

Time: 1806.54

a utility function,

Time: 1807.84

those are all different terms for the same thing,

Time: 1809.669

is there's a like any equation that says what's good,

Time: 1815.12

and then you're trying to optimize that equation.

Time: 1817.55

So, there's a clear goal for these systems.

Time: 1820.602

- Because it's a game, like with chess, there's a goal.

Time: 1823.96

- But for anything.

Time: 1825.32

Anything you want machine learning to solve,

Time: 1827.69

there needs to be an objective function.

Time: 1829.97

In machine learning, it's usually called Loss Function,

Time: 1832.307

that you're optimizing.

Time: 1834.38

The interesting thing about evolution,

Time: 1837.54

it's complicated of course,

Time: 1838.74

but the goal also seems to be evolving.

Time: 1841.74

Like it's a, I guess, adaptation to the environment,

Time: 1844.05

is the goal,

Time: 1844.99

but it's unclear that you can convert that always.

Time: 1848.32

It's like survival of the fittest.

Time: 1852.18

It's unclear what the fittest is.

Time: 1853.89

In machine learning,

Time: 1855.37

the starting point,

Time: 1856.85

and this is like what human ingenuity provides,

Time: 1860.49

is that fitness function of what's good and what's bad,

Time: 1864.45

which it lets you know which of the systems is going to win.

Time: 1868.38

So, you need to have a equation like that.

Time: 1870.579

One of the fascinating things about humans,

Time: 1872.89

is we figure out objective functions for ourselves.

Time: 1877.17

Like it's the meaning of life,

Time: 1880.54

like why the hell are we here?

Time: 1882.99

And a machine currently

Time: 1885.11

has to have a hard-coded statement about why.

Time: 1889.27

- It has to have a meaning of-

Time: 1890.53

- Yeah.

Time: 1891.363

- Artificial intelligence-based life.

Time: 1893.28

- Right.

Time: 1894.113

It can't.

Time: 1894.946

So, like there's a lot of interesting explorations

Time: 1897.63

about that function being more about curiosity,

Time: 1902.44

about learning new things and all that kind of stuff,

Time: 1905.27

but it's still hard coded.

Time: 1906.72

If you want a machine to be able to be good at stuff,

Time: 1909.62

it has to be given very clear statements

Time: 1913.48

of what good at stuff means.

Time: 1916.11

That's one of the challenges of artificial intelligence,

Time: 1918.4

is you have to formalize the,

Time: 1921.63

in order to solve a problem, you have to formalize it

Time: 1923.897

and you have to provide

Time: 1926.19

both like the full sensory information,

Time: 1928.29

you have to be very clear about what is the data

Time: 1930.667

that's being collected,

Time: 1932.208

and you have to also be clear about the objective function.

Time: 1935.9

What is the goal that you're trying to reach?

Time: 1938.88

And that's a very difficult thing

Time: 1940.75

for artificial intelligence.

Time: 1942.14

- I love that you mentioned curiosity,

Time: 1943.99

I am sure this definition falls short in many ways,

Time: 1946.98

but I define curiosity

Time: 1948.47

as a strong interest in knowing something,

Time: 1953.19

but without an attachment to the outcome.

Time: 1955.598

It's sort of a, it could be a random search,

Time: 1959.36

but there's not really an emotional attachment,

Time: 1962.1

it's really just a desire

Time: 1963.37

to discover and unveil what's there

Time: 1965.57

without hoping it's a gold coin under a rock,

Time: 1968.91

you're just looking under rocks.

Time: 1970.269

Is that more or less how the,

Time: 1972.39

within machine learning,

Time: 1973.91

it sounds like there are elements

Time: 1975.32

of reward prediction and rewards.

Time: 1978.6

The machine has to know when it's done the right thing.

Time: 1981.5

So, can you make machines that are curious,

Time: 1985.32

or are the sorts of machines that you are describing,

Time: 1988.05

curious by design?

Time: 1990.25

- Yeah, curiosity is a kind of a symptom, not the goal.

Time: 1996.31

So, what happens,

Time: 1998.2

is one of the big trade-offs in reinforcement learning,

Time: 2002.33

is this exploration versus exploitation.

Time: 2005.195

So, when you know very little, it pays off to explore a lot,

Time: 2009.64

even suboptimal, like even trajectories

Time: 2012.47

that seem like they're not going to lead anywhere,

Time: 2014.67

that's called exploration.

Time: 2016.21

The smarter and smarter and smarter you get,

Time: 2018.247

the more emphasis you put on exploitation,

Time: 2021.86

meaning you take the best solution, you take the best path.

Time: 2025.369

Now, through that process,

Time: 2027.25

the exploration can look like curiosity by us humans,

Time: 2032.567

but it's really just trying to get out of the local optimal,

Time: 2035.71

the thing it's already discovered.

Time: 2038.78

From an AI perspective,

Time: 2040.2

it's always looking to optimize the objective function,

Time: 2044.41

it derives, and we can talk about the slot more,

Time: 2048.19

but in terms of the tools of machine learning today,

Time: 2051.55

it derives no pleasure from just the curiosity

Time: 2055.99

of like, I don't know, discovery.

Time: 2059.522

- So, there's no dopamine

Time: 2060.605

for machine learning. - There's no dopamine.

Time: 2061.86

- There's no reward, system chemical,

Time: 2064.03

or I guess electronic-reward system.

Time: 2066.93

- That said, if you look at machine learning literature

Time: 2070.41

and reinforcement learning literature,

Time: 2072.1

that will use,

Time: 2073.01

like deep mind, we use terms like dopamine,

Time: 2075.78

we're constantly trying to use the human brain

Time: 2078.86

to inspire totally new solutions to these problems.

Time: 2081.87

So, they'll think like,

Time: 2082.75

how does dopamine function in the human brain,

Time: 2085

and how can it lead to more interesting ways

Time: 2089.16

to discover optimal solutions?

Time: 2091.52

But ultimately currently,

Time: 2094.14

there has to be a formal objective function.

Time: 2097.5

Now, you could argue

Time: 2098.333

the humans also has a set of objective functions

Time: 2100.51

we try and optimize, we're just not able to introspect them.

Time: 2104.55

- Yeah, we don't actually know what we're looking for

Time: 2107.93

and seeking and doing.

Time: 2109.35

- Well, like Lisa Feldman Barrett

Time: 2111.05

who we spoken with at least on Instagram, I hope you-

Time: 2113.56

- I met her through you, yeah.

Time: 2114.81

- Yeah, I hope you actually have are on this podcast.

Time: 2117.58

- Yes, she's terrific.

Time: 2118.85

- So, she has a very,

Time: 2122.482

it has to do with homeostasis like that.

Time: 2126.82

Basically, there's a very dumb objective function

Time: 2128.743

that the brain is trying to optimize,

Time: 2130.89

like to keep like body temperature the same.

Time: 2132.97

Like there's a very dom

Time: 2134.35

kind of optimization function happening.

Time: 2136.48

And then what we humans do with our fancy consciousness

Time: 2139.58

and cognitive abilities, is we tell stories to ourselves

Time: 2142.045

so we can have nice podcasts,

Time: 2144.08

but really it's the brain trying to maintain a,

Time: 2148.11

just like healthy state, I guess.

Time: 2150.76

That's fascinating.

Time: 2151.91

I also see the human brain,

Time: 2155.89

and I hope artificial intelligence systems,

Time: 2158.99

as not just systems that solve problems, or optimize a goal,

Time: 2164.21

but are also storytellers.

Time: 2166.3

I think there's a power to telling stories.

Time: 2168.86

We tell stories to each other, that's what communication is.

Time: 2171.374

Like when you're alone, that's when you solve problems,

Time: 2176.72

that's when it makes sense to talk about solving problems.

Time: 2179.09

But when you're a community,

Time: 2180.86

the capability to communicate, tell stories,

Time: 2185.47

share ideas in such a way that those ideas are stable

Time: 2188.24

over a long period of time,

Time: 2190.03

that's like, that's being a charismatic storyteller.

Time: 2193.34

And I think both humans are very good at this.

Time: 2195.9

Arguably, I would argue that's why we are who we are,

Time: 2200.26

is we're great storytellers.

Time: 2201.993

And then AI I hope will also become that.

Time: 2204.82

So, it's not just about being able to solve problems

Time: 2207.48

with a clear objective function,

Time: 2209.06

it's afterwards, be able to tell like a way better,

Time: 2211.95

like make up a way better story about why you did something,

Time: 2214.82

or why you failed.

Time: 2215.8

- So, you think that robots or, and/or machines of some sort

Time: 2219.88

are going to start telling human stories?

Time: 2222.37

- Well, definitely.

Time: 2223.48

So, the technical field for that is called Explainable AI,

Time: 2227.38

Explainable Artificial Intelligence,

Time: 2229.35

is trying to figure out how you get the AI system

Time: 2233.877

to explain to us humans why the hell it failed,

Time: 2237.56

or why it succeeded,

Time: 2239.075

or there's a lot of different sort of versions of this,

Time: 2242.36

or to visualize how it understands the world.

Time: 2246.35

That's a really difficult problem,

Time: 2248.14

especially with neural networks

Time: 2249.67

that are famously opaque,

Time: 2252.824

that we don't understand in many cases,

Time: 2256.36

why a particular neural network does what it does so well,

Time: 2260.51

and to try to figure out where it's going to fail,

Time: 2263.73

that requires the AI to explain itself.

Time: 2266.28

There's a huge amount of money,

Time: 2269.57

like there's a huge amount of money in this,

Time: 2272.39

especially from government funding and so on.

Time: 2274.6

Because if you want to deploy AI systems in the real world,

Time: 2279.149

we humans at least, want to ask it a question like,

Time: 2282.9

why the hell did you do that?

Time: 2284.3

Like in a dark way,

Time: 2286.7

why did you just kill that person, right?

Time: 2289.17

Like if a car ran over a person,

Time: 2290.66

we want to understand why that happened.

Time: 2293.05

And now again, we're sometimes very unfair to AI systems

Time: 2297.93

because we humans can often not explain why very well.

Time: 2301.92

But that's the field of Explainable AI

Time: 2305.6

that people are very interested in

Time: 2308.21

because the more and more we rely on AI systems,

Time: 2311.54

like the Twitter recommender system,

Time: 2314.06

that AI algorithm that's, I would say impacting elections,

Time: 2319.19

perhaps starting wars, or at least military conflict,

Time: 2322.02

that algorithm, we want to ask that algorithm,

Time: 2326.05

first of all, do you know what the hell you're doing?

Time: 2328.53

Do you understand the society-level effects you're having?

Time: 2332.77

And can you explain the possible other trajectories?

Time: 2335.86

Like we would have that kind of conversation with a human,

Time: 2338.35

we want to be able to do that with an AI.

Time: 2340.07

And in my own personal level,

Time: 2342.06

I think it would be nice to talk to AI systems

Time: 2345.48

for stupid stuff, like robots when they fail to-

Time: 2351.65

- Why'd you fall down the stairs?

Time: 2352.94

- Yeah.

Time: 2353.773

But not an engineering question,

Time: 2355.91

but almost like an endearing question,

Time: 2358.842

like I'm looking for,

Time: 2361.65

if I fell and you and I were hanging out,

Time: 2365.04

I don't think you need an explanation

Time: 2368.07

exactly what were the dynamics,

Time: 2369.89

like what was the under actuated system problem here?

Time: 2372.81

Like what was the texture of the floor?

Time: 2375.65

Or so on.

Time: 2376.483

Or like, what was the-

Time: 2377.427

- No, I want to know what you're thinking.

Time: 2379.06

- That, or you might joke about like,

Time: 2381.25

you're drunk again, go home, or something,

Time: 2383.14

like there could be humor in it, that's an opportunity.

Time: 2386.99

Like storytelling, isn't just explanation of what happened,

Time: 2390.662

it's something that makes people laugh,

Time: 2394.44

it makes people fall in love, it makes people dream,

Time: 2397.51

and understand things

Time: 2399.01

in a way that poetry makes people understand things

Time: 2402.02

as opposed to a rigorous log

Time: 2404.68

of where every sensor was, where every actuator was.

Time: 2409.59

- I mean, I find this incredible

Time: 2411.51

because one of the hallmarks

Time: 2413.91

of severe autism spectrum disorders is,

Time: 2417.67

a report of experience from the autistic person

Time: 2421.9

that is very much a catalog of action steps.

Time: 2425.35

It's like, how do you feel today?

Time: 2426.42

And they'll say, well, I got up and I did this,

Time: 2428.01

and then I did this, and I did this.

Time: 2429.13

And it's not at all the way that a person

Time: 2431.72

who doesn't have autism spectrum disorder would respond.

Time: 2435.75

And the way you describe these machines

Time: 2438.8

has so much humanism,

Time: 2442.45

or so much of a human and biological element,

Time: 2445.46

but I realized that we were talking about machines.

Time: 2448.926

I want to make sure that I understand

Time: 2451.72

if there's a distinction between a machine that learns,

Time: 2457.9

a machine with artificial intelligence and a robot.

Time: 2460.84

Like at what point does a machine become a robot?

Time: 2463.89

So, if I have a ballpoint pen,

Time: 2466.5

I'm assuming I wouldn't call that a robot,

Time: 2468.7

but if my ballpoint pen can come to me

Time: 2472.44

when I moved to the opposite side of the table,

Time: 2475.36

if it moves by whatever mechanism,

Time: 2477.97

at that point, does it become a robot?

Time: 2480.7

- Okay, there's 1 million ways to explore this question.

Time: 2483.46

It's a fascinating one.

Time: 2485.13

So, first of all, there's a question of what is life?

Time: 2489.26

Like how do you know something as a living form and not?

Time: 2492.5

And it's to the question of when does sort of a,

Time: 2495.5

maybe a cold computational system becomes a,

Time: 2500.237

or already loading these words with a lot of meaning,

Time: 2502.72

robot and machine,

Time: 2506.24

So, one, I think movement is important,

Time: 2510.69

but that's a kind of a boring idea

Time: 2512.37

that a robot is just a machine

Time: 2514.227

that's able to act in the world.

Time: 2516.65

So, one artificial intelligence

Time: 2518.86

could be both just the thinking thing,

Time: 2521.84

which I think is what machine learning is,

Time: 2524.08

and also the acting thing,

Time: 2525.76

which is what we usually think about robots.

Time: 2527.93

So, robots are the things that have a perception system

Time: 2529.757

that's able to take in the world

Time: 2531.64

however you define the world,

Time: 2533.19

is able to think and learn

Time: 2534.64

and do whatever the hell it does inside,

Time: 2536.75

and then act on the world.

Time: 2538.73

So, that's the difference

Time: 2539.563

between maybe an AI system learning machine and a robot,

Time: 2543.26

it's something that's able,

Time: 2544.3

a robot is something that's able to perceive the world

Time: 2547.26

and act in the world.

Time: 2548.21

- So, it could be through language or sound,

Time: 2551.12

or it could be through movement or both.

Time: 2552.69

- Yeah.

Time: 2553.734

And I think it could also be in the digital space

Time: 2556.12

as long as there's a aspect of entity

Time: 2559.07

that's inside the machine

Time: 2560.954

and a world that's outside the machine.

Time: 2564.24

And there's a sense in which the machine

Time: 2566.48

is sensing that world and acting in it.

Time: 2569.37

- So, we could,

Time: 2570.203

for instance, there could be a version of a robot,

Time: 2572.64

according to the definition that I think you're providing,

Time: 2575.44

where the robot, where I go to sleep at night

Time: 2578.28

and this robot goes and forges for information

Time: 2581.47

that it thinks I want to see loaded onto my desktop

Time: 2584.83

in the morning.

Time: 2585.663

There was no movement of that machine,

Time: 2587.04

there was no language,

Time: 2587.873

but it essentially, has movement in cyberspace.

Time: 2591.17

- Yeah, there's a distinction that I think is important

Time: 2598.43

in that there's an element of it being an entity,

Time: 2604.21

whether it's in the digital or the physical space.

Time: 2606.66

So, when you have something like Alexa in your home,

Time: 2612.31

most of the speech recognition, most of what Alexa is doing,

Time: 2616.76

is constantly being sent back to the mothership.

Time: 2622.09

When Alexa is there on its own, that's to me, a robot,

Time: 2626.99

when it's there interacting with the world.

Time: 2629.094

When it's simply a finger of the main mothership,

Time: 2634.374

then the Alexa is not a robot,

Time: 2636.7

then it's just an interaction device,

Time: 2638.91

then may be the main Amazon Alexa AI,

Time: 2642.4

big, big system is the robot.

Time: 2644.83

So, that's important because there's some element,

Time: 2648.89

to us humans, I think, where we want there to be an entity,

Time: 2652.6

whether in the digital or the physical space,

Time: 2654.78

that's where ideas of consciousness come in

Time: 2656.78

and all those kinds of things

Time: 2658.61

that we project our understanding

Time: 2660.753

of what it means to be a being.

Time: 2664.3

And so, to take that further,

Time: 2667.05

when does a machine become a robot,

Time: 2671.4

I think there's a special moment.

Time: 2675.26

There's a special moment in a person's life

Time: 2677.757

and in a robot's life where it surprises you.

Time: 2681.64

I think surprise is a really powerful thing,

Time: 2684.31

where you know how the thing works and yet it surprises you,

Time: 2690.03

that's a magical moment for us humans.

Time: 2692.01

So, whether it's a chess-playing program

Time: 2694.68

that does something that you haven't seen before,

Time: 2697.65

that makes people smile like, huh,

Time: 2701.2

those moments happen with AlphaZero for the first time

Time: 2704.35

in chess playing,

Time: 2705.6

where grand masters were really surprised by a move.

Time: 2708.79

They didn't understand the move

Time: 2710.29

and then they studied and studied

Time: 2711.68

and then they understood it.

Time: 2713.41

But that moment of surprise,

Time: 2715.32

that's for grandmasters in chess.

Time: 2717.088

I find that moment of surprise really powerful,

Time: 2720.45

really magical in just everyday life.

Time: 2723.36

- Because it supersedes the human brain in that moment?

Time: 2728.14

- So, it's not supersedes, like outperforms,

Time: 2731.04

but surprises you in a positive sense.

Time: 2735.42

Like I didn't think he could do that,

Time: 2738.27

I didn't think that you had that in you.

Time: 2740.77

And I think that moment is a big transition for a robot

Time: 2745.21

from a moment of being a servant

Time: 2748.313

that accomplishes a particular task

Time: 2751.2

with some level of accuracy, with some rate of failure,

Time: 2755.75

to an entity, a being that's struggling

Time: 2759.62

just like you are in this world.

Time: 2761.94

And that's a really important moment

Time: 2764.49

that I think you're not going to find many people

Time: 2767.55

in the AI community that talk like I just did.

Time: 2771.119

I'm not speaking like some philosopher or some hippie,

Time: 2774.41

I'm speaking from purely engineering perspective.

Time: 2776.82

I think it's really important for robots to become entities

Time: 2780.125

and explore that as a real engineering problem,

Time: 2783.39

as opposed to everybody treats robots

Time: 2785.92

in the robotics community,

Time: 2787.71

they don't even call them or he or she,

Time: 2789.58

they don't give them, try to avoid giving them names,

Time: 2791.954

they've really want to see like a system, like a servant.

Time: 2796.74

They see it as a servant that's trying to accomplish a task.

Time: 2800.32

To me, and don't think I'm just romanticizing the notion,

Time: 2804.7

I think it's a being, it's a currently perhaps a dumb being,

Time: 2808.594

but in the long arc of history,

Time: 2813.1

humans are pretty dumb beings too, so-

Time: 2815.22

- I would agree with that statement.

Time: 2816.233

[Andrew laughing]

Time: 2817.14

- So, I tend to really want to explore this treating robots

Time: 2821.55

really as entities, yeah.

Time: 2825.76

So, like anthropomorphization,

Time: 2828.44

which is the sort of the act

Time: 2829.95

of looking at a inanimate object

Time: 2832.53

and projecting onto it life-like features,

Time: 2835.49

I think robotics generally sees that as a negative,

Time: 2841.26

I see it as a superpower.

Time: 2843.89

Like that, we need to use that.

Time: 2846.77

- Well, I'm struck

Time: 2847.75

by how that really grabs onto the relationship

Time: 2850.61

between human and machine, or human and robot.

Time: 2854.11

So, I guess the simple question is,

Time: 2856.71

and I think you've already told us the answer,

Time: 2858.92

but does interacting with a robot change you?

Time: 2864.01

In other words, do we develop relationships to robots?

Time: 2868.07

- Yeah, I definitely think so.

Time: 2870.28

I think the moment you see a robot or AI systems

Time: 2875.225

as more than just servants but entities,

Time: 2880.02

they begin to change you, in just like good friends do,

Time: 2882.62

just like relationships just to other humans.

Time: 2886.437

I think for that,

Time: 2888.33

you have to have certain aspects of that interaction.

Time: 2891.49

Like the robot's ability to say no,

Time: 2895.98

to have its own sense of identity,

Time: 2899.13

to have its own set of goals,

Time: 2901.77

that's not constantly serving you,

Time: 2902.603

but instead, trying to understand the world

Time: 2904.95

and do that dance of understanding

Time: 2907.25

through communication with you.

Time: 2908.96

So, I definitely think there's a,

Time: 2911.82

I mean, I have a lot of thoughts about this as you may know,

Time: 2914.573

and that's at the core of my life-long dream actually

Time: 2919.15

of what I want to do,

Time: 2919.983

which is I believe that most people

Time: 2924.41

have a notion of loneliness in them

Time: 2928.87

that we haven't discovered,

Time: 2930.34

that we haven't explored, I should say.

Time: 2932.742

And I see AI systems as helping us explore that

Time: 2937.89

so that we can become better humans,

Time: 2940.28

better people towards each other.

Time: 2942.31

So, I think that connection

Time: 2945.13

between human and AI, human and robot,

Time: 2949.195

is not only possible, but will help us understand ourselves

Time: 2954.56

in ways that are like several orders of magnitude

Time: 2958.8

deeper than we ever could have imagined.

Time: 2961.36

I tend to believe that [sighing]

Time: 2964.38

well, I have very wild levels of belief

Time: 2972.53

in terms of how impactful that will be, right?

Time: 2975.64

- So, when I think about human relationships,

Time: 2977.78

I don't always break them down into variables,

Time: 2981.19

but we could explore a few of those variables

Time: 2983.45

and see how they map to human-robot relationships.

Time: 2987.39

One is just time, right?

Time: 2989.19

If you spend zero time with another person at all

Time: 2992.879

in cyberspace or on the phone or in person,

Time: 2995.42

you essentially have no relationship to them.

Time: 2997.88

If you spend a lot of time, you have a relationship,

Time: 2999.71

this is obvious.

Time: 3000.543

But I guess one variable would be time,

Time: 3001.84

how much time you spend with the other entity,

Time: 3005.22

robot or human.

Time: 3006.58

The other would be wins and successes.

Time: 3010.05

You enjoy successes together.

Time: 3013.87

I'll give a absolutely trivial example this in a moment,

Time: 3016.72

but the other would be failures.

Time: 3019.21

When you struggle with somebody,

Time: 3021.56

whether or not you struggle between one another,

Time: 3023.61

you disagree,

Time: 3024.55

like I was really struck by the fact

Time: 3025.7

that you said that robot saying no,

Time: 3027.1

I've never thought about a robot saying no to me,

Time: 3030.13

but there it is.

Time: 3031.15

- I look forward to you being one of the first people

Time: 3034.07

I send this robots to.

Time: 3035.33

- So do I.

Time: 3036.35

So, there's struggle.

Time: 3037.856

When you struggle with somebody, you grow closer.

Time: 3041.21

Sometimes the struggles

Time: 3042.59

are imposed between those two people,

Time: 3044.7

so called trauma bonding,

Time: 3045.95

they call it in the whole psychology literature

Time: 3048.35

and pop psychology literature.

Time: 3050.17

But in any case, I can imagine.

Time: 3052.25

So, time successes together, struggle together,

Time: 3057.64

and then just peaceful time,

Time: 3060.06

hanging out at home, watching movies,

Time: 3063.01

waking up near one another,

Time: 3066.21

here, we're breaking down

Time: 3067.21

the elements of relationships of any kind.

Time: 3070.7

So, do you think that these elements

Time: 3073.44

apply to robot-human relationships?

Time: 3076.45

And if so, then I could see how,

Time: 3081.18

if the robot has its own entity

Time: 3084.11

and has some autonomy in terms of how it reacts you,

Time: 3087.2

it's not just there just to serve you,

Time: 3088.99

it's not just a servant,

Time: 3090.23

it actually has opinions,

Time: 3091.94

and can tell you when maybe your thinking is flawed,

Time: 3094.48

or your actions are flawed.

Time: 3095.82

- It can also leave.

Time: 3097.43

- It could also leave.

Time: 3099.41

So, I've never conceptualized

Time: 3100.95

robot-human interactions this way.

Time: 3103.95

So, tell me more about how this might look.

Time: 3106.54

Are we thinking about a human-appearing robot?

Time: 3111.41

I know you and I have both had intense relationships to our,

Time: 3114.45

we have separate dogs obviously,

Time: 3116.21

but to animals,

Time: 3117.35

it sounds a lot like human-animal interaction.

Time: 3119.17

So, what is the ideal human-robot relationship?

Time: 3124.51

- So, there's a lot to be said here,

Time: 3126.4

but you actually pinpointed one of the big, big first steps,

Time: 3130.71

which is this idea of time.

Time: 3133.79

And it's a huge limitation

Time: 3135.54

in machine-learning community currently.

Time: 3138.56

Now we're back to like the actual details.

Time: 3141.5

Life-long learning is a problem space

Time: 3146.36

that focuses on how AI systems

Time: 3149.18

can learn over a long period of time.

Time: 3151.97

What's currently most machine learning systems

Time: 3154.932

are not able to do,

Time: 3157.43

is to all of the things you've listed under time,

Time: 3159.82

the successes, the failures,

Time: 3161.87

or just chilling together watching movies,

Time: 3164.64

AI systems are not able to do that,

Time: 3168

which is all the beautiful, magical moments

Time: 3171.08

that I believe are the days filled with,

Time: 3173.92

they're not able to keep track of those together with you.

Time: 3177.372

- 'Cause they can't move with you and be with you.

Time: 3179.25

- No, no, like literally we don't have the techniques

Time: 3181.966

to do the learning,

Time: 3183.49

the actual learning of containing those moments.

Time: 3187.04

Current machine-learning systems

Time: 3188.634

are really focused on understanding the world

Time: 3191.89

in the following way,

Time: 3192.723

it's more like the perception system,

Time: 3194.38

like looking around, understand like what's in the scene.

Time: 3200.03

That there's a bunch of people sitting down,

Time: 3202.28

that there is cameras and microphones,

Time: 3204.593

that there's a table, understand that.

Time: 3207.147

But the fact that we shared this moment of talking today,

Time: 3211

and still remember that

Time: 3212.82

for like next time you're doing something,

Time: 3216.69

remember that this moment happened.

Time: 3218.4

We don't know how to do that technique-wise.

Time: 3220.44

This is what I'm hoping to innovate on

Time: 3224.18

as I think it's a very, very important component

Time: 3227.13

of what it means to create a deeper relationship,

Time: 3229.99

that sharing of moments together.

Time: 3232.09

- Could you post a photo of you in the robot,

Time: 3234.11

like selfie with robot

Time: 3235.88

and the robot sees that image

Time: 3238.35

and recognizes that was time spent,

Time: 3241.063

there were smiles, or there were tears-

Time: 3243.42

- Yeah.

Time: 3244.253

- And create some sort of metric of emotional depth

Time: 3249.08

in the relationship and update its behavior?

Time: 3251.69

- So.

Time: 3253.41

- Could it...

Time: 3254.512

It texts you in the middle of the night and say,

Time: 3255.345

why haven't you texted me back?

Time: 3256.84

- Well, yes, all of those things, but we can dig into that.

Time: 3261.46

But I think that time element, forget everything else,

Time: 3264.86

just sharing moments together, that changes everything.

Time: 3269.25

I believe that changes everything.

Time: 3270.673

Now, there's specific things

Time: 3272.24

that are more in terms of systems that I can explain you.

Time: 3277.35

It's more technical and probably a little bit offline,

Time: 3279.42

'cause I have kind of wild ideas

Time: 3281.34

how that can revolutionize social networks

Time: 3284.79

and operating systems.

Time: 3287.268

But the point is that element alone,

Time: 3290.67

forget all the other things we're talking about

Time: 3293.23

like emotions, saying no, all that,

Time: 3296.09

just remembering sharing moments together

Time: 3298.64

would change everything.

Time: 3300.3

We don't currently have systems that share moments together.

Time: 3305.64

Like even just you and your fridge,

Time: 3308.13

just all those times, you went late at night

Time: 3310.712

and ate thing you shouldn't have eaten,

Time: 3313.33

that was a secret moment you have with your refrigerator.

Time: 3316.73

You shared that moment,

Time: 3318

that darkness or that beautiful moment

Time: 3320.4

where you were just like heartbroken for some reason,

Time: 3324.1

you're eating that ice cream or whatever,

Time: 3326.35

that's a special moment.

Time: 3327.67

And that refrigerator was there for you,

Time: 3329.67

and the fact that it missed the opportunity

Time: 3331.86

to remember that is tragic.

Time: 3336.04

And once it does remember that,

Time: 3338.8

I think you're going to be very attached to the refrigerator.

Time: 3342.47

You're going to go through some hell with that refrigerator.

Time: 3345.75

Most of us have, like in the developed world,

Time: 3349.68

have weird relationships with food, right?

Time: 3351.56

So, you can go through some deep moments

Time: 3354.91

of trauma and triumph with food,

Time: 3357.3

and at the core of that, is the refrigerator.

Time: 3359.23

So, a smart refrigerator, I believe would change society.

Time: 3365

Not just the refrigerator,

Time: 3366.24

but these ideas in the systems all around us.

Time: 3370.25

So, I just want to comment

Time: 3371.95

on how powerful that idea of time is.

Time: 3374.26

And then there's a bunch of elements

Time: 3375.85

of actual interaction of allowing you as a human

Time: 3383.34

to feel like you're being heard.

Time: 3386.42

Truly heard, truly understood,

Time: 3389.15

that we human, like deep friendship is like that, I think,

Time: 3393.08

but there's still an element of selfishness,

Time: 3396.98

there's still an element

Time: 3397.93

of not really being able to understand another human.

Time: 3400.76

And a lot of the times

Time: 3401.975

when you're going through trauma together,

Time: 3404.91

through difficult times and through successes,

Time: 3407.63

you actually starting

Time: 3408.463

to get that inkling of understanding of each other,

Time: 3411.25

but I think

Time: 3412.083

that could be done more aggressively, more efficiently.

Time: 3417.46

Like if you think of a great therapist,

Time: 3419.31

I think I've never actually been to a therapist,

Time: 3421.84

but I'm a believer I used to want to be a psychiatrist.

Time: 3425

- Do Russians go to therapists?

Time: 3426.37

- No, they don't.

Time: 3427.34

They don't.

Time: 3428.173

And if they do, the therapist don't live to tell the story.

Time: 3434.004

I do believe in talk therapy,

Time: 3436.05

which friendship is to me, is it's talk therapy,

Time: 3438.62

or like you don't even necessarily need to talk [laughing]

Time: 3443.74

it's like just connecting through in the space of ideas

Time: 3446.927

and the space of experiences.

Time: 3448.93

And I think there's a lot of ideas

Time: 3450.81

of how to make AI systems

Time: 3452.21

to be able to ask the right questions

Time: 3454.569

and truly hear another human.

Time: 3457.36

This is what we try to do with podcasting, right?

Time: 3460.59

I think there's ways to do that with AI.

Time: 3462.66

But above all else,

Time: 3464.48

just remembering the collection of moments

Time: 3468.49

that make up the day, the week, the months,

Time: 3472.05

I think you maybe have some of this as well.

Time: 3475.343

Some of my closest friends

Time: 3476.88

still are the friends from high school.

Time: 3479.76

That's time, we've been through a bunch of together,

Time: 3483.01

and that like we're very different people.

Time: 3486.26

But just the fact that we've been through that,

Time: 3487.697

and we remember those moments,

Time: 3489.35

and those moments somehow create a depth of connection

Time: 3492.86

like nothing else, like you and your refrigerator.

Time: 3497.13

- I love that because my graduate advisor,

Time: 3500.57

she unfortunately, she passed away,

Time: 3501.76

but when she passed away,

Time: 3502.63

somebody said at her at her memorial

Time: 3507.288

all these amazing things she had done, et cetera.

Time: 3509.45

And then her kids got up there,

Time: 3511.79

and she had young children and that I knew

Time: 3513.45

as when she was pregnant with them.

Time: 3515.52

And so, it was really,

Time: 3516.76

you're even now I can feel like your heart gets heavy,

Time: 3519.3

thinking about this,

Time: 3520.133

they're going to grow up without their mother.

Time: 3521.86

And it was really amazing,

Time: 3522.693

very strong young girls, and now the young women.

Time: 3526.954

And what they said was incredible,

Time: 3529.34

they said what they really appreciated most

Time: 3531.77

about their mother, who was an amazing person,

Time: 3535.85

is all the unstructured time they spent together.

Time: 3539.31

- Mm-hmm.

Time: 3540.385

- So, it wasn't the trips to the zoo,

Time: 3541.218

it wasn't she woke up at five in the morning

Time: 3543.65

and drove us to school.

Time: 3544.483

She did all those things too.

Time: 3545.55

She had two hour commute in each direction,

Time: 3547.37

it was incredible, ran a lab, et cetera,

Time: 3549.4

but it was the unstructured time.

Time: 3551.49

So, on the passing of their mother,

Time: 3552.97

that's what they remembered was that the biggest give

Time: 3556.56

and what bonded them to her,

Time: 3557.81

was all the time where they just kind of hung out.

Time: 3560.45

And the way you describe the relationship to a refrigerator

Time: 3564.06

is so, I want to say human-like,

Time: 3567.23

but I'm almost reluctant to say that.

Time: 3568.925

Because what I'm realizing as we're talking,

Time: 3571.48

is that what we think of as human-like

Time: 3574.44

might actually be the lower form of relationship.

Time: 3579.04

There may be relationships that are far better

Time: 3582.4

than the sorts of relationships

Time: 3583.77

that we can conceive in our minds right now

Time: 3586.87

based on what these machine relationship interactions

Time: 3589.668

could teach us.

Time: 3591.14

Do I have that right?

Time: 3592.418

- Yeah, I think so.

Time: 3594.31

I think there's no reason to see machines

Time: 3595.77

as somehow incapable of teaching us something

Time: 3599.93

that's deeply human.

Time: 3601.62

I don't think humans have a monopoly on that.

Time: 3605.02

I think we understand ourselves very poorly

Time: 3606.98

and we need to have the kind of prompting from a machine.

Time: 3614

And definitely part of that,

Time: 3615.21

is just remembering the moments.

Time: 3618.96

I think the unstructured time together,

Time: 3624.23

I wonder if it's quite so unstructured.

Time: 3627.55

That's like calling this podcast on structured time.

Time: 3630.57

- Maybe what they meant, was it wasn't a big outing,

Time: 3633.91

there was no specific goal,

Time: 3636.21

but a goal was created through the lack of a goal.

Time: 3639.68

Like we would just hang out

Time: 3640.59

and then you start playing, thumb war,

Time: 3642.55

and you end up playing thumb war for an hour.

Time: 3644.63

So, it's the structure emerges from lack of structure.

Time: 3648.95

- No, but the thing is the moments,

Time: 3652.34

there's something about those times

Time: 3654.36

that creates special moments,

Time: 3656.5

and I think those could be optimized for.

Time: 3660.33

I think we think of like a big outing as,

Time: 3662.1

I don't know, going to Six Flags or something,

Time: 3663.85

or some big, the Grand Canyon,

Time: 3666.39

or go into some, I don't know, I think we would need to,

Time: 3671.67

we don't quite yet understand, as humans,

Time: 3673.84

what creates magical moments.

Time: 3675.73

I think it's possible to optimize a lot of those things.

Time: 3678.35

And perhaps like podcasting is helping people discover that,

Time: 3681.41

like maybe the thing we want to optimize for

Time: 3684.23

isn't necessarily like some sexy, like quick clips,

Time: 3691.49

maybe what we want, is long-form authenticity.

Time: 3694.54

- Depth.

Time: 3695.373

- Depth.

Time: 3696.29

So, we were trying to figure that out,

Time: 3698.264

certainly from a deep connection

Time: 3700.323

between humans and humans and NAS systems,

Time: 3704.18

I think long conversations, or long periods of communication

Time: 3708.181

over a series of moments like my new,

Time: 3713.93

perhaps, seemingly insignificant to the big ones,

Time: 3717.21

the big successes, the big failures,

Time: 3720.78

those are all just stitching those together

Time: 3723.06

and talking throughout.

Time: 3725.31

I think that's the formula

Time: 3726.65

for a really, really deep connection.

Time: 3728.28

That from like a very specific engineering perspective,

Time: 3731.97

is I think a fascinating open problem

Time: 3735.61

that has been really worked on very much.

Time: 3738.45

And for me, from a, if I have the guts

Time: 3741.88

and, I mean there's a lot of things to say,

Time: 3744.84

but one of it is guts, is I'll build a startup around it.

Time: 3749.53

- So, let's talk about this startup

Time: 3752.4

and let's talk about the dream.

Time: 3754.84

You mentioned this dream before

Time: 3756.08

in our previous conversations,

Time: 3757.32

always as little hints dropped here and there.

Time: 3760.17

Just for anyone listening,

Time: 3761.42

there's never been an offline conversation about this dream,

Time: 3763.75

I'm not privy to anything, except what Lex says now.

Time: 3768.22

And I realized that there's no way

Time: 3769.8

to capture the full essence of a dream

Time: 3772.89

in any kind of verbal statement

Time: 3776.85

in a way that captures all of it.

Time: 3778.24

But what is this dream

Time: 3781.01

that you've referred to now several times

Time: 3783.89

when we've sat down together and talked on the phone?

Time: 3786.97

Maybe it's this company, maybe it's something distinct.

Time: 3789.13

If you feel comfortable,

Time: 3790.98

it'd be great if you could share a little bit

Time: 3792.53

about what that is. - Sure.

Time: 3793.62

So, the way people express long-term vision,

Time: 3799.1

I've noticed is quite different.

Time: 3800.92

Like Elon is an example of somebody

Time: 3802.71

who can very crisply say exactly what the goal is.

Time: 3807.062

Also has to do with the fact that problems he's solving

Time: 3809.85

have nothing to do with humans.

Time: 3812.44

So, my long-term vision

Time: 3815.17

is a little bit more difficult to express in words,

Time: 3819.02

I've noticed, as I've tried,

Time: 3820.9

it could be my brain's failure,

Time: 3822.51

but there's a way to sneak up to it.

Time: 3825.43

So, let me just say a few things.

Time: 3828.511

Early on in life and also in the recent years,

Time: 3833.03

I've interacted with a few robots

Time: 3835.18

where I understood there's magic there.

Time: 3838.23

And that magic could be shared by millions

Time: 3842.16

if it's brought to light.

Time: 3844.69

When I first met Spot from Boston Dynamics,

Time: 3847.673

I realized there's magic there that nobody else is seeing.

Time: 3850.44

- Is the dog.

Time: 3851.47

- The dog, sorry.

Time: 3852.38

The Spot is the four-legged robot from Boston Dynamics.

Time: 3857.28

Some people might have seen it, it's this yellow dog.

Time: 3860.42

And sometimes in life,

Time: 3865.157

you just notice something that just grabs you.

Time: 3868.47

And I believe that this is something

Time: 3870.81

that this magic is something that could be

Time: 3874.75

in every single device in the world.

Time: 3878.28

The way that I think maybe Steve Jobs

Time: 3881.42

thought about the personal computer,

Time: 3883.89

laws didn't think about the personal computer this way,

Time: 3886.75

but Steve did.

Time: 3888.11

Which is like, he thought that the personal computer

Time: 3890.37

should be as thin as a sheet of paper

Time: 3892.16

and everybody should have one,

Time: 3893.37

and this idea,

Time: 3894.808

I think it is heartbreaking that we're getting,

Time: 3901.25

the world is being filled up with machines

Time: 3903.87

that are soulless.

Time: 3906.11

And I think every one of them can have that same magic.

Time: 3910.002

One of the things that also inspired me

Time: 3914.83

in terms of a startup,

Time: 3916.09

is that magic can engineered much easier than I thought.

Time: 3919.6

That's my intuition

Time: 3920.81

with everything I've ever built and worked on.

Time: 3924

So, the dream is to add a bit of that magic

Time: 3928.9

in every single computing system in the world.

Time: 3932.4

So, the way that Windows Operating System for a long time

Time: 3936.53

was the primary operating system everybody interacted with,

Time: 3939.49

they built apps on top of it.

Time: 3940.982

I think this is something that should be as a layer,

Time: 3945.513

it's almost as an operating system

Time: 3947.428

in every device that humans interacted with in the world.

Time: 3951.04

Now, what that actually looks like,

Time: 3953.08

the actual dream when I was officially a kid,

Time: 3957.97

it didn't have this concrete form of a business,

Time: 3961.17

it had more of a dream of exploring your own loneliness

Time: 3968.29

by interacting with machines, robots.

Time: 3973.38

This deep connection between humans and robots

Time: 3975.75

was always a dream.

Time: 3977.21

And so, for me,

Time: 3978.99

I'd love to see a world where there's every home as a robot,

Time: 3982.49

and not a robot that washes the dishes, or a sex robot,

Time: 3987.89

or I don't know,

Time: 3990.02

think of any kind of activity the robot can do,

Time: 3992.27

but more like a companion.

Time: 3993.88

- A family member.

Time: 3994.86

- A family member, the way a dog is.

Time: 3996.47

- Mm-hmm.

Time: 3997.8

- But a dog that's able to speak your language too.

Time: 4002.01

So, not just connect the way a dog does by looking at you

Time: 4006.02

and looking away

Time: 4006.853

and almost like smiling with its soul in that kind of way,

Time: 4011.15

but also to actually understand what the hell,

Time: 4014.62

like, why are you so excited about the successes?

Time: 4016.78

Like understand the details, understand the traumas.

Time: 4019.92

And that, I just think [sighing]

Time: 4022.48

that has always filled me with excitement that I could,

Time: 4027.09

with artificial intelligence, bring joy to a lot of people.

Time: 4032.41

More recently, I've been more and more heartbroken

Time: 4038.07

to see the kind of division, derision,

Time: 4043.37

even hate that's boiling up on the internet

Time: 4047.14

through social networks.

Time: 4048.68

And I thought this kind of mechanism is exactly applicable

Time: 4052.8

in the context of social networks as well.

Time: 4054.83

So, it's an operating system that serves as your guide

Time: 4061.558

on the internet.

Time: 4063.38

One of the biggest problems

Time: 4064.84

with YouTube and social networks currently,

Time: 4067.99

is they're optimizing for engagement.

Time: 4070.8

I think if you create AI systems

Time: 4072.58

that know each individual person,

Time: 4076.26

you're able to optimize for long-term growth

Time: 4079.14

for a long-term happiness.

Time: 4080.82

- Of the individual, or-

Time: 4081.653

- Of the individual, of the individual.

Time: 4083.76

And there's a lot of other things to say,

Time: 4086.4

which is in order for AI systems

Time: 4090.75

to learn everything about you,

Time: 4095.8

they need to collect,

Time: 4097.12

they need to just like you and I when we talk offline,

Time: 4100.02

we're collecting data

Time: 4100.853

about each others secrets about each other,

Time: 4103.56

the same way AI has to do that.

Time: 4106.66

And that allows you to,

Time: 4108.77

and that requires you to rethink ideas of ownership of data.

Time: 4115.5

I think each individual should own all of their data

Time: 4119.97

and very easily be able to leave

Time: 4121.95

just like AI systems can leave,

Time: 4123.61

humans can disappear

Time: 4126.667

and delete all of their data in a moment's notice.

Time: 4130.67

Which is actually better than we humans can do,

Time: 4133.97

is once we load the data into each other, it's there.

Time: 4136.54

I think it's very important to be both,

Time: 4140.48

give people complete control over their data

Time: 4143.634

in order to establish trust that they can trust you.

Time: 4146.23

And the second part of trust is transparency.

Time: 4149.5

Whenever the data is used

Time: 4150.98

to make it very clear what is being used for.

Time: 4153.08

And not clear in a lawyerly legal sense,

Time: 4156.17

but clear in a way that people

Time: 4158.14

really understand what it's used for.

Time: 4159.65

I believe when people have the ability

Time: 4161.55

to delete all their data and walk away

Time: 4164.2

and know how the data is being used, I think they'll stay.

Time: 4169.86

- The possibility of a clean breakup,

Time: 4171.584

is actually what will keep people together.

Time: 4173.42

- Yeah, I think so.

Time: 4174.48

I think, exactly.

Time: 4176.43

I think a happy marriage

Time: 4178.69

acquires the ability to divorce easily

Time: 4182.109

without the divorce industrial complex or whatever.

Time: 4186.72

Things currently going on

Time: 4187.553

and then there's so much money to be made

Time: 4189.44

from lawyers and divorce.

Time: 4190.77

But yeah, the ability to leave

Time: 4192.89

is what enables love, I think.

Time: 4195.444

- It's interesting.

Time: 4196.277

I've heard the phrase from a semi-cynical friend,

Time: 4198.91

that marriage is the leading cause of divorce,

Time: 4201.62

but now we've heard that divorce,

Time: 4203.292

or the possibility of divorce

Time: 4205.01

could be the leading cause of marriage.

Time: 4206.67

- Of a happy marriage.

Time: 4208.15

- Good point.

Time: 4208.983

- Of a happy marriage.

Time: 4209.96

So, yeah.

Time: 4210.793

So, there's a lot of details there,

Time: 4212.79

but the big dream

Time: 4213.97

is that connection between AI system and a human.

Time: 4217.52

And I haven't.

Time: 4220.3

There's so much fear

Time: 4221.133

about artificial intelligence systems and about robots

Time: 4223.91

that I haven't quite found the right words

Time: 4226.46

to express that vision

Time: 4227.62

because the vision I have,

Time: 4230.146

it's not like some naive, delusional vision

Time: 4233.54

of like technology is going to save everybody,

Time: 4236.47

it's I really do just have a positive view

Time: 4240.34

of ways AI systems can help humans explore themselves.

Time: 4244.73

- I love that positivity

Time: 4246.08

and I agree that the stance everything

Time: 4249.64

is doomed is equally bad

Time: 4254.44

to say that everything's going to turn out all right,

Time: 4256.6

there has to be a dedicated effort.

Time: 4258.33

And clearly, you're thinking

Time: 4260.42

about what that dedicated effort would look like.

Time: 4262.55

You mentioned two aspects to this dream [clears throat]

Time: 4266.63

and I want to make sure

Time: 4267.463

that I understand where they connect if they do,

Time: 4270.07

or if they are independent streams.

Time: 4272.43

One was this hypothetical robot family member,

Time: 4277.31

or some other form of robot

Time: 4279.39

that would allow people to experience the kind of delight

Time: 4283.43

that you experienced many times,

Time: 4287.06

and that you would like the world to be able to have,

Time: 4290.33

and it's such a beautiful idea of this give.

Time: 4293.54

And the other is social media

Time: 4296.19

or social network platforms that really serve individuals

Time: 4300.525

and their best selves and their happiness and their growth.

Time: 4304.2

Is there crossover between those,

Time: 4305.64

or are these two parallel dreams?

Time: 4307.07

- It's 100% of the same thing.

Time: 4308.27

It's difficult to kind of explain

Time: 4310.93

without going through details,

Time: 4312.14

but maybe one easy way to explain

Time: 4314.67

the way I think about social networks,

Time: 4316.137

is to create an AI system that's yours.

Time: 4319.71

It's not like Amazon Alexa that's centralized,

Time: 4323.26

you own the data,

Time: 4324.526

it's like your little friend

Time: 4327.117

that becomes your representative on Twitter

Time: 4331.36

that helps you find things that will make you feel good,

Time: 4336.25

that will also challenge your thinking to make you grow,

Time: 4339.93

but not get to that,

Time: 4342.47

not let you get lost in the negative spiral of dopamine,

Time: 4347.111

that gets you to be angry,

Time: 4349.47

or most just get you to be not open to learning.

Time: 4354.28

And so, that little representative

Time: 4356.27

is optimizing your long-term health.

Time: 4360.41

And I believe that that is not only good for human beings,

Time: 4365.64

it's also good for business.

Time: 4367.4

I think longterm, you can make a lot of money

Time: 4370.52

by challenging this idea that the only way to make money,

Time: 4374.55

is maximizing engagement.

Time: 4377.41

And one of the things that people disagree with me on,

Time: 4379.73

is they think Twitter's always going to win.

Time: 4382.45

Like maximizing engagement is always going to win,

Time: 4384.92

I don't think so.

Time: 4386.34

I think people have woken up now to understanding that,

Time: 4389.73

like they don't always feel good,

Time: 4392.667

the ones who are on Twitter a lot,

Time: 4396.27

that they don't always feel good at the end of the week.

Time: 4399.33

- I would love feedback

Time: 4400.78

from whatever this creature, whatever,

Time: 4404.44

I don't know what to call it,

Time: 4406.77

as to maybe at the end of the week,

Time: 4408.76

it would automatically unfollow

Time: 4410.56

some of the people that I follow

Time: 4411.97

because it realized through some really smart data

Time: 4415.64

about how I was feeling inside, or how I was sleeping,

Time: 4418.09

or something that, I don't know,

Time: 4419.41

that just wasn't good for me.

Time: 4420.65

But it might also put things and people in front of me

Time: 4423.64

that I ought to see,

Time: 4425.55

is that kind of a sliver of what this look like?

Time: 4428.81

- The whole point, because of the interaction,

Time: 4430.59

because of sharing the moments and learning a lot about you,

Time: 4436.65

you're now able to understand what interactions

Time: 4441.13

led you to become a better version of yourself.

Time: 4444.32

Like the person you yourself are happy with.

Time: 4446.849

This isn't,

Time: 4448.57

if you're into flat earth and you feel very good about it,

Time: 4452.85

that you believe that earth is flat,

Time: 4455.76

like the idea that you should sensor that is ridiculous.

Time: 4459.63

If it makes you feel good

Time: 4461.05

and you becoming the best version of yourself,

Time: 4463.47

I think you should be getting

Time: 4464.65

as much flat earth as possible.

Time: 4466.47

Now, it's also good to challenge your ideas,

Time: 4469.37

but not because the centralized committee decided,

Time: 4474.407

but because you tell to the system

Time: 4477.32

that you like challenging your ideas, I think all of us do.

Time: 4480.82

And then, which actually YouTube doesn't do that well,

Time: 4484.09

once you go down the flat-earth rabbit hole,

Time: 4485.96

that's all you're going to see.

Time: 4487.2

It's nice to get some really powerful communicators

Time: 4491.69

to argue against flat earth.

Time: 4493.64

And it's nice to see that for you,

Time: 4497.05

and potentially, at least long-term to expand your horizons,

Time: 4501.15

maybe the earth is not flat.

Time: 4503.29

But if you continue to live your whole life

Time: 4505.21

thinking the earth is flat,

Time: 4506.7

I think, and you're being a good father or son or daughter,

Time: 4511.72

and like you're being the best version of yourself

Time: 4514.38

and you're happy with yourself, I think the earth is flat.

Time: 4518.98

So, like I think this kind of idea,

Time: 4521.43

and I'm just using that kind of silly, ridiculous example

Time: 4523.938

because I don't like the idea of centralized forces

Time: 4530.45

controlling what you can and can't see,

Time: 4533.27

but I also don't like this idea

Time: 4534.98

of like not censoring anything.

Time: 4539.7

Because that's always the biggest problem with that,

Time: 4542.58

is this there's a central decider,

Time: 4545.395

I think you yourself can decide what you want to see and not,

Time: 4549.5

and it's good to have a companion that reminds you

Time: 4554.07

that you felt last time you did this,

Time: 4555.838

or you felt good last time you did this.

Time: 4558.61

- Man, I feel like in every good story,

Time: 4560.15

there's a guide or a companion that flies out,

Time: 4563.57

or forges a little bit further, a little bit differently

Time: 4566.48

and brings back information that helps us,

Time: 4568.46

or at least tries to steer us in the right direction.

Time: 4571.39

- So, yeah, that's exactly what I'm thinking

Time: 4576.64

and what I've been working on.

Time: 4577.85

I should mention as a bunch of difficulties here,

Time: 4580.332

you see me up and down a little bit recently.

Time: 4584.67

So, there's technically a lot of challenges here.

Time: 4588.21

Just like with a lot of technologies,

Time: 4590.367

and the reason I'm talking about it

Time: 4592.75

on a podcast comfortably as opposed to working in secret,

Time: 4596.482

is it's really hard and maybe it's time has not come.

Time: 4601.98

And that's something you have to constantly struggle with

Time: 4604.17

in terms of like entrepreneurially as a startup,

Time: 4608.11

like I've also mentioned to you, maybe offline,

Time: 4610.68

I really don't care about money,

Time: 4612.223

I don't care about business success,

Time: 4615.33

all those kinds of things.

Time: 4618.54

So, it's a difficult decision to make,

Time: 4621.15

how much of your time do you want to go all in here

Time: 4625.29

and give everything to this?

Time: 4628.053

It's a big roll the dice.

Time: 4629.8

Because I've also realized

Time: 4631.58

that's working on some of these problems,

Time: 4634.51

both with the robotics and the technical side

Time: 4638.09

in terms of the machine-learning system that I'm describing,

Time: 4643.16

it's lonely, it's really lonely.

Time: 4646.84

Because both on a personal level and a technical level.

Time: 4651.46

So, on the technical level,

Time: 4652.53

I'm surrounded by people that kind of doubt me,

Time: 4657.94

which I think all entrepreneurs go through.

Time: 4660.56

And they doubt you in the following sense,

Time: 4662.84

they know how difficult it is.

Time: 4666.85

Like the people that the colleagues of mine,

Time: 4669.57

they know how difficult life-long learning is,

Time: 4672.31

they also know how difficult it is

Time: 4674.236

to build a system like this,

Time: 4675.807

to build the competitive social network.

Time: 4679.37

And in general,

Time: 4682.12

there's a kind of a loneliness

Time: 4685.06

to just working on something on your own

Time: 4688.83

for long periods of time.

Time: 4690.29

And you start to doubt whether,

Time: 4693.36

given that you don't have a track record of success,

Time: 4696.35

like that's a big one.

Time: 4697.677

But when you look in the mirror,

Time: 4699.03

especially when you're young,

Time: 4700.55

but I still have that, I'm most things,

Time: 4702.71

you look in the mirror,

Time: 4703.73

it's like, and you have these big dreams,

Time: 4706.688

how do you know

Time: 4708.003

you're actually as smart as you think you are?

Time: 4712.53

Like how do you know

Time: 4713.363

you're going to be able to accomplish this dream?

Time: 4715.46

You have this ambition.

Time: 4716.57

- You sort of don't,

Time: 4717.72

but you're kind of pulling on a string

Time: 4720.95

hoping that there's a bigger ball of yarn.

Time: 4721.897

- Yeah.

Time: 4723.243

[Andrew laughing]

Time: 4724.076

But you have this kind of intuition.

Time: 4725.168

I think I pride myself in knowing what I'm good at,

Time: 4731.066

because the reason I have that intuition,

Time: 4734.48

is 'cause I think I'm very good

Time: 4736.27

at knowing all the things I suck at,

Time: 4738.95

which is basically everything.

Time: 4740.943

So, like whenever I notice like,

Time: 4743.76

wait a minute, I'm kind of good at this,

Time: 4746.26

which is very rare for me.

Time: 4748.21

I think like that,

Time: 4749.1

that might be a ball of yarn worth pulling at.

Time: 4751.55

And the thing with in terms of engineering systems

Time: 4754.81

that are able to interact with humans,

Time: 4756.59

I think I'm very good at that.

Time: 4758.54

And because we talk about podcasting and so on,

Time: 4762.07

I don't know if I'm very good at podcasts.

Time: 4763.59

- You're very good at podcasting, right?

Time: 4765.24

- But I certainly don't,

Time: 4767.22

I think maybe it is compelling for people

Time: 4770.81

to watch a kindhearted idiot struggle with this form,

Time: 4775.4

maybe that's what's compelling.

Time: 4777.24

But in terms of like actual being a good engineer

Time: 4780.63

of human-robot interaction systems, I think I'm good.

Time: 4785.6

But it's hard to know until you do it,

Time: 4788.02

and then the world keeps telling you you're not,

Time: 4791.28

and it's full of doll, it's really hard.

Time: 4793.85

And I've been struggling with that recently,

Time: 4795.37

it's kind of a fascinating struggle.

Time: 4797.22

But then that's where the Goggins thing comes in,

Time: 4799.82

is like, aside from the state-hard motherfucker, is the,

Time: 4805.15

like whenever you're struggling,

Time: 4807.9

that's a good sign that if you keep going,

Time: 4811.21

that you're going to be alone in the success, right?

Time: 4815.881

- Well, in your case, however, I agree.

Time: 4818.7

And actually David had a post recently

Time: 4820.38

that I thought was among his many brilliant posts,

Time: 4823.26

was one of the more brilliant

Time: 4824.88

about how he talked about this myth

Time: 4827.38

of the light at the end of the tunnel.

Time: 4829.177

And instead, what he replaced that myth with,

Time: 4833.054

was a concept that eventually, your eyes adapt to the dark.

Time: 4838.057

That the tunnel, it's not about a light at the end,

Time: 4840.35

that it's really about adapting to the dark of the tunnel.

Time: 4842.553

It's very Goggins-

Time: 4843.523

- I love him so much.

Time: 4844.85

- Yeah.

Time: 4845.683

You got share a lot in common knowing you both a bit,

Time: 4850.511

share a lot in common.

Time: 4852.55

But in this loneliness and the pursuit of this dream,

Time: 4857.88

it seems to me,

Time: 4858.713

it has a certain component to it that is extremely valuable,

Time: 4864.09

which is that the loneliness itself

Time: 4866.38

could serve as a driver

Time: 4868.04

to build the companion for the journey.

Time: 4870.51

- Well, I'm very deeply aware of that.

Time: 4873.71

So, like some people can make,

Time: 4877.009

'cause I talk about love a lot,

Time: 4878.67

I really love everything in this world,

Time: 4881.98

but I also love humans, friendship and romantic,

Time: 4886.66

like even the cheesy stuff, just-

Time: 4891.07

- You like romantic movies?

Time: 4892.84

- Yeah, not those, not necessarily.

Time: 4894.42

So, well, I got so much from Rogan about like,

Time: 4898.48

was it the Tango scene from "Scent of a Woman,"

Time: 4901.13

but I find like there's nothing better

Time: 4904.04

than a woman in a red dress, just like the classy-

Time: 4909.46

- You should move to Argentina my friend.

Time: 4910.6

My father's Argentine, and you know what he said

Time: 4913.12

when I went on your podcast for the first time?

Time: 4915.84

He said, he dresses well.

Time: 4918.38

Because in Argentina,

Time: 4919.22

the men go to a wedding or a party or something,

Time: 4921.52

in the U.S., by halfway through the night,

Time: 4923.71

10 minutes in the night, all the jackets are off.

Time: 4925.63

- Yeah.

Time: 4926.463

- It looks like everyone's undressing for the party

Time: 4927.64

they just got dressed up for.

Time: 4928.94

And he said, you know, I liked the way he dresses.

Time: 4931.64

And then when I started, he was talking about you.

Time: 4933.23

And then when I started my podcast, he said,

Time: 4935.86

why don't you wear a real suit like your friend Lex?

Time: 4938.768

[laughing]

Time: 4940.569

- I remember that.

Time: 4941.402

- No, you can't.

Time: 4943.48

But let's talk about this pursuit just a bit more,

Time: 4947.88

because I think what you're talking about,

Time: 4949.431

is building a, not just a solution for loneliness,

Time: 4953.596

but you've alluded to the loneliness as itself,

Time: 4956.81

an important thing.

Time: 4957.85

And I think you're right.

Time: 4958.683

I think within people,

Time: 4960.046

there is like caverns of faults and shame,

Time: 4965.13

but also just the desire to have resonance,

Time: 4969.5

to be seen and heard.

Time: 4971.25

And I don't even know

Time: 4972.083

that it's seen and heard through language.

Time: 4974.98

But these reservoirs of loneliness,

Time: 4977.66

I think, well, they're interesting,

Time: 4981.27

maybe you could comment a little bit about it.

Time: 4982.66

Because just as often as you talk about love,

Time: 4985.2

I haven't quantified it,

Time: 4986.13

but it seems that you talk about this loneliness,

Time: 4988.48

and maybe you just if you're willing,

Time: 4990.22

you'll share a little bit more about that,

Time: 4992.28

and what that feels like now

Time: 4994.66

in the pursuit of building this robot-human relationship.

Time: 4998.157

And you've been, let me be direct,

Time: 5000.49

you've been spending a lot of time

Time: 5001.96

on building a robot-human relationship, where's that at?

Time: 5008.41

- Oh, in terms of business, in terms of systems?

Time: 5013.45

- No, I'm talking about a specific robot.

Time: 5015.56

- Oh [laughing]

Time: 5017.243

so, okay, I should mention a few things.

Time: 5019.99

So, one, is there's a startup with an idea

Time: 5022.7

where I hope millions of people can use.

Time: 5026.04

And then there's my own personal

Time: 5028.55

like almost like Frankenstein explorations

Time: 5031.646

with particular robots.

Time: 5035.09

So, I'm very fascinated with the legged robots in my own,

Time: 5041.249

private sounds like dark,

Time: 5042.538

but like in of one experiments

Time: 5046.47

to see if I can recreate the magic.

Time: 5049.64

And that's been,

Time: 5051.36

I have a lot of really good already perception systems

Time: 5055.195

and control systems that are able to communicate affection

Time: 5059.28

in a dog-like fashion.

Time: 5060.81

So, I'm in a really good place there.

Time: 5062.55

The stumbling blocks,

Time: 5063.61

which also have been part of my sadness recently,

Time: 5066.785

is that I also have to work with robotics companies

Time: 5070.78

that I gave so much of my heart, soul

Time: 5074.557

and love and appreciation towards Boston Dynamics,

Time: 5077.99

but Boston Dynamics has also,

Time: 5081.97

as a company, it has to make a lot of money

Time: 5083.89

and they have marketing teams.

Time: 5085.6

And they're like looking at this silly Russian kid

Time: 5088.13

in a suit and tie,

Time: 5089.37

it's like, what's he trying to do

Time: 5090.58

with all this love and robot interaction

Time: 5092.65

and dancing and so on?

Time: 5093.52

So, there was a, I think let's say for now,

Time: 5099.05

it's like, when you break up with a girlfriend or something,

Time: 5101.23

right now, we decided to part ways on this particular thing.

Time: 5104.11

They're huge supporters of mine, they're huge fans,

Time: 5105.94

but on this particular thing,

Time: 5108.7

Boston Dynamics is not focusing on,

Time: 5112.68

or interested in human-robot interaction.

Time: 5115.08

In fact, their whole business currently,

Time: 5117.21

is keep the robot as far away from humans as possible

Time: 5120.465

because it's an in the industrial setting

Time: 5123.84

where it's doing monitoring in dangerous environments.

Time: 5127.34

It's almost like a remote security camera,

Time: 5129.47

essentially is its application.

Time: 5132.07

To me, I thought it's still,

Time: 5134.6

even in those applications, exceptionally useful

Time: 5137.15

for the robot to be able to perceive humans, like see humans

Time: 5141.49

and to be able to, in a big map,

Time: 5145.55

localize where those humans are and have human intention.

Time: 5148.44

For example, like this,

Time: 5149.84

I did this a lot of work with pedestrians,

Time: 5151.793

for a robot to be able to anticipate

Time: 5153.88

what the how the human is doing, like where it's walking.

Time: 5157.481

The humans are not ballistics object,

Time: 5159.67

just because you're walking this way one moment,

Time: 5161.64

it doesn't mean you'll keep walking that direction,

Time: 5163.85

you have to infer a lot of signals,

Time: 5165.45

especially with the head movement and the eye movement.

Time: 5167.427

And so, I thought that's super interesting to explore,

Time: 5169.81

but they didn't feel that.

Time: 5171.73

So, I'll be working with a few other robotics companies

Time: 5174.53

that are much more open to that kind of stuff,

Time: 5178.39

and they're super excited and fans of mine.

Time: 5180.7

And hopefully, Boston Dynamics, my first love,

Time: 5182.962

like getting back with an ex-girlfriend, will come around.

Time: 5186.04

But so, the algorithmically, it's basically a done there.

Time: 5192.825

The rest, is actually getting some of these companies

Time: 5196.74

to work with.

Time: 5197.573

And then there's,

Time: 5199.69

for people who'd worked with robots

Time: 5201.33

know that one thing is to write software that works,

Time: 5204.93

and the other is to have a real machine that actually works.

Time: 5208.46

And it breaks down all kinds of different ways

Time: 5210.91

that are fascinating.

Time: 5211.75

And so, there's a big challenge there.

Time: 5213.98

But that's almost,

Time: 5217.05

it may sound a little bit confusing

Time: 5218.8

in the context of our previous discussion

Time: 5220.856

because the previous discussion

Time: 5223.36

was more about the big dream,

Time: 5224.62

how I hoped to have millions of people

Time: 5227.035

enjoy this moment of magic.

Time: 5229.24

This current discussion about a robot,

Time: 5232.19

is something I personally really enjoy,

Time: 5235.14

just brings me happiness,

Time: 5236.32

I really try to do now everything that just brings me joy,

Time: 5240.79

I maximize that 'cause robots are awesome.

Time: 5244.23

But two, given my like little bit growing platform,

Time: 5248.46

I want to use the opportunity to educate people.

Time: 5252.83

Like robots are cool.

Time: 5254.14

And if I think they're cool,

Time: 5256.66

I hope be able to communicate why they're cool to others.

Time: 5259.66

So, this little robot experiment

Time: 5262.19

is a little bit of research project too,

Time: 5264.02

there's a couple of publications with MIT folks around that.

Time: 5267.05

But the other is just to make some cool videos

Time: 5269.82

and explain to people how they actually work.

Time: 5273.34

And as opposed to people being scared of robots,

Time: 5276.373

they could still be scared, but also excited,

Time: 5279.9

like see the dark side, the beautiful side,

Time: 5283.17

the magic of what it means to bring,

Time: 5287.6

for a machine to become a robot.

Time: 5290.12

I want to inspire people with that.

Time: 5292.29

But that's less,

Time: 5293.89

it's interesting because I think the big impact

Time: 5298.17

in terms of the dream does not have to do with embodied AI.

Time: 5302.38

So, it does not need to have a body.

Time: 5304.06

I think the refrigerator's enough,

Time: 5306.82

that for an AI system just to have a voice and to hear you,

Time: 5311.34

that's enough for loneliness.

Time: 5313.45

The embodiment is just-

Time: 5316.59

- By embodiment, you mean the physical structure.

Time: 5318.48

- Physical instantiation of intelligence.

Time: 5321.29

So, it's a legged robot, or even just a thing.

Time: 5326.04

I have a few other of humanoid robot,

Time: 5328.403

a little humanoid robot maybe I'll keep them on the table,

Time: 5331.37

is like walks around,

Time: 5332.88

or even just like a mobile platform

Time: 5334.77

they can just like, turn around and look at you,

Time: 5337.878

it's like we mentioned with the pen.

Time: 5339.23

Something that moves and can look at you,

Time: 5342.076

it's like that Butter Robot that asks, what is my purpose?

Time: 5351.371

That is really, it's almost like art.

Time: 5355.68

There's something about a physical entity that moves around,

Time: 5359.71

that's able to look at you and interact with you,

Time: 5362.137

that makes you wonder what it means to be human.

Time: 5365.97

It like challenges you to think,

Time: 5367.54

if that thing looks like he has consciousness,

Time: 5372.58

what the hell am I?

Time: 5373.97

And I like that feeling,

Time: 5375.14

I think that's really useful for us,

Time: 5376.55

it's humbling for us humans.

Time: 5378.02

But that's less about research,

Time: 5381.13

it certainly less about business

Time: 5382.7

and more about exploring our own selves,

Time: 5386.01

and challenging others to think like,

Time: 5389.26

to think about what makes them human.

Time: 5392.93

- I love this desire

Time: 5394.81

to share the delight of an interaction with a robot.

Time: 5398.07

And as you describe it,

Time: 5398.93

I actually find myself starting to crave that

Time: 5401.37

because we all have those elements from childhood where,

Time: 5404.96

or from adulthood, where we experience something,

Time: 5406.81

we want other people to feel that.

Time: 5409.84

And I think that you're right,

Time: 5410.96

I think a lot of people are scared of AI,

Time: 5412.59

I think a lot of people are scared of robots.

Time: 5414.49

My only experience, and of a robotic-like thing

Time: 5418.498

is my Roomba Vacuum, where it goes about,

Time: 5421.89

actually, it was pretty good at picking up Costello's hair

Time: 5424.3

when he was shed and I was grateful for it.

Time: 5428.3

But then when I was on a call or something,

Time: 5430.027

and it would get caught on a wire or something,

Time: 5433.22

I would find myself getting upset with the Roomba

Time: 5435.31

in that moment, I'm like, what are you doing?

Time: 5437.07

And obviously it's just doing what it does.

Time: 5439.92

But that's a kind of mostly positive,

Time: 5442.51

but slightly negative interaction.

Time: 5445.53

But what you're describing,

Time: 5446.7

it has so much more richness and layers of detail

Time: 5450.131

that I can only imagine what those relationships are like.

Time: 5453.71

- Well, there's a few, just a quick comment.

Time: 5455.43

So, I've had, they're currently in Boston

Time: 5457.64

and I have a bunch of Roombas from iRobot.

Time: 5460.59

And I did this experiment-

Time: 5461.93

- Wait, how many Roombas?

Time: 5463.891

[Lex laughing]

Time: 5464.984

It sounds like a fleet of Roombas.

Time: 5465.96

- Yeah.

Time: 5467.037

So, it's probably seven or eight, yeah.

Time: 5469.03

- Well, it's a lot of Roombas.

Time: 5469.863

This place is very clean.

Time: 5472.82

- Well, so, this, I'm kind of waiting,

Time: 5474.78

this is the place we're currently in Austin,

Time: 5478.46

is way larger than I need.

Time: 5481.48

But I basically got it to make sure I have room for robots.

Time: 5487.21

- So, you have these seven or so Roombas,

Time: 5490.45

you deploy all seven at once?

Time: 5492.13

- Oh, no, I do different experience with them,

Time: 5493.916

different experiments with them.

Time: 5496.06

So, one of the things I want to mention, is this is a,

Time: 5498.87

I think there was a YouTube video

Time: 5500.18

that inspired me to try this,

Time: 5502.83

is I got them to scream in pain and moan in pain

Time: 5508.77

whenever they were kicked or contacted.

Time: 5513.43

And I did that experiment to see how I would feel.

Time: 5517.11

I meant to do like a YouTube video on it,

Time: 5518.84

but then it just seemed very cruel.

Time: 5520.4

- Did any Roomba rights activists come at you?

Time: 5522.42

- Like I think if I released that video,

Time: 5527.66

I think it's going to make me look insane,

Time: 5529.73

which I know people know I'm already insane-

Time: 5532.78

- Now, you have to release the video.

Time: 5533.866

[Lex laughing]

Time: 5534.699

- Sure.

Time: 5535.532

Well, I think maybe if I contextualize it

Time: 5538.33

by showing other robots

Time: 5540.42

like to show why this is fascinating,

Time: 5543.24

because ultimately,

Time: 5544.34

I felt like there were human almost immediately.

Time: 5547.4

And that display of pain was what did that.

Time: 5550.38

- Giving them a voice.

Time: 5551.53

- Giving them a voice,

Time: 5552.4

especially a voice of dislike, of pain.

Time: 5557.21

- I have to connect you to my friend Eddie Chang,

Time: 5559.25

he studied speech and language, he's a neurosurgeon,

Time: 5561.4

and we're life-long friends.

Time: 5563.79

He studied speech and language,

Time: 5566.39

but he describes some of these more primitive,

Time: 5570.19

visceral vocalizations, cries, groans, moans of delight,

Time: 5576.47

other sounds as well, use your imagination,

Time: 5578.99

as such powerful rudders for the other,

Time: 5582.87

for the emotions of other people.

Time: 5584.143

And so, I find it fascinating,

Time: 5586.12

I can't wait to see this video.

Time: 5587.61

So, is the video available online?

Time: 5589.52

- No, I haven't recorded it,

Time: 5592.08

I just hit a bunch of Roombas

Time: 5593.71

that are able to scream in pain in my Boston place.

Time: 5598.252

[Andrew laughing]

Time: 5599.41

So, like people already-

Time: 5602.37

- Next podcast episode with Lex,

Time: 5604.87

maybe we'll have that one, who knows?

Time: 5606.8

- So, the thing is like people,

Time: 5608.28

I've noticed because I talk so much about love

Time: 5611.01

and it's really who I am,

Time: 5612.41

I think they want to,

Time: 5614.164

to a lot of people,

Time: 5615.72

seems like there's there got to be a dark person

Time: 5618.57

in there somewhere.

Time: 5619.48

And I thought if I release videos of Roombas screaming

Time: 5621.93

and they're like, yep, yep, that guy's definitely insane.

Time: 5624.59

- What about like shouts of glee and delight,

Time: 5627.81

you could do that too, right?

Time: 5628.78

- Well, I don't know how to,

Time: 5631.67

to me, delight is quiet, right?

Time: 5634.85

- You're a Russian.

Time: 5636.325

Americans are much louder than Russians.

Time: 5639.89

- Yeah. - Yeah.

Time: 5640.723

- Yeah.

Time: 5641.556

But like I don't, unless you're talking about like,

Time: 5644.37

I don't know

Time: 5645.203

how you would have sexual relationships with the Roomba.

Time: 5647.12

- Well, I wasn't necessarily saying a sexual delight, but-

Time: 5650.082

- Trust me, I tried.

Time: 5651.356

I'm just kidding, that's a joke, internet.

Time: 5654.13

Okay [giggles]

Time: 5654.963

but I was fascinated in the psychology

Time: 5656.81

of how little it took.

Time: 5657.75

'Cause you mentioned

Time: 5658.73

you had a negative relationship with a Roomba.

Time: 5661.521

- Well, I'd find that mostly, I took it for granted.

Time: 5664.93

- Yeah.

Time: 5665.763

- It just served me, it collected Costello's hair.

Time: 5667.58

And then when it would do something I didn't like,

Time: 5669.23

I would get upset with it.

Time: 5670.27

So, that's not a good relationship, it was taken for granted

Time: 5673.41

and I would get upset and then I'd park it again,

Time: 5676.09

and I just like you're in the corner.

Time: 5679.06

- Yeah.

Time: 5679.893

- But there's a way to frame it being quite dumb

Time: 5685.41

as almost cute.

Time: 5687.72

You're almost connecting with it for its dumbness.

Time: 5691.16

And I think that's artificial intelligence problem.

Time: 5694.752

- [Andrew] It's interesting.

Time: 5695.585

- I think flaws should be feature, not a bug.

Time: 5699.33

- So, along the lines of this,

Time: 5701.18

the different sorts of relationships

Time: 5702.59

that one could have with robots,

Time: 5703.607

and the fear, but also some of the positive relationships

Time: 5706.94

that one could have,

Time: 5708.93

there's so much dimensionality, there's so much to explore.

Time: 5712.44

But power dynamics in relationships are very interesting,

Time: 5716.34

because the obvious ones,

Time: 5718.822

the unsophisticated view of this,

Time: 5721.04

is one, there's a master and a servant, right?

Time: 5725.07

But there's also manipulation,

Time: 5727.84

there's benevolent manipulation,

Time: 5730.86

children do this with parents, puppies do this.

Time: 5733.16

Puppies turn their head and look cute

Time: 5734.96

and maybe give out a little noise,

Time: 5737.6

kids coup and parents always think that they're doing this

Time: 5741.58

because they love the parent,

Time: 5745.01

but in many ways,

Time: 5746.58

studies show that those coups are ways

Time: 5748.5

to extract the sorts of behaviors and expressions

Time: 5750.86

from the parent that they want.

Time: 5751.85

The child doesn't know it's doing this,

Time: 5753.2

it's completely subconscious,

Time: 5754.29

but it's benevolent manipulation.

Time: 5756.5

So, there's one version of fear of robots

Time: 5759.75

that I hear a lot about

Time: 5761.043

that I think most people can relate to

Time: 5762.75

where the robots take over

Time: 5764.57

and they become the masters and we become the servants.

Time: 5768.2

But there could be another version

Time: 5770.04

that in certain communities

Time: 5773.68

that I'm certainly not a part of,

Time: 5775.17

but they call topping from the bottom,

Time: 5777.25

where the robot is actually manipulating you

Time: 5780.84

into doing things,

Time: 5782.39

but you are under the belief that you are in charge,

Time: 5786.78

but actually they're in charge.

Time: 5789.24

And so, I think that's one

Time: 5791

that if we could explore that for a second,

Time: 5794.56

you could imagine it wouldn't necessarily be bad,

Time: 5797.19

although it could lead to bad things.

Time: 5800.17

The reason I want to explore this,

Time: 5801.12

is I think people always default to the extreme,

Time: 5804.33

like the robots take over and we're in little jail cells

Time: 5807.07

and they're out having fun and ruling the universe.

Time: 5810.354

What sorts of manipulation

Time: 5813.05

can a robot potentially carry out, good or bad?

Time: 5816.55

- Yeah.

Time: 5817.383

Just so, there's a lot of good and bad manipulation

Time: 5819.83

between humans, right?

Time: 5820.81

Just like you said.

Time: 5823.354

[Lex sighing]

Time: 5824.44

To me [sighing] especially like you said [giggles]

Time: 5829.97

topping from the bottom, is that the term?

Time: 5831.7

- I think someone from MIT told me that term.

Time: 5834.551

[Lex laughing]

Time: 5835.815

It wasn't Lex.

Time: 5837.86

- I think.

Time: 5838.693

So, first of all, there's power dynamics in bed,

Time: 5842.08

and power dynamics in relationships,

Time: 5844.25

and power dynamics on the street,

Time: 5845.9

and in the work environment, those are all very different.

Time: 5849.359

I think power dynamics can make human relationships,

Time: 5854.78

especially romantic relationships fascinating and rich

Time: 5859.88

and fulfilling and exciting and all those kinds of things.

Time: 5862.78

So, I don't think in themselves they're bad,

Time: 5868.67

and the same goes with robots.

Time: 5870.86

I really love the idea

Time: 5872.106

that a robot will be at top or a bottom

Time: 5875

in terms of like power dynamics.

Time: 5877.54

And I think everybody should be aware of that.

Time: 5880.02

And the manipulation is not so much manipulation,

Time: 5882.49

but a dance of like pulling away, a push and pull,

Time: 5886.74

and all those kinds of things.

Time: 5888.003

In terms of control,

Time: 5890.4

I think we're very, very, very far away from AI systems

Time: 5893.95

that are able to lock us up.

Time: 5897.76

To lock us up in,

Time: 5900.23

like to have so much control

Time: 5901.67

that we basically cannot live our lives

Time: 5905.17

in the way that we want.

Time: 5906.95

I think there's a, in terms of dangers of AI systems,

Time: 5909.73

there's much more dangers

Time: 5910.87

that have to do with autonomous weapon systems

Time: 5913.087

and all those kinds of things.

Time: 5914.28

So, the power dynamics as exercised in the struggle

Time: 5917.95

between nations and war and all those kinds of things.

Time: 5920.7

But in terms of personal relationships,

Time: 5923.71

I think power dynamics are a beautiful thing.

Time: 5925.95

Now, there's of course,

Time: 5927.21

going to be all those kinds of discussions

Time: 5929.61

about consent and rights and all those kinds of things.

Time: 5932.66

- Well, here, we're talking about,

Time: 5934

I always say in any discussion around this,

Time: 5936.47

if we need to define a really the context,

Time: 5939.21

it always should be consensual, age-appropriate,

Time: 5943.32

context-appropriate, species-appropriate,

Time: 5946.11

but now we're talking about human-robot interactions.

Time: 5949.25

And so, I guess that-

Time: 5950.82

- No, I actually was trying to make a different point,

Time: 5953.66

which is I do believe

Time: 5955.02

that robots will have rights down the line.

Time: 5957.88

And I think in order for us

Time: 5960.86

to have deep, meaningful relationships with robots,

Time: 5963.1

we would have to consider them as entities in themselves

Time: 5966.58

that deserve respect.

Time: 5969.77

And that's a really interesting concept

Time: 5971.277

that I think people are starting to talk about

Time: 5974.38

a little bit more,

Time: 5975.52

but it's very difficult for us to understand

Time: 5977.66

how entities that are other than human.

Time: 5979.75

I mean, the same as with dogs and other animals

Time: 5982.51

can have rights on a level as humans.

Time: 5984.94

- Well, yeah.

Time: 5985.86

We can't and nor should we do whatever we want with animals,

Time: 5989.98

we have a USDA, we have Department of Agriculture

Time: 5994.74

that deal with animal care and use committees for research,

Time: 5999.05

for farming and ranching and all that.

Time: 6002.11

So, when you first said it, I thought, wait, why would,

Time: 6005.94

there'll be a bill of robotic rights,

Time: 6007.41

but it absolutely makes sense

Time: 6008.833

in the context of everything we've been talking about

Time: 6012.23

up until now.

Time: 6015.293

If you're willing, I'd love to talk about dogs,

Time: 6018.74

because you've mentioned dogs a couple of times,

Time: 6021.31

a robot dog, you had a biological dog.

Time: 6026.08

Yeah. - Yeah.

Time: 6026.913

I had a Newfoundland named Homer for many years growing up.

Time: 6034.27

- In Russia or in the U.S.?

Time: 6035.64

- In the United States.

Time: 6037.33

And he was about, he's over 200 pounds, that's a big dog.

Time: 6040.97

- That's a big dog.

Time: 6042.04

- People know Newfoundland.

Time: 6043.76

So, he's this black dog that's a really a long hair

Time: 6048.62

and just a kind soul.

Time: 6050.5

I think perhaps that's true for a lot of large,

Time: 6053.27

but he thought he was a small dog.

Time: 6055.23

So, he moved like that, and-

Time: 6056.52

- Was he your dog?

Time: 6057.43

- Yeah, yeah.

Time: 6058.263

- So, you had him since he was fairly young?

Time: 6060.89

- Since the very, very beginning, till the very, very end.

Time: 6063.81

And one of the things, I mean, he had this kind of a,

Time: 6068.66

we mentioned like the Roombas,

Time: 6071.33

he had kind-hearted dumbness about him

Time: 6075.156

that was just overwhelming,

Time: 6076.71

it's part of the reason I named him Homer,

Time: 6080.06

because it's after Homer Simpson,

Time: 6082.52

in case people are wondering which Homer I'm referring to.

Time: 6087.041

[Andrew laughing]

Time: 6087.874

And so, there's a.

Time: 6088.97

Yeah, exactly.

Time: 6092.33

There's a clumsiness that was just something

Time: 6095.24

that immediately led to a deep love for each other.

Time: 6097.685

And one of the,

Time: 6100.685

I mean, he was always, it's the shared moments,

Time: 6103.37

he was always there for so many a nights together,

Time: 6106.55

that's a powerful thing about a dog

Time: 6108.64

that he was there through all the loneliness,

Time: 6112.35

through all the tough times, through the successes

Time: 6114.44

and all those kinds of things.

Time: 6115.87

And I remember,

Time: 6117.55

I mean, that was a really moving moment for me,

Time: 6120.15

I still miss him to this day.

Time: 6122.19

- How long ago did he die?

Time: 6125.4

- Maybe 15 years ago.

Time: 6127.1

So, it's been awhile.

Time: 6129.17

But it was the first time

Time: 6131.964

I've really experienced like the feeling of death.

Time: 6137.01

So, what happened, is he got a cancer.

Time: 6142.87

And so, he was dying slowly.

Time: 6146.06

And then there's a certain point he couldn't get up anymore.

Time: 6149.396

Now, there's a lot of things that could say here

Time: 6152.71

that I struggle with.

Time: 6154.184

Maybe he suffered much longer than he needed to,

Time: 6159.12

that's something I really think about a lot.

Time: 6162.08

But I remember when I had to take him to the hospital

Time: 6167.2

and the nurses couldn't carry him, right?

Time: 6172.35

So, you're talking about a 200 pound dog

Time: 6174.12

and I was really into power lifting at the time.

Time: 6176.73

And I remember they tried to figure out

Time: 6179.39

all these kinds of ways to.

Time: 6182.015

So, in order to put them to sleep,

Time: 6183.62

they had to take them into a room.

Time: 6187.27

And so, I had to carry him everywhere.

Time: 6189.54

And here's this dying friend of mine that I just had to,

Time: 6195.26

first of all, that was really difficult to carry,

Time: 6197.02

somebody that heavy when they're not helping you out.

Time: 6199.791

And, yeah.

Time: 6202.76

So, I remember it was the first time

Time: 6205.96

seeing a friend laying there

Time: 6208.47

and seeing life drain from his body.

Time: 6214.138

And that realization that we're here for a short time

Time: 6218.39

was made so real,

Time: 6220.59

that here is friend that was there for me the week before,

Time: 6223.75

the day before, and now he's gone.

Time: 6226.27

And that was,

Time: 6228.5

I don't know, that spoke to the fact

Time: 6230.8

that he could be deeply connected with the dog.

Time: 6234.56

Also spoke to the fact that the shared moments together

Time: 6240.86

that led to that deep friendship will make life so amazing,

Time: 6248.56

but it also spoke to the fact that death is a motherfucker.

Time: 6253.44

So, I know you've lost Costello recently.

Time: 6255.87

- Yeah.

Time: 6256.872

- And you can- - And as you're saying this,

Time: 6257.705

I'm definitely fighting back the tears.

Time: 6260.09

I thank you for sharing that.

Time: 6263.36

That I guess we're about to both cry over,

Time: 6266.19

I don't want to say dogs [laughing]

Time: 6268.69

that it was bound to happen

Time: 6270.34

just given when this is happening.

Time: 6273.53

Yeah, it's-

Time: 6275.266

- How long did you know that Costello was not doing well?

Time: 6279.7

- We'll, let's see, a year ago during the start of,

Time: 6284.31

about six months into the pandemic,

Time: 6286.48

he started getting abscesses and he was not,

Time: 6289.11

his behavior change and something really changed.

Time: 6291.62

And then I put him on testosterone

Time: 6296.09

which helped a lot of things.

Time: 6298.14

It certainly didn't cure everything,

Time: 6299.33

but it helped a lot of things,

Time: 6300.66

he was dealing with joint pain, sleep issues,

Time: 6303.92

and then it just became a very slow decline

Time: 6308.91

to the point where two, three weeks ago,

Time: 6311.41

he had a closet full of medication.

Time: 6315.37

I mean, this dog was, it was like a pharmacy.

Time: 6317.39

It's amazing to me when I looked at it the other day,

Time: 6319.6

I still haven't cleaned up and removed all those things,

Time: 6322.12

'cause I can't quite bring myself to do it, but-

Time: 6325.961

- Do you think he was suffering?

Time: 6327.51

- Well, so, what happened, was about a week ago,

Time: 6330.33

it was really just about a week ago, it's amazing.

Time: 6332.163

He was going up the stairs, I saw him slip.

Time: 6335.08

And he was a big guy,

Time: 6335.913

he wasn't 200 pounds, but he was about 90 pounds,

Time: 6337.86

he's a bulldog, he was pretty big and he was fit.

Time: 6341.24

And then I noticed

Time: 6342.073

that he wasn't carrying a foot in the back

Time: 6344.45

like it was injured, it had no feeling at all.

Time: 6346.44

He never liked me to touch his hind paws, and I could do,

Time: 6348.92

that thing was just flopping there.

Time: 6350.52

And then the vet found some spinal degeneration

Time: 6353.89

and I was told that the next one would go.

Time: 6355.76

Did he suffer?

Time: 6356.863

Sure, hope not, but something changed in his eyes.

Time: 6360.96

- Yeah. - Yeah, it's the eyes again,

Time: 6362.84

I know you and I spend long hours on the phone

Time: 6365.16

and tell you about like the eyes

Time: 6366.41

and what they convey and what they mean

Time: 6368.05

about internal states

Time: 6369.23

and for sake of robots and biology of other kinds, but-

Time: 6372.797

- You think something about him was gone in his eyes?

Time: 6377.92

- I think he was real,

Time: 6380.66

here, I am anthropomorphizing,

Time: 6382.513

I think he was realizing that one of his great joys in life,

Time: 6386.87

which was to walk and sniff and pee on things.

Time: 6392.167

[Lex laughing]

Time: 6393.15

This dog.

Time: 6393.983

- The fundamental. - Loved to pee on things,

Time: 6396.057

it was amazing.

Time: 6396.89

I've wondered where he put it,

Time: 6398.94

he was like a reservoir of urine, it was incredible.

Time: 6402.33

I think, oh, that's Eddie,

Time: 6403.55

he'd put like one drop on the 50 millionth plant.

Time: 6406.59

And then we get to the 50 millionth in one plant,

Time: 6409.017

and he just have, leave a puddle.

Time: 6411.26

And here I am talking about Costello peeing.

Time: 6414.54

He was losing that ability to stand up and do that,

Time: 6417.08

he was falling down while he was doing that.

Time: 6418.92

And I do think he started to realize.

Time: 6421.6

And the passage was easy and peaceful,

Time: 6424.6

but I'll say this, I'm not ashamed to say it,

Time: 6428.803

I mean, I wake up every morning since then just,

Time: 6431.214

I don't even make the conscious decision

Time: 6433.27

to allow my self to cry, I wake up crying.

Time: 6436.11

And I'm unfortunately able to make it through the day,

Time: 6438.2

thanks to the great support of my friends

Time: 6440.01

and you and my family, but I miss him, man.

Time: 6443.847

- You miss him?

Time: 6444.68

- Yeah, I miss him.

Time: 6445.56

And I feel like, Homer, Costello,

Time: 6449.44

the relationship to one's dog is so specific part.

Time: 6454.14

- So, that part of you is gone?

Time: 6457.97

- That's the hard thing.

Time: 6465.12

What I think is different,

Time: 6466.98

is that I made the mistake, I think.

Time: 6469.807

Moreover, I hope it was a good decision,

Time: 6471.85

but sometimes I think I made the mistake

Time: 6473.3

of I brought Costello a little bit to the world

Time: 6476.52

through the podcast or posting about him,

Time: 6479

I anthropomorphized about him in public.

Time: 6481.58

Let's be honest, I have no idea what his mental life was,

Time: 6483.73

or his relationship to me.

Time: 6484.91

And I'm just exploring all this for the first time

Time: 6486.397

'cause he was my first dog,

Time: 6487.79

but I raised him since he was seven weeks.

Time: 6489.71

- Yeah, you got hold it together,

Time: 6491.05

I noticed the episode you released on Monday,

Time: 6494.97

you mentioned Costello.

Time: 6496.61

Like you brought them back to life for me

Time: 6498.95

for that brief moment.

Time: 6500.35

- Yeah.

Time: 6501.401

But he's gone.

Time: 6502.42

- That's the,

Time: 6504.66

he's going to be gone for a lot of people too.

Time: 6508.53

- Well, this is what I'm struggling with.

Time: 6509.86

I think that maybe you're pretty good at this.

Time: 6513.42

Like, have you done this before?

Time: 6516.083

[Andrew laughing]

Time: 6517.977

This is the challenge.

Time: 6518.81

Is that actually part of me.

Time: 6520.6

I know how to take care of myself pretty well.

Time: 6522.57

- Yeah.

Time: 6523.403

- Not perfectly, but pretty well.

Time: 6524.76

And I have good support.

Time: 6526.04

I do worry a little bit about how it's going to land

Time: 6528.81

and how people will feel.

Time: 6530.23

I'm concerned about their internalization.

Time: 6534.121

So that's something I'm still iterating on.

Time: 6536.3

- And you have to watch you struggle,

Time: 6538.4

which is fascinating.

Time: 6539.4

- Right.

Time: 6540.233

And I've mostly been shielding them from this,

Time: 6541.66

but what would make me happiest

Time: 6544.01

is if people would internalize

Time: 6546.42

some of Costello's best traits.

Time: 6548.24

And his best traits were that he was incredibly tough.

Time: 6554.26

I mean, he was a,

Time: 6555.84

22-inch-neck bulldog, the whole thing.

Time: 6557.72

He was just born that way.

Time: 6558.999

What was so beautiful is that his toughness

Time: 6562.42

has never what he rolled forward.

Time: 6564.1

It was just how sweet and kind he was.

Time: 6567.14

And so if people can take that,

Time: 6569.18

then there's a win in there someplace, so.

Time: 6573.2

- I think there's some ways

Time: 6574.65

in which she should probably live on in your podcast too.

Time: 6577.607

You should,

Time: 6579.1

I mean, it's such a,

Time: 6581.83

one of the things I loved about his role in your podcast

Time: 6585.83

is that he brought so much joy to you.

Time: 6588.64

We mentioned the robots.

Time: 6589.925

- Mm-hmm.

Time: 6590.758

- Right?

Time: 6591.633

I think that's such a powerful thing

Time: 6594.35

to bring that joy into,

Time: 6596.33

like, allowing yourself to experience that joy,

Time: 6599.42

to bring that joy to others, to share it with others.

Time: 6602.48

That's really powerful.

Time: 6603.85

And I mean, not to,

Time: 6605.64

this is like the Russian thing,

Time: 6607.21

is [chuckles] it touched me when Lucy Kay had that moment

Time: 6614.08

that I keep thinking about in this show "Louie,"

Time: 6618

or like an old man was criticizing Louis

Time: 6620.29

for whining about breaking up with his girlfriend,

Time: 6622.63

and he was saying like the most beautiful thing about love,

Time: 6630.46

they made a song that's catchy

Time: 6632.34

now that's now making me feel horrible saying it,

Time: 6635.69

but like, is the loss.

Time: 6637.74

The loss really also is making you realize

Time: 6641.26

how much that person,

Time: 6644.49

that dog meant to you.

Time: 6646.37

And like allowing yourself to feel that loss

Time: 6648.74

and not run away from that loss is really powerful.

Time: 6651.45

And in some ways that's also sweet,

Time: 6655.1

just like the love was the loss is also sweet

Time: 6657.735

because you know that you felt a lot for that

Time: 6662.085

through your friend.

Time: 6663.81

So I, and like continue to bring that joy.

Time: 6666.812

I think it would be amazing to the podcast.

Time: 6670.435

I hope to do the same with [laughing] robots,

Time: 6673.71

or whatever else is the source of joy, right?

Time: 6677.75

And maybe, do you think about one day getting another dog?

Time: 6682.49

- Yeah, in time.

Time: 6685.06

You're hitting on all the key buttons here.

Time: 6688.61

I want that to,

Time: 6690.09

we're thinking about ways to kind of immortalize Costello

Time: 6694.78

in a way that's real,

Time: 6695.66

not just creating some little logo or something silly.

Time: 6699.95

Costello much like David Goggins is a person,

Time: 6703.88

but Goggins also has grown into kind of a verb.

Time: 6707.19

You're going to Goggins this or you're going to,

Time: 6708.87

and there's an adjective.

Time: 6710.15

Like that's extreme.

Time: 6711.06

Like it,

Time: 6712.46

I think that for me, Costello was all those things.

Time: 6714.49

He was a being, he was his own being,

Time: 6716.85

he was a noun, a verb and an adjective.

Time: 6720.38

So, and he had this amazing super power

Time: 6722.47

that I wish I could get,

Time: 6723.34

which is this ability to get everyone else

Time: 6725.09

to do things for you without doing a thing.

Time: 6727.893

[Lex laughing]

Time: 6728.726

The Costello effect,

Time: 6729.83

as I call it.

Time: 6730.663

- So is an idea I hope he lives on.

Time: 6732.29

- Yes.

Time: 6733.81

Thank you for that.

Time: 6734.643

This actually has been very therapeutic for me,

Time: 6738.1

which actually brings me to a question,

Time: 6742.38

we're friends, we're not just co-scientists,

Time: 6745.87

colleagues working on a project together

Time: 6748.22

and in the world,

Time: 6753.33

that's somewhat similar.

Time: 6754.8

- Just two dogs.

Time: 6757.23

- Just two dogs basically.

Time: 6760.57

But let's talk about friendship,

Time: 6764.26

because I think that I certainly know as a scientist

Time: 6769.19

that there are elements that are very lonely

Time: 6771.15

of the scientific pursuit.

Time: 6772.78

There are elements of many pursuits that are lonely.

Time: 6777.03

Music,

Time: 6778.35

Math always seem to me like they're like

Time: 6780.76

the loneliest people.

Time: 6782.125

Who knows if that's true or not.

Time: 6783.657

Also people work in teams.

Time: 6785.41

And sometimes people are surrounded

Time: 6786.63

by people interacting with people and they feel very lonely.

Time: 6789.33

But for me, and I think as well for you,

Time: 6794.47

friendship is an incredibly strong force

Time: 6797.96

in making one feel like certain things are possible

Time: 6803.56

or worth reaching for,

Time: 6805.82

maybe even making us compulsively reach for them.

Time: 6808.25

So, when you were growing up,

Time: 6810.75

you grew up in Russia until what age?

Time: 6812.82

- 13.

Time: 6813.73

- Okay.

Time: 6814.563

And then you moved directly to Philadelphia?

Time: 6818.32

- To Chicago.

Time: 6819.33

- [Andrew] Chicago.

Time: 6820.163

- And then Philadelphia,

Time: 6822.43

and San Francisco and Boston and so on.

Time: 6825.06

But really to Chicago, that's where I went to high school.

Time: 6828.48

- Do you have siblings?

Time: 6829.88

- Older brother.

Time: 6830.73

- But most people don't know that.

Time: 6831.851

[Lex laughing]

Time: 6833.87

- Yeah, he is a very different person.

Time: 6838.01

But somebody I definitely look up to.

Time: 6839.64

So he's a wild man.

Time: 6840.81

He's extrovert,

Time: 6841.86

he was into, I mean,

Time: 6845.06

so he's also scientists, a bio engineer,

Time: 6847.58

but he's, when we were growing up,

Time: 6849.826

he was the person who did,

Time: 6854.91

drank and did every drug.

Time: 6856.92

And, but also as the life of the party.

Time: 6859.4

And I just thought he was the,

Time: 6861.12

when your older brother, five years older,

Time: 6863.48

he was the coolest person that I was wanting to be him.

Time: 6868.34

So for that,

Time: 6869.92

he definitely had a big influence.

Time: 6871.23

But I think for me in terms of friendship, growing up,

Time: 6875.926

I had one really close friend.

Time: 6880.09

And then when I came here I had another close friend,

Time: 6882.33

but I'm very,

Time: 6884.36

I believe,

Time: 6886.255

I don't know if I believe,

Time: 6887.56

but I draw a lot of strength from deep connections

Time: 6893.06

with other people,

Time: 6896.2

and just the small number of people.

Time: 6897.78

Just a really small number of people.

Time: 6899.09

That's when I moved to this country,

Time: 6900.53

I was really surprised how,

Time: 6902.29

like, there were

Time: 6903.366

these large groups of friends, quote, unquote.

Time: 6908.33

But the depth of connection was not there at all

Time: 6912.28

from my sort of perspective.

Time: 6914.3

Now, I moved to the suburb of Chicago, was Naperville.

Time: 6917.74

It's more like a middle class, maybe upper middle class.

Time: 6921.01

So it's like people that cared more

Time: 6923.58

about material possessions than deep human connection.

Time: 6926.51

So that added to the thing.

Time: 6928.44

But I drove more meaning than almost anything else

Time: 6934.05

was from friendship.

Time: 6935.05

Early on I had a best friend.

Time: 6937.22

His name was,

Time: 6938.41

his name is Yura.

Time: 6941.121

I don't know how to say it in English.

Time: 6943.47

- How do you say in Russian?

Time: 6944.72

- Yura.

Time: 6945.553

What's his last name?

Time: 6946.47

Do you remember if was...

Time: 6947.473

[Lex chuckles]

Time: 6948.93

- Mikolov.

Time: 6950.928

Yura Mikolov.

Time: 6953.75

So we just spent all our time together.

Time: 6956.37

There's also a group of friends.

Time: 6958.51

Like, I dunno, it's like eight guys.

Time: 6959.611

In Russia growing up,

Time: 6964.87

it's like parents didn't care

Time: 6967.51

if you're coming back at certain hour.

Time: 6969.7

So we'll spent all day, all night just playing soccer,

Time: 6973.38

usually called football,

Time: 6976.006

and just talking about life and all those kinds of things,

Time: 6978.94

even at that young age.

Time: 6980.82

I think people in Russia

Time: 6982.74

and Soviet Union grew up much quicker.

Time: 6985.234

[Lex chuckles]

Time: 6986.536

I think the education system at the university level

Time: 6990.93

is world-class in the United States in terms of like,

Time: 6994.59

really creating really big, powerful minds.

Time: 6998.98

At least they used to be,

Time: 7000.09

but I think that they aspire to that.

Time: 7001.715

But the education system for like,

Time: 7004.81

for younger kids in the Soviet Union was incredible.

Time: 7009.07

Like they did not treat us as kids.

Time: 7011.127

The level of literature, Tolstoy, Dostoevsky.

Time: 7014.798

- When you were a small child?

Time: 7016

- Yeah.

Time: 7016.926

- Amazing.

Time: 7017.759

Amazing.

Time: 7018.592

- And like the level of mathematics,

Time: 7020.44

and you're made to feel like shit

Time: 7022.11

if you're not good at mathematics.

Time: 7023.81

Like we, I think in this country,

Time: 7026.46

there's more like,

Time: 7027.4

especially young kids

Time: 7028.24

'cause they're so cute.

Time: 7029.74

Like they're being babied.

Time: 7032.38

We only start to really push adults later in life.

Time: 7035.28

Like, so if you want to be the best in the world at this,

Time: 7037.89

then you get to be pushed.

Time: 7039.51

But we were pushed at a young age.

Time: 7041.81

Everybody was pushed.

Time: 7042.95

And that brought out the best in people.

Time: 7045.06

I think it really forced people to discover,

Time: 7049.06

like discover themselves in the Goggin style,

Time: 7051.89

but also discover what they're actually passionate about,

Time: 7054.983

what they're not.

Time: 7055.816

- Is this true for boys and girls?

Time: 7057.49

Were they pushed equally there?

Time: 7059.01

- Yeah.

Time: 7059.843

They were pushed.

Time: 7060.676

Yeah, they were pushed equally, I would say.

Time: 7061.99

There was a,

Time: 7063.03

obviously there was more,

Time: 7064.93

not obviously,

Time: 7065.763

but at least from my memories,

Time: 7068.63

more of a,

Time: 7070.4

what's the right way to put it?

Time: 7071.725

But there was like gender roles,

Time: 7073.91

but not in a negative connotation.

Time: 7076.65

It was the red dress versus the suit and tie

Time: 7079.52

kind of connotation, which is like,

Time: 7081.6

there's a,

Time: 7084.63

like guys like lifting heavy things

Time: 7088.07

and girls like creating beautiful art,

Time: 7091.54

and, like there's-

Time: 7094.053

- A more traditional view of gender,

Time: 7095.733

more 1950, '60s.

Time: 7098.25

- But we didn't think in terms of, at least at that age,

Time: 7100.44

in terms of like roles and then like a homemaker

Time: 7103.83

or something like that or not,

Time: 7105.48

it was more about what people care about.

Time: 7108.27

Like girls cared about this set of things,

Time: 7111.42

and guys cared about the set of things.

Time: 7113.28

I think mathematics and engineering

Time: 7115.99

was something that guys cared about and sort of,

Time: 7118.63

at least my perception of that time.

Time: 7120.53

And then girls cared about beauty.

Time: 7124.02

So like guys want to create machines,

Time: 7126.1

girls want to create beautiful stuff.

Time: 7128.035

[Lex laughing]

Time: 7128.882

And now, of course,

Time: 7129.89

that I don't take that forward

Time: 7132.79

in some kind of philosophy of life,

Time: 7134.413

but it's just the way I grew up and the way I remember it.

Time: 7137.56

But all, everyone worked hard.

Time: 7141.63

The value of hard work was instilled in everybody.

Time: 7146.53

And through that,

Time: 7150.11

I think it's a little bit of hardship.

Time: 7152.078

Of course also economically everybody was poor,

Time: 7154.98

especially with the collapse of the Soviet Union.

Time: 7157.14

There's poverty everywhere.

Time: 7158.65

You didn't notice it as much,

Time: 7159.84

but there was a,

Time: 7161.07

because there's not much material possessions,

Time: 7164.14

there was a huge value placed on human connection.

Time: 7167.673

Just meeting with neighbors,

Time: 7170.15

everybody knew each other.

Time: 7171.23

We lived in an apartment building very different

Time: 7174.34

than you have in the United States these days.

Time: 7176.5

Everybody knew each other.

Time: 7179.03

You would get together,

Time: 7179.96

drink vodka, smoke cigarettes,

Time: 7181.69

and play guitar and sing sad songs about life.

Time: 7187.61

- What's with the sad songs in the Russian thing?

Time: 7190.63

I mean, Russians do I express joy from time to time.

Time: 7194.56

- Yeah, they do.

Time: 7195.97

- Certainly you do.

Time: 7197.59

But what do you think that's about?

Time: 7200.28

Is it 'cause it's cold there?

Time: 7201.36

But it's called other places too, right?

Time: 7204.49

- I think,

Time: 7205.44

let's just, first of all the Soviet Union,

Time: 7210.38

the echoes of World War II and the millions and millions

Time: 7213.522

and millions of people that,

Time: 7214.72

civilians that were slaughtered,

Time: 7217.04

and also starvation is there, right?

Time: 7220.21

So like the echoes of that,

Time: 7223.213

of the ideas, the literature, the art is there.

Time: 7225.95

Like that's a grandparents,

Time: 7227.34

that's parents, that's all there.

Time: 7229.37

So that contributes to it,

Time: 7230.6

that life can be absurdly unexplainably cruel.

Time: 7235.62

At any moment everything can change.

Time: 7237.41

So that's in there.

Time: 7238.64

Then I think there's an empowering aspect

Time: 7240.96

to finding beauty in suffering

Time: 7243.27

that then everything else is beautiful too.

Time: 7245.103

It's like, if you just linger or it's like,

Time: 7247.68

why you meditate on death?

Time: 7249.28

Is like, if you just think about the worst possible case

Time: 7252.08

and find beauty in that,

Time: 7253.13

then everything else is beautiful too.

Time: 7254.417

And so you write songs about the dark stuff.

Time: 7257.44

And that's somehow helps you deal with whatever comes.

Time: 7262.2

There's a hopelessness to the Soviet Union that,

Time: 7267.49

like, inflation,

Time: 7269.39

all those kinds of things where people were sold dreams

Time: 7273.24

and never delivered.

Time: 7275.44

And so like, there's a,

Time: 7278.22

if you don't sing songs about sad things,

Time: 7281.42

you're going to become cynical about this world.

Time: 7284

- Mm-hmm.

Time: 7284.833

Interesting.

Time: 7285.692

- So they don't want to give in to cynicism.

Time: 7287.39

Now, a lot of people did, one of the,

Time: 7291.54

but that is the battle against cynicism.

Time: 7294.28

One of the things that may be common in Russia

Time: 7298.47

is a kind of cynicism about,

Time: 7301.07

like, if I told you the thing I said earlier

Time: 7303.23

about dreaming about robots,

Time: 7304.853

it's very common for people to dismiss that dream,

Time: 7308.84

of saying, no, that's not, that's too wild.

Time: 7312.21

Like, who else do you know that did that?

Time: 7314.35

Or you want to start a podcast?

Time: 7315.92

Like who else?

Time: 7317.03

Like nobody's making money on podcasts.

Time: 7318.82

Like, why do you want to start a podcast?

Time: 7320.05

That kind of mindset I think is quite common,

Time: 7323.41

which is why I would say entrepreneurship in Russia

Time: 7327.55

is still not very good.

Time: 7329.36

Which to be a business,

Time: 7331.06

like, to be an entrepreneur you have to dream big,

Time: 7333.47

and you have to have others around you,

Time: 7335.02

like friends and support group that make you dream big.

Time: 7339.12

But if you don't give in to cynicism

Time: 7342.39

and appreciate the beauty in the unfairness of life,

Time: 7347.83

the absurd unfairness of life,

Time: 7349.96

then I think it just makes you appreciative of everything.

Time: 7354.52

It's like a,

Time: 7355.42

it's a prerequisite for gratitude.

Time: 7358.05

And so, yeah,

Time: 7360.34

I think that instilled in me ability

Time: 7362.58

to appreciate everything,

Time: 7364.02

just like everything.

Time: 7365.24

Everything's amazing.

Time: 7366.367

And then also there is a culture

Time: 7371.267

of like romanticizing everything.

Time: 7376.14

Like, it's almost like romantic relationships

Time: 7381.37

were very like soap opera,

Time: 7383.43

like is very like over the top dramatic.

Time: 7387.72

And I think that it was instilled in me too,

Time: 7390.93

not only do I appreciate everything about life,

Time: 7393.68

but I get like emotional about it.

Time: 7395.616

In a sense, like,

Time: 7397.08

I get like a visceral feeling of joy for everything.

Time: 7401.7

And the same with friends or people of the opposite sex.

Time: 7406.13

Like, there's a deep,

Time: 7408.42

like emotional connection there that like [laughing]

Time: 7413

that's like way too dramatic too.

Time: 7414.96

Like, I guess relative to what the actual moment is.

Time: 7418.8

But I derive so much deep,

Time: 7422.38

like dramatic joy from so many things in life.

Time: 7426.66

And I think I would attributed that

Time: 7428.94

to the upbringing in Russia.

Time: 7430.37

But the thing that sticks most of all is the friendship.

Time: 7434.04

And have now since then had one other friend like that

Time: 7439.244

in the United States, he lives in Chicago.

Time: 7442.66

His name is Matt.

Time: 7443.832

And slowly here and there accumulating

Time: 7448.32

really fascinating people,

Time: 7449.89

but I'm very selective with that.

Time: 7451.57

Funny enough, the few times,

Time: 7456.18

it's not few it's a lot of times now

Time: 7458.01

interacting with Joe Rogan [chuckles]

Time: 7459.94

it's sounds surreal to say,

Time: 7461.414

but there was a kindred spirit there.

Time: 7464.21

Like I've connected with him.

Time: 7466.811

And there's been people like that

Time: 7468.02

also in the grappling sports that are really connected with.

Time: 7471.25

I've actually struggled,

Time: 7473.42

which is why I'm so glad to be your friend,

Time: 7476.93

is I've struggled to connect with scientists.

Time: 7480.29

- They can be a little bit wooden sometimes.

Time: 7482.69

- [Lex] Yeah.

Time: 7483.523

- Even the biologist.

Time: 7484.356

I mean, one thing that I,

Time: 7486.559

well, I'm so struck by the fact that you work with robots,

Time: 7490.13

you're an engineer, AI,

Time: 7491.408

science technology,

Time: 7492.98

and that all sounds like hardware, right?

Time: 7495.35

But what you're describing and I know is true about you

Time: 7498.31

is this deep emotional life and this resonance

Time: 7501.69

and it's really wonderful.

Time: 7502.74

I actually think it's one of the reasons why so many people,

Time: 7506.53

scientists and otherwise have gravitated towards you

Time: 7509.04

and your podcast is because you hold both elements.

Time: 7512.1

In the Hermann Hesse's book,

Time: 7514.27

I don't know if you, "Narcissus and Goldmund," right?

Time: 7516.14

It's about these elements of the logical rational mind

Time: 7519.43

and the emotional mind and how those are woven together.

Time: 7522.98

And if people haven't read it, they should,

Time: 7524.9

and you embody the full picture.

Time: 7527.36

And I think that's so much of what draws people to you.

Time: 7530.06

- I've read every Hermann Hesse book by the way.

Time: 7531.96

- As usual [chuckles]

Time: 7533.49

as usual I've done about 9% of what life is.

Time: 7537.01

No, it's true.

Time: 7537.843

You mentioned Joe,

Time: 7539.74

who is a phenomenal human being,

Time: 7542.38

not just for his amazing accomplishments,

Time: 7544.3

but for how he shows up to the world one on one.

Time: 7548.53

I think I heard him say the other day on an interview,

Time: 7551.64

he said, there is no public or private version of him.

Time: 7555.73

And he's like, this is me.

Time: 7557.03

He said that it was beautiful.

Time: 7558.35

He said, I'm like the fish that got through the net.

Time: 7560.95

And there is no onstage offstage version.

Time: 7563.6

And you're absolutely right.

Time: 7564.61

And I.

Time: 7565.88

- Fish.

Time: 7566.713

[Lex laughing]

Time: 7567.68

- So, but, well, you guys,

Time: 7568.513

I have a question actually about-

Time: 7569.86

- But that's a really good point

Time: 7570.89

about public and private life.

Time: 7572.67

He was a huge,

Time: 7573.503

if I could just comment real quick.

Time: 7575.68

Like that, he was a,

Time: 7577

I've been a fan of Joe for a long time,

Time: 7578.61

but he's been an inspiration

Time: 7579.887

to not have any difference between public and private life.

Time: 7584.74

I actually had a conversation with Naval about this,

Time: 7588.2

and he said that you can't have a rich life,

Time: 7594.29

like exciting life

Time: 7596.66

if you're the same person publicly and privately.

Time: 7599.38

And I think I understand that idea,

Time: 7602.46

but I don't agree with it.

Time: 7605.41

I think that's really fulfilling and exciting

Time: 7609.15

to be the same person privately and publicly

Time: 7611.39

with very few exceptions.

Time: 7612.54

Now that said, I don't have any really strange sex kinks.

Time: 7618.09

So like,

Time: 7618.923

I feel like it can be open with basically everything.

Time: 7620.46

I don't have anything I'm ashamed of.

Time: 7623.58

There's some things that could be perceived poorly,

Time: 7625.7

like the screaming Roombas,

Time: 7627.77

but I'm not ashamed of them.

Time: 7629.1

I just have to present them in the right context.

Time: 7631.44

But there's a freedom to being the same person in private

Time: 7635.62

as in public.

Time: 7636.67

And that Joe made me realize that you can be that.

Time: 7642.5

And also to be kind to others.

Time: 7645.28

It sounds kind of absurd,

Time: 7648.67

but I really always enjoyed like being good to others.

Time: 7658.49

Like just being kind towards others.

Time: 7661.13

But I always felt like the world didn't want me to be.

Time: 7665.42

Like, there's so much negativity when I was growing up,

Time: 7668.55

like just around people.

Time: 7669.74

If you actually just notice how people talk,

Time: 7673.12

they, from like, complaining about the weather,

Time: 7677.161

this could be just like the big cities

Time: 7679.01

that I've visited.

Time: 7679.843

But there's a general negativity,

Time: 7681.735

and positivity is kind of suppressed.

Time: 7685.07

You're not,

Time: 7685.903

one, you're not seen as very intelligent,

Time: 7688.75

and two, there's a kind of,

Time: 7690.45

you're seen as like a little bit of a weirdo.

Time: 7692.86

And so I always felt like I had to hide that.

Time: 7695.79

And what Joe made me realize,

Time: 7697.16

one, I could be fully

Time: 7699.13

just the same person private and public,

Time: 7702.33

and two, I can embrace being kind,

Time: 7705.18

in just, in the way that I like,

Time: 7708.31

in the way I know how to do.

Time: 7711.27

And sort of for me on like, on Twitter or like publicly,

Time: 7716.43

whenever I say stuff,

Time: 7717.56

that means saying stuff simply

Time: 7719.67

almost to the point of cliche.

Time: 7721.24

And like, I have the strength now to say it,

Time: 7724.59

even if I'm being mocked.

Time: 7726.157

[Lex laughing]

Time: 7726.99

You know what I mean?

Time: 7727.823

Like, just it's okay.

Time: 7728.72

If everything's going to be okay.

Time: 7730.43

Okay, some people will think you're dumb.

Time: 7732.071

They're probably right.

Time: 7733.56

The point is,

Time: 7734.393

like, just enjoy being yourself.

Time: 7736.32

And that Joe, more than almost anybody else,

Time: 7738.35

because he's so successful at it, inspired me to do that.

Time: 7743.1

Be kind and be the same person, private and public.

Time: 7746.23

- I love it.

Time: 7747.063

And I love the idea that authenticity

Time: 7748.69

doesn't have to be oversharing, right?

Time: 7751.085

That it doesn't mean you reveal every detail of your life.

Time: 7754.99

It's a way of being true to an essence of oneself.

Time: 7759.06

- Right.

Time: 7759.98

There's never a feeling when you deeply think and introspect

Time: 7764.46

that you're hiding something from the world

Time: 7766.15

or you're being dishonest in some fundamental way.

Time: 7768.77

So, yeah, that's truly liberating.

Time: 7773.52

It allows you to think,

Time: 7774.76

it allows you to like think freely,

Time: 7777.08

to speak freely,

Time: 7778.73

to just to be freely.

Time: 7782.14

That said,

Time: 7783.17

it's not like there's not still a responsibility

Time: 7789.55

to be the best version of yourself.

Time: 7792.15

So, I'm very careful with the way I say something.

Time: 7797.22

So, the whole point,

Time: 7798.65

it's not so simple to express the spirit

Time: 7802.61

that's inside you with words.

Time: 7804.997

I mean, some people are much better than others.

Time: 7808.8

I struggle.

Time: 7809.633

Like oftentimes when I say something

Time: 7812.331

and I hear myself say it,

Time: 7814.25

it sounds really dumb and not at all what I meant.

Time: 7816.75

So that's the responsibility you have.

Time: 7818.45

It's not just like being

Time: 7820.33

the same person publicly and privately

Time: 7821.87

means you can just say whatever the hell.

Time: 7824.5

It means there's still responsibility to try to be,

Time: 7827.67

to express who you truly are.

Time: 7829.33

And that's hard.

Time: 7832.24

[Lex chuckles]

Time: 7833.073

- It hard.

Time: 7833.906

And I think that,

Time: 7834.739

we have this pressure,

Time: 7837.67

all people, when I say we, I mean all humans,

Time: 7840.568

and maybe robots too,

Time: 7842.75

feel this pressure to be able to express ourselves

Time: 7846.68

in that one moment in that one form.

Time: 7848.93

And it is beautiful when somebody, for instance,

Time: 7851.3

can capture some essence of love or sadness or anger

Time: 7855.19

or something in a song or in a poem or in a short quote,

Time: 7858.79

but perhaps it's also possible to do it in aggregate,

Time: 7863.21

all the things how you show up.

Time: 7865.658

For instance, one of the things that initially drew me

Time: 7868.72

to want to get to know you as a human being and a scientist

Time: 7871.46

and eventually we became friends,

Time: 7873.73

was the level of respect

Time: 7875.87

that you brought to your podcast listeners

Time: 7878.02

by wearing a suit.

Time: 7879.38

- Yeah. - I'm being serious here.

Time: 7880.942

I was raised thinking that if you overdress a little bit,

Time: 7884.39

overdressed by American, certainly by American standards,

Time: 7887.3

you're overdressed for a podcast,

Time: 7888.68

but it's genuine.

Time: 7890.58

You're not doing it for any reason,

Time: 7891.86

except I have to assume, and I assumed at the time,

Time: 7894.798

that it was because you have a respect for your audience,

Time: 7897.99

you respect them enough to show up a certain way for them.

Time: 7902.44

It's for you also, but it's for them.

Time: 7904.19

- Yeah.

Time: 7905.023

- And I think between that

Time: 7905.96

and your commitment to your friendship,

Time: 7907.893

just the way that you talk about friendships and love

Time: 7910.07

and the way you hold up these higher ideals,

Time: 7912.88

I think at least as a consumer of your content

Time: 7917.94

and as your friend,

Time: 7919.617

what I find, is that in aggregate,

Time: 7921.8

you're communicating who you are,

Time: 7923.4

it doesn't have to be one quote or something.

Time: 7925.85

And I think that we're sort of obsessed

Time: 7928.18

by like the one Einstein quote,

Time: 7930.16

or the one line of poetry or something,

Time: 7933.05

but I think you so embody the way that, and Joe as well,

Time: 7938.32

it's about how you live your life

Time: 7940.04

and how you show up as a collection of things

Time: 7942.88

and said and done.

Time: 7944.91

- Yeah, that that's,

Time: 7945.743

and so, the aggregate is the goal, the tricky thing,

Time: 7950.42

and Jordan Peterson talks about this

Time: 7952.25

because he's under attack

Time: 7953.28

way more than you and I will ever be, but-

Time: 7956.26

- Right now?

Time: 7957.26

- For now, right?

Time: 7958.093

This is very true for now.

Time: 7960.47

That the people who attack on the internet,

Time: 7966.99

this is one of the problems with Twitter,

Time: 7968.777

is they don't consider the aggregate,

Time: 7973.21

they take a single statements.

Time: 7975.34

And so, one of the defense mechanisms,

Time: 7978.46

like again why Joe has been an inspiration,

Time: 7981.19

is that when you in aggregate, a good person,

Time: 7985.57

a lot of people will know that.

Time: 7986.897

And so, that makes you much more immune

Time: 7988.98

to the attacks of people

Time: 7990.05

that bring out an individual statement

Time: 7991.627

that might be a misstatement of some kind,

Time: 7994.27

or doesn't express who you are.

Time: 7996.99

And so, that, I like that idea is the aggregate.

Time: 8000.99

And the power of the podcast,

Time: 8003.82

is you have hundreds of hours out there,

Time: 8007.24

and being yourself and people get to know who you are.

Time: 8010.202

And once they do,

Time: 8011.561

and you post pictures of screaming Roombas

Time: 8014.78

as you kick them,

Time: 8015.764

they will understand that you don't mean well.

Time: 8018.16

By the way, as a side comment,

Time: 8021.32

I don't know if I want to release this

Time: 8022.97

because it's not just the Roombas-

Time: 8026.79

- You have a whole dungeon of robots.

Time: 8028.52

- Okay.

Time: 8029.353

So, this is a problem,

Time: 8031.65

the Boston Dynamics came up against this problem.

Time: 8034.47

But, let me work this out like workshop this out with you,

Time: 8039.87

and maybe because we'll post this, people will let me know.

Time: 8045.59

So, there's legged robots, they look like a dog,

Time: 8049.085

I'm trying to create a very real human-robot connection,

Time: 8054.03

but like they're also incredible

Time: 8055.45

because you can throw them like off of a building

Time: 8059.51

and they'll land fine.

Time: 8061.45

And this beautiful.

Time: 8062.43

- That's amazing.

Time: 8063.263

I've seen the Instagram videos

Time: 8064.36

of like cats jumping off of like fifth story buildings

Time: 8067.32

and then walking away.

Time: 8068.69

But no one should throw their cat

Time: 8070.41

out of a window of a building.

Time: 8071.452

- Well, this is the problem I'm experiencing,

Time: 8072.57

all certainly kicking the robots,

Time: 8074.72

its really fascinating how they recover from those kicks.

Time: 8078.04

But like just seeing myself do it

Time: 8080.47

and also seeing others do it, it just does not look good,

Time: 8083.4

and I don't know what to do with that.

Time: 8084.94

'Cause it's such a- - Ill 'do it.

Time: 8088.093

[Lex laughing]

Time: 8089.43

- See.

Time: 8090.263

But you don't,

Time: 8091.91

'cause you are- - A robot, no, I'm kidding.

Time: 8095.4

What's interesting?

Time: 8096.233

- Yeah.

Time: 8097.066

- Before today's conversation, I probably could do it,

Time: 8100.02

and I'm thinking about robots, bills of rights and things.

Time: 8104.611

Not to satisfy you or to satisfy anything,

Time: 8108.6

except that if they have some sentience aspect

Time: 8113.04

to their being, then I would load to kick it.

Time: 8116.44

- I don't think we'd be able to kick it,

Time: 8117.7

you might be able to kick the first time,

Time: 8119.03

but not the second, this is the problem of experience.

Time: 8121.687

One of the cool things,

Time: 8122.875

is one of the robots I'm working with,

Time: 8126.61

you can pick it up by one leg and is dangling,

Time: 8129.38

and you can throw it in any kind of way

Time: 8131.772

and it'll land correctly.

Time: 8133.71

So, it's really-

Time: 8134.704

- I had a friend who had a cat like that.

Time: 8135.72

[Lex laughing]

Time: 8137.778

- Oh man, we look forward to the letters on the cat-

Time: 8140.6

- Oh no, I'm not suggesting anyone did that,

Time: 8142.277

but he had this cat,

Time: 8143.97

and the cat, he would just throw it onto the bed

Time: 8146.08

from across the room, and then it would run back for more,

Time: 8148.84

or somehow that was the nature of the relationship.

Time: 8151.62

I think no one should do that to an animal,

Time: 8154.13

but this cat seemed to return for whatever reason.

Time: 8157.58

- But a robot is a robot,

Time: 8158.777

and it's fascinating to me how hard it is for me to do that.

Time: 8162.27

So, it's unfortunate,

Time: 8164.05

but I don't think I can do that to a robot.

Time: 8166.28

Like I struggle with that.

Time: 8169.661

So, for me to be able to do that with a robot,

Time: 8172.879

I have to almost get like into the state

Time: 8175.86

that I imagine like doctors get into

Time: 8177.75

when they're doing surgery,

Time: 8179.38

like I have to do what robotics colleagues of mine do,

Time: 8183.02

which is like start seeing it as an object.

Time: 8185.26

- Dissociate.

Time: 8186.093

- Like dissociate.

Time: 8186.926

So, it was just fascinating

Time: 8187.95

that I have to do that in to do that with a robot.

Time: 8190.73

I just want to take that little bit of a tangent.

Time: 8193.65

- No, I think it's an important thing.

Time: 8195.03

I mean, I'm not shy about the fact that for many years

Time: 8200.71

I've worked on experimental animals,

Time: 8202.46

and that's been a very challenging aspect

Time: 8204.56

of being a biologist.

Time: 8206.1

Mostly mice, but in the past no longer,

Time: 8208.64

thank goodness 'cause I just don't like doing it,

Time: 8211.25

larger animals as well.

Time: 8212.71

And now I work on humans,

Time: 8213.76

which I can give consent, verbal consent.

Time: 8215.95

So, I think that it's extremely important

Time: 8220.84

to have an understanding of what the guidelines are

Time: 8223.92

and where one's own boundaries are around this.

Time: 8226.22

It's not just an important question,

Time: 8229.06

it might be the most important question

Time: 8231.08

before any work can progress.

Time: 8232.91

- So, you asked me about friendship.

Time: 8234.96

I know you have a lot of thoughts about friendship,

Time: 8238.21

what do you think is the value of friendship in life?

Time: 8242.35

- Well, for me personally,

Time: 8244.85

just because of my life trajectory and arc friendship,

Time: 8249.91

and I should say,

Time: 8251.54

I do have some female friends that are just friends,

Time: 8255.65

they're completely platonic relationships,

Time: 8257.15

but it's been mostly male friendship to me, has been-

Time: 8259.7

- It has been all male friendships to me actually, yeah.

Time: 8261.96

- Interesting. - Yeah.

Time: 8263.06

- It's been an absolute lifeline.

Time: 8265.73

They are my family,

Time: 8267.07

I have a biological family

Time: 8268.31

and I have great respect and love for them

Time: 8270.09

and an appreciation for them,

Time: 8271.29

but it's provided me the,

Time: 8277.912

I won't even say confidence

Time: 8278.92

because there's always an anxiety in taking any good risk,

Time: 8281.997

or any risk worth taking.

Time: 8284.25

It's given me the sense that I should go for certain things

Time: 8288.9

and try certain things to take risk to weather that anxiety.

Time: 8292.21

And I don't consider myself

Time: 8294.67

a particularly competitive person,

Time: 8296.81

but I would sooner die than disappoint,

Time: 8301.91

or let down one of my friends.

Time: 8304.66

I can think of nothing worse actually,

Time: 8306.96

than disappointing one of my friends,

Time: 8309.17

everything else is secondary to me.

Time: 8311.29

- What disappointment?

Time: 8313.16

- Disappoint, meaning not,

Time: 8315.93

I mean, certainly I strive always to show up as best I can

Time: 8320.76

for the friendship, and that can be in small ways.

Time: 8323

That can mean making sure the phone is away,

Time: 8325.03

sometimes it's about,

Time: 8328.58

I'm terrible with punctuality 'cause I'm an academic.

Time: 8330.86

And so, I just get lost in time and I don't mean anything,

Time: 8333.27

but it's striving to, to listen to,

Time: 8336.32

to enjoy good times and to make time.

Time: 8338.54

It kind of goes back to this first variable we talked about,

Time: 8341.89

to make sure that I spend time

Time: 8343.646

and to get time in person and check in.

Time: 8346.77

And I think there's so many ways

Time: 8350.69

in which friendship is vital to me,

Time: 8352.38

it's it's actually to me, what makes life worth living.

Time: 8354.86

- Yeah.

Time: 8355.693

Well, there's a,

Time: 8357.43

I am surprised like with the high school friends

Time: 8360.11

how we don't actually talk that often these days

Time: 8362.361

in terms of time,

Time: 8363.78

but every time we see each other,

Time: 8364.98

it's immediately right back to where we started.

Time: 8367.12

So, I struggled that how much time you really allocate,

Time: 8372.62

for the friendship to be deeply meaningful

Time: 8374.43

because they're always there with me

Time: 8376.293

even if we don't talk often.

Time: 8379.56

So, there's a kind of loyalty.

Time: 8381.02

I think maybe it's a different style,

Time: 8384.04

but I think to me,

Time: 8387.04

friendship is being there in the hard times, I think.

Time: 8391.64

Like I'm much more reliable when you're going through shit

Time: 8396.52

than in like-

Time: 8397.61

- You're pretty reliable anyway.

Time: 8399.6

- No, but if you're like a wedding or something like that,

Time: 8403.17

or like, I don't know, like you want an award of some kind,

Time: 8408.77

yeah, I'll congratulate the shit out of you,

Time: 8411.92

but like that's not, and I'll be there,

Time: 8414.5

but that's not as important to me as being there

Time: 8416.74

when like nobody else is like just being there

Time: 8420.28

when shit hits the fan, or something's tough

Time: 8423.85

where the world turns their back on you,

Time: 8425.91

all those kinds of things,

Time: 8426.91

that, to me, that's where friendship is meaningful.

Time: 8429.37

- Well, I know that to be true about you

Time: 8430.96

and that's a felt thing and a real thing with you.

Time: 8433.049

Let me ask one more thing about that actually,

Time: 8435.61

because I'm not a practitioner Jujitsu,

Time: 8438.46

I know you are, Joe is,

Time: 8439.7

but years ago, I read a book that I really enjoyed,

Time: 8442.45

which is Sam Sheridan's book, " A Fighter's Heart,"

Time: 8444.83

he talks about all these different forms of martial arts.

Time: 8447.185

And maybe it was in the book, maybe it was in an interview,

Time: 8450.68

but he said that fighting,

Time: 8453.33

or being in physical battle with somebody,

Time: 8456.24

Jujitsu boxing or some other form

Time: 8458.45

of direct physical contact between two individuals

Time: 8461.87

creates this bond unlike any other.

Time: 8464.34

Because he said it's like a one night stand,

Time: 8466.61

you're sharing bodily fluids

Time: 8468.06

with somebody that you barely know.

Time: 8469.83

- Yeah.

Time: 8470.663

- And I chuckled about it 'cause it's kind of funny

Time: 8473.14

and it kind of tongue in cheek.

Time: 8474.85

But at the same time,

Time: 8476.3

I think this is a fundamental way

Time: 8479

in which members of a species bond

Time: 8482.31

is through physical contact.

Time: 8484.35

And certainly, there are other forms,

Time: 8485.92

there's cuddling, and there's hand holding,

Time: 8487.4

and there's in their sexual intercourse,

Time: 8489.71

and there's all sorts of things.

Time: 8490.543

- What's cuddling?

Time: 8491.579

I haven't heard of it.

Time: 8492.77

- I heard this recently, I didn't know this term,

Time: 8494.91

but there's a term,

Time: 8496.18

they've turned the noun cupcake into a verb,

Time: 8499.2

cupcaking it turns out, I just learned about this.

Time: 8501.89

Cupcaking is when you spend time just cuddling.

Time: 8505.1

I didn't know about this.

Time: 8506.34

You heard it here first,

Time: 8507.33

although I heard it first just the other day.

Time: 8508.89

Cupcaking is actually a-

Time: 8510.03

- Cuddling is everything,

Time: 8510.92

it's not just like,

Time: 8511.81

is it in bed, or is it in the coach?

Time: 8513.79

Like what's cuddling?

Time: 8515.44

I do look up what cuddling is-

Time: 8516.39

- We need to look at this up

Time: 8517.223

and we need to define the variables.

Time: 8519.03

I think it definitely has to do

Time: 8520.67

with physical contact, I'm told,

Time: 8524.23

but in terms of battle, a competition,

Time: 8530.74

and the Sheridan quote, I'm just curious.

Time: 8532.66

So, do you get close, or feel a bond with people that,

Time: 8538.39

for instance, you rolled Jujitsu with,

Time: 8540.13

or even though you don't know anything else about them,

Time: 8544.01

was he right about this?

Time: 8545.54

- Yeah, I mean on many levels.

Time: 8547.17

He also has the book, what?

Time: 8548.357

"A Fighter's Mind."

Time: 8549.44

- Yeah, that was the third one.

Time: 8550.92

He's actually an excellent writer.

Time: 8552.17

What's interesting about him, just briefly about Sheridan,

Time: 8554.54

I don't know him, but I did a little bit of research,

Time: 8556.14

he went to Harvard, he was an art major at Harvard,

Time: 8560.12

he claims all he did was smoke cigarettes and do art.

Time: 8563.47

I don't know if his art was any good.

Time: 8565.14

And I think his father was in the SEAL teams.

Time: 8568.348

And then when he got out of Harvard, graduated,

Time: 8571.76

he took off around the world

Time: 8572.93

learning all the forms of martial arts,

Time: 8574.47

and was early to the kind of ultimate fighting

Time: 8576.7

kind of mixed martial arts and things.

Time: 8579.05

Great book.

Time: 8580

Yeah, yeah.

Time: 8580.833

- It's amazing.

Time: 8581.666

I don't actually remember it, but I read it,

Time: 8583.178

and I remember thinking that was an amazing encapsulation

Time: 8586.667

of what makes fighting like the art,

Time: 8590.55

like what makes it compelling.

Time: 8591.98

I would say that there's so many ways

Time: 8595.82

that Jujitsu grappling, wrestling, combat sports in general,

Time: 8600.825

is like one of the most intimate things you could do.

Time: 8604.28

I don't know if I would describe

Time: 8605.6

in terms of bodily liquids and all those kinds of things.

Time: 8607.643

- I think he was more or less joking.

Time: 8611.266

- I think there's a few ways that it does that.

Time: 8615.1

So, one, because you're so vulnerable [sighs]

Time: 8621.04

So, the honesty of stepping on the mat

Time: 8625.57

and often all of us have ego

Time: 8628.39

thinking we're better than we are at this particular art.

Time: 8632.38

And then the honesty of being submitted,

Time: 8635.576

or being worse than you thought you are

Time: 8638.75

and just sitting with that knowledge,

Time: 8640.623

that kind of honesty,

Time: 8642.04

we don't get to experience it in most of daily life.

Time: 8645.743

We can continue living

Time: 8646.956

somewhat of an illusion of our conceptions of ourselves

Time: 8650.349

'cause people are not going to hit us with the reality,

Time: 8653.79

the mat speaks only the truth, the reality just hits you.

Time: 8657.86

And that vulnerability

Time: 8658.989

is the same as like the loss of a loved one,

Time: 8662.16

though it's the loss of a reality that you knew before,

Time: 8666.33

you now have to deal with this new reality.

Time: 8668.21

And when you're sitting there in that vulnerability,

Time: 8670.211

and there's these other people

Time: 8672.15

that are also sitting in that vulnerability,

Time: 8674.35

you get to really connect like, fuck,

Time: 8676.96

like I'm not as special as I thought I was,

Time: 8680.127

and life is like not, life is harsher than I thought I was,

Time: 8686.202

and we're just sitting there with that reality,

Time: 8687.71

some of us can put words to them, some we can't.

Time: 8690.04

So, I think that definitely,

Time: 8691.3

is a thing that at least the intimacy.

Time: 8694.65

The other thing is the human contact.

Time: 8698.64

There's something about, I mean, like a big hug,

Time: 8703.71

like during COVID,

Time: 8705.32

very few people hugged me and I hugged them,

Time: 8707.81

and I always felt good when they did.

Time: 8710.28

Like we were all tested,

Time: 8711.61

and especially now we're vaccinated,

Time: 8713.89

but there's still people,

Time: 8715.12

this is true of San Francisco's, it's true in Boston,

Time: 8717.36

they want to keep, not only six feet away,

Time: 8719.41

but stay at home and never touch you.

Time: 8721.76

That loss of basic humanity

Time: 8724.68

is the opposite of what I feel in Jujitsu,

Time: 8729.37

where it was like that contact where you're like,

Time: 8733.621

I don't give a shit

Time: 8734.94

about whatever rules we're supposed to have in society

Time: 8737.45

where you have to keep a distance

Time: 8739.42

and all that kind of stuff.

Time: 8740.48

Just the hug,

Time: 8741.86

like the intimacy of a hug, that's like a good bear hug,

Time: 8746.47

and you're like just controlling another person,

Time: 8749.75

and also there is some kind of love communicated

Time: 8752.14

through just trying to break each other's arms.

Time: 8754.205

I don't exactly understand

Time: 8756.024

why violence is the such a close neighbor to love,

Time: 8761.24

but it is.

Time: 8762.227

- Well, in the hypothalamus,

Time: 8764.8

the neurons that control sexual behavior,

Time: 8768.27

but also non-sexual contact,

Time: 8771.018

are not just nearby the neurons

Time: 8773.56

that control aggression and fighting,

Time: 8775.81

they are salt and pepper with those neurons.

Time: 8779.172

It's a very interesting,

Time: 8781.58

and it almost sounds

Time: 8783.16

kind of risky and controversial and stuff,

Time: 8785.07

I'm not anthropomorphizing about what this means,

Time: 8787.87

but in the brain, those structures are interdigitated,

Time: 8792.486

you can't separate them except at a very fine level.

Time: 8796.04

And here, the way you describe it,

Time: 8797.75

is the same as a real thing.

Time: 8799.74

- I do want to make an interesting comment.

Time: 8802.99

Again, these are the things

Time: 8803.823

that could be taken out of context,

Time: 8805.69

but one of the amazing things about Jujitsu,

Time: 8810.57

is both guys and girls train it.

Time: 8813.14

And I was surprised.

Time: 8814.409

So, like I'm a big fan of yoga pants [giggles]

Time: 8819.313

at the gym kind of thing.

Time: 8821.04

It reveals the beauty of the female form.

Time: 8824.81

But the thing is,

Time: 8826.24

like girls are dressed in skintight clothes

Time: 8828.8

in Jujitsu often.

Time: 8830.21

And I found myself not at all,

Time: 8832.62

thinking like that at all when training with girls.

Time: 8835.03

- Well, the context is very non-sexual.

Time: 8837.19

- But I was surprised to learn that.

Time: 8839.636

When I first started to Jujitsu,

Time: 8841.19

I thought, wouldn't that be kind of weird

Time: 8842.69

to train with the opposites that in something so intimate.

Time: 8845.279

- So, boys and girls, men and women,

Time: 8847.88

they roll Jujitsu together?

Time: 8850.56

- Completely. - Interesting.

Time: 8851.44

- And the only times girls kind of try to stay away from guys,

Time: 8854.894

I mean, there's two contexts,

Time: 8856.44

of course, there's always going to be creeps in this world.

Time: 8858.89

So, everyone knows who kind of stay away from,

Time: 8862.27

and the other is there's a size disparity.

Time: 8864.31

So, girls will often try to roll with people

Time: 8866.29

a little bit closer weight-wise,

Time: 8868.31

But no, that's one of the things

Time: 8871.13

that are empowering to women,

Time: 8872.689

that's what they fall in love with

Time: 8874.3

when they started doing Jujitsu,

Time: 8875.53

is first of all,

Time: 8877.31

they gain an awareness and a pride over their body,

Time: 8880.27

which is great.

Time: 8881.103

And then second, they get to [chuckles]

Time: 8882.906

especially later on, start submitting big dudes

Time: 8885.98

like these bros that come in

Time: 8889.28

who are all shredded and like muscular,

Time: 8891.33

and they get to technique to exercise dominance over them,

Time: 8896.1

and that's a powerful feeling.

Time: 8897.364

- You've seen women force a larger guy to tap her,

Time: 8901.57

or even choke them up.

Time: 8902.46

- Well, I was deadlifting like a four,

Time: 8909.53

oh boy, I think it's 495.

Time: 8911.58

So, I was really into power-lifting

Time: 8913.16

when I started at Jujitsu,

Time: 8915.03

and I remember being submitted by,

Time: 8917.44

I thought I walked in feeling like I'm going to be,

Time: 8920.67

if not the greatest fighter ever, at least top three.

Time: 8923.03

And so, as a white belt, you roll in like all happy.

Time: 8927.45

And then you realize

Time: 8928.639

that as long as you're not applying too much force,

Time: 8931.77

that you're having,

Time: 8932.603

I remember being submitted many times

Time: 8934.39

by like 130, 120-pound girls

Time: 8936.829

at our Balance Studios in Philadelphia,

Time: 8939.9

that a lot of incredible female Jujitsu players.

Time: 8942.096

And that's really humbling too

Time: 8943.854

that technique can overpower in combat pure strength.

Time: 8951.87

And that's the other thing,

Time: 8953.65

there is something about combat that's primal.

Time: 8958.16

Like there, it just feels,

Time: 8962.07

it feels like we were born to do this.

Time: 8966.5

Like that-

Time: 8967.333

- We have circuits in our brain

Time: 8968.624

that are dedicated to this kind of interaction.

Time: 8972.42

There's no question.

Time: 8974.11

- And that's what it felt like,

Time: 8975.45

it wasn't that I'm learning a new skill.

Time: 8978.96

It was like,

Time: 8979.793

somehow I am a remembering echoes

Time: 8982.84

of something I've learned in the past.

Time: 8984.44

- Well, it's like hitting puberty.

Time: 8985.69

A child before puberty has no concept

Time: 8987.8

of boys and girls having this attraction,

Time: 8991.04

regardless of whether or not they're attracted

Time: 8992.65

to boys or girl, doesn't matter.

Time: 8993.69

At some point, most people, not all,

Time: 8995.63

but certainly, but most people, when they hit puberty,

Time: 8998.4

suddenly people appear differently,

Time: 9000.991

and certain people take on a romantic or sexual interest

Time: 9005.58

for the very first time.

Time: 9007.2

- Yeah.

Time: 9008.033

- And so it's like,

Time: 9008.866

it's revealing a circuitry in the brain.

Time: 9010.736

It's not like they learn that it's innate.

Time: 9014.36

And I think it,

Time: 9015.193

when I hear the way you describe Jujitsu

Time: 9018.45

and enrolling Jujitsu,

Time: 9019.95

it reminds me a little bit,

Time: 9021.11

Joe was telling me recently

Time: 9022.083

about the first time he went hunting

Time: 9024.47

and he felt like it revealed a circuit

Time: 9026.374

that was in him all along,

Time: 9028.56

but he hadn't experienced before.

Time: 9030.95

- Yeah.

Time: 9031.783

That's definitely there.

Time: 9032.616

And of course, there's the physical activity.

Time: 9034.96

One of the interesting things about Jujitsu

Time: 9037.443

is it's one of the really strenuous exercises

Time: 9040.643

that you can do late into your adult life,

Time: 9043.92

like into your 50, 60, 70s, 80s.

Time: 9047.011

When I came up,

Time: 9049.22

there's a few people in their 80s that were training.

Time: 9051.4

And as long as you're smart,

Time: 9053.13

as long as you practice techniques

Time: 9054.89

and pick your partners correctly,

Time: 9055.98

you can do that kind of art.

Time: 9057.34

That's late into life.

Time: 9058.173

And so you're getting exercise.

Time: 9059.57

There's not many activities I find

Time: 9061.92

that are amenable to that.

Time: 9064.52

So, because it's such a thinking game,

Time: 9068.17

the Jujitsu in particular is an art

Time: 9070.91

or technique pays off a lot.

Time: 9073.026

So you can still maintain,

Time: 9075.707

first of all, remain injury free if you use good technique,

Time: 9080.86

and also through good technique be able to go,

Time: 9085.93

be active with people that are much, much younger.

Time: 9088.066

And so that was, to me,

Time: 9090.61

that and running are the two activities

Time: 9092.51

you can kind of do late in life.

Time: 9093.87

Because to me a healthy life

Time: 9096.23

has exercises as the piece of the puzzle.

Time: 9099.4

- No, absolutely.

Time: 9100.44

And I'm glad that we're on the physical component,

Time: 9102.75

because I know that there's for you,

Time: 9107.99

you've talked before about the crossover

Time: 9109.71

between the physical and the intellectual and the mental.

Time: 9115.03

Are you still running at ridiculous hours of the night

Time: 9117.71

for ridiculously long?

Time: 9119.65

- Yeah, so, definitely.

Time: 9121.55

I've been running late at night here in Austin.

Time: 9123.54

People tell,

Time: 9124.529

the area we're in now,

Time: 9125.84

people say it's a dangerous area,

Time: 9127.27

which I find laughable coming from the bigger cities.

Time: 9130.63

No, I run late at night.

Time: 9132.83

There's something.

Time: 9135.2

- If you see a guy running through Austin at 2:00 a.m.

Time: 9139.368

in a suit and tie, it's probably.

Time: 9140.201

[Lex laughing]

Time: 9142.25

- Well, yeah.

Time: 9143.083

I mean, I do think about that

Time: 9144.11

'cause I get recognized more and more in Austin.

Time: 9146.67

I worry that,

Time: 9147.847

but not really,

Time: 9149.08

that I get recognized late at night.

Time: 9152.63

But there is something about the night

Time: 9156.65

that brings out those deep philosophical thoughts

Time: 9159

and self-reflection, that really enjoy.

Time: 9160.76

But recently I started getting back to the grind.

Time: 9164.59

So I'm going to be competing

Time: 9166.49

or hoping to be compete in September and October.

Time: 9170.06

- In Jujitsu?

Time: 9170.893

- In Jujitsu, yeah.

Time: 9171.726

To get back to competition.

Time: 9172.782

And so that requires getting back into a great cardio shape.

Time: 9178.63

I've been getting,

Time: 9179.87

running as part of my daily routine.

Time: 9182.34

- Got it.

Time: 9183.23

- [Lex] Yeah.

Time: 9184.215

- Well, I always know I can reach you

Time: 9185.18

regardless of time zone in the middle of the night,

Time: 9187.69

wherever that happens.

Time: 9189.37

- Well, part of that has to be just being single

Time: 9191.16

and being a programmer.

Time: 9193.589

Those two things just don't work well

Time: 9196.1

in terms of a steady sleep schedule.

Time: 9198.09

- It's not bankers hours kind of work.

Time: 9200.1

Nine to five.

Time: 9201.5

I want to, you mentioned single.

Time: 9203.31

I want to ask you a little bit

Time: 9204.88

about the other form of relationship,

Time: 9206.44

which is romantic love.

Time: 9209.89

So, your parents are still married?

Time: 9212.41

- Still married,

Time: 9213.243

still happily married.

Time: 9214.16

- That's impressive.

Time: 9214.993

- [Lex] Yeah.

Time: 9215.826

- A rare thing nowadays.

Time: 9216.66

- [Lex] Yeah.

Time: 9217.493

- So you grew up with that example?

Time: 9219.01

- Yeah.

Time: 9219.843

I guess that's a powerful thing, right?

Time: 9220.95

If there's an example that I think can work.

Time: 9224.841

- Yeah.

Time: 9225.674

I didn't have that in my own family,

Time: 9226.54

but when I see it,

Time: 9229.785

it's inspiring and it's beautiful.

Time: 9232.07

The fact that they have that,

Time: 9233.247

and that was the norm for you,

Time: 9235.07

I think is really wonderful.

Time: 9236.9

- Well, it was a,

Time: 9238.04

in the case of my parents it was interesting to watch

Time: 9240.12

'cause there's obviously tension.

Time: 9243.27

Like, there'll be times where they fought

Time: 9244.96

and all those kinds of things.

Time: 9246.752

They obviously get frustrated with each other and they like,

Time: 9251.78

but they find mechanisms

Time: 9253.36

how to communicate that to each other,

Time: 9255.04

like to make fun of each other a little bit,

Time: 9256.56

like to tease, to get some of that frustration out,

Time: 9259.44

and then ultimately to reunite

Time: 9261.04

and find their joyful moments

Time: 9263.53

and be that the energy.

Time: 9265.33

I think it's clear

Time: 9266.6

'cause I got together in there I think early 20s,

Time: 9268.87

like very, very young.

Time: 9269.913

I think you grow together as people.

Time: 9272.46

- Yeah.

Time: 9273.579

You're still in the critical period of brain plasticity.

Time: 9275.504

[laughing]

Time: 9277.09

- And also, I mean,

Time: 9278.49

it's just like divorce was so frowned upon

Time: 9282.31

that you stick it out.

Time: 9283.532

And I think a lot of couples especially from that time,

Time: 9285.92

the Soviet Union,

Time: 9286.753

that's probably applies to a lot of cultures.

Time: 9288.324

You stick it out and you put in the work,

Time: 9290.67

you learn how to put in the work.

Time: 9292.36

And once you do,

Time: 9293.31

you start to get to

Time: 9294.23

some of those rewarding aspects of being,

Time: 9297.882

like through time has sharing so many moments together.

Time: 9301.93

That's definitely something that was an inspiration to me,

Time: 9307.86

but maybe that's where I have.

Time: 9309.74

So I have a similar kind of longing

Time: 9311.39

to have a lifelong partner,

Time: 9312.86

like to have that kind of view,

Time: 9314.94

where same with friendship,

Time: 9316.79

lifelong friendship is the most meaningful kind.

Time: 9320.65

That there is something with that time

Time: 9322.09

of sharing all that time together.

Time: 9323.716

Like till death do us part is a powerful thing.

Time: 9326.85

Not by force,

Time: 9327.69

not because of the religion said it

Time: 9329.25

or the government said it or your culture said it,

Time: 9331.62

but because you want to.

Time: 9333.29

- Do you want children?

Time: 9334.56

- Definitely, yeah.

Time: 9335.96

Definitely want children.

Time: 9338.42

It's-

Time: 9339.253

- How many Roombas do you have?

Time: 9341.47

- Oh, I thought-

Time: 9342.303

- You should, no, no-

Time: 9343.136

- Human children?

Time: 9344.042

- No, human to human children.

Time: 9344.875

- 'Cause I already have the children.

Time: 9345.708

- Exactly.

Time: 9346.541

What I was saying,

Time: 9347.418

you probably need to at least as many human children

Time: 9348.84

as you do Roombas.

Time: 9349.673

Big family, small family.

Time: 9353.184

- So.

Time: 9354.017

- In your mind's eyes,

Time: 9354.85

they're a bunch of Fridman's running around.

Time: 9359.17

- So I'll tell you, like realistically,

Time: 9361.27

I can explain exactly my thinking,

Time: 9363.74

and this is similar to the robotics work is,

Time: 9367.44

if I'm like purely logical right now,

Time: 9370.13

my answer would be I don't want kids.

Time: 9372.39

Because I just don't have enough time.

Time: 9375.76

I have so much going on.

Time: 9377.32

But when I'm using the same kind of vision I use

Time: 9379.61

for the robots

Time: 9380.77

is I know my life will be transformed with the first.

Time: 9385.03

Like I know I would love being a father.

Time: 9387.67

And so the question of how many,

Time: 9389.689

that's on the other side of that hill.

Time: 9392.961

It could be some ridiculous number.

Time: 9395.93

So I just know that-

Time: 9397.01

- I have a feeling and I could be,

Time: 9398.67

I don't have a crystal ball, but I don't know.

Time: 9402.66

I see an upwards of,

Time: 9404.8

certainly three or more come comes to mind.

Time: 9407.85

- So much of that has to do

Time: 9409.538

with the partner you're with too.

Time: 9411.96

So like that, that's such an open question,

Time: 9415.127

especially in this society of what the right partnership is.

Time: 9418.595

'Cause I'm deeply empathetic.

Time: 9422.85

I want to see, like to me,

Time: 9425.15

what I look for in your relationship

Time: 9426.63

is for me to be really excited

Time: 9429.81

about the passions of another person,

Time: 9432.11

like whatever they're into,

Time: 9433.07

it doesn't have to be a career success.

Time: 9435.77

Any kind of success,

Time: 9436.9

just to be excited for them,

Time: 9438.49

and for them to be excited for me.

Time: 9440.57

And like share in that excitement and build,

Time: 9442.48

and build and build.

Time: 9443.399

But there was also practical aspects of like,

Time: 9445.84

what kind of shit do you enjoy doing together?

Time: 9448.677

And I think family is a real serious undertaking.

Time: 9452.564

- It certainly is.

Time: 9454.07

I mean, I think that I have a friend who said it,

Time: 9457.52

I think best,

Time: 9458.353

which is that you first have,

Time: 9461.26

he's in a very successful relationship

Time: 9463.02

and has a family.

Time: 9464.49

And he said,

Time: 9465.323

you first have to define the role

Time: 9467.51

and then you have to cast the right person for the role.

Time: 9469.954

[Lex laughing]

Time: 9471.41

- Well, yeah, there's some deep aspects of that,

Time: 9473.827

but there's also an aspect to which you're not smart enough

Time: 9478.32

from this side of it to define the role.

Time: 9482.77

I think there's part of it that has to be a leap

Time: 9484.87

that you have to take.

Time: 9486.55

And I see having kids that way.

Time: 9492.16

You just have to go with it and figure it out also.

Time: 9496.15

As long as there's love there,

Time: 9497.68

like what the hell is life for even?

Time: 9500.62

So I've,

Time: 9501.9

there's so many incredibly successful people that I know

Time: 9506.35

that I've gotten to know that all have kids.

Time: 9509

And the presence of kids for the most part

Time: 9512.344

has only been something that energizes them,

Time: 9516.64

something they gave them meaning,

Time: 9517.93

something that made them the best version of themselves,

Time: 9520.24

like made them more productive, not less,

Time: 9522.42

which is fascinating to me.

Time: 9523.47

- It is fascinating.

Time: 9524.52

I mean, you can imagine

Time: 9525.353

if the way that you felt about Homer,

Time: 9527.29

the way that I feel and felt about Costello

Time: 9529.79

is at all a glimpse of what that must be like then.

Time: 9534.65

- Exactly.

Time: 9536.45

The downside,

Time: 9537.76

the thing I worry more about is the partner side of that.

Time: 9544.61

I've seen,

Time: 9546.35

the kids are almost universally

Time: 9547.87

a source of increased productivity and joy and happiness.

Time: 9551.75

Like, yeah, they're a pain in the ass.

Time: 9553.31

Yeah, is complicated.

Time: 9554.143

Yeah, so and so forth, people like to complain about kids.

Time: 9557.48

But then when you actually look past

Time: 9559.65

that little shallow layer of complaint, kids are great.

Time: 9562.98

The source of pain for a lot of people

Time: 9564.81

is if when the relationship doesn't work.

Time: 9567.62

And so I'm very kind of concerned about like,

Time: 9572.64

dating is very difficult,

Time: 9573.94

and I'm a complicated person.

Time: 9576.6

And so it's been very difficult

Time: 9578.27

to find the right kind of person.

Time: 9581.575

But that statement doesn't even make sense

Time: 9585.08

because I'm not on dating apps, I don't see people.

Time: 9588.3

You're like the first person I saw in awhile.

Time: 9590.42

It's like you and Michael Malice and like Joe.

Time: 9593.08

So, like, I don't think I've seen like a female.

Time: 9599.88

What is it?

Time: 9600.89

An element of the female species in quite a while.

Time: 9603.71

So, I think you have to put yourself out there.

Time: 9606.56

What is it?

Time: 9607.48

Daniel Johnston says, true love will find you,

Time: 9610.049

but only if you're looking.

Time: 9611.78

So there's some element of really taking the leap

Time: 9613.72

and putting yourself out there

Time: 9614.72

in kind of different situations.

Time: 9617.03

And I don't know how to do that

Time: 9618.46

when you're behind a computer all the time.

Time: 9620.5

- Well, you're a builder and you're a problem solver,

Time: 9625.24

and you find solutions,

Time: 9629.05

and I'm confident the solution is out there, and.

Time: 9634.569

- I think you're implying

Time: 9635.402

that I'm going to build the girlfriend, which I think.

Time: 9638.568

- Well, and maybe we shouldn't separate this friendship,

Time: 9643.057

the notion of friendship and community,

Time: 9645.5

and if we go back to this concept of the aggregate,

Time: 9648.95

maybe you'll meet this woman through a friend,

Time: 9652.132

or maybe you'll or something of that sort.

Time: 9653.653

- So, one of the things,

Time: 9655.253

I dunno if you feel the same way,

Time: 9656.95

I definitely one of those people that just falls in love

Time: 9661.55

and that's it.

Time: 9662.51

- Yeah, I can't say I'm like that.

Time: 9664.03

With Costello it was instantaneous.

Time: 9666.62

- Yeah.

Time: 9667.453

- It really was.

Time: 9668.286

I mean, I know it's not romantic love,

Time: 9669.72

but it was instantaneous.

Time: 9670.77

No, I, but that's me.

Time: 9672.447

And I think that if you know, you know,

Time: 9674.72

because that's a good thing that you have there.

Time: 9678.457

- It's, I'm very careful with that,

Time: 9681.78

because you don't want to fall in love with the wrong person.

Time: 9684.88

So I try to be very kind of careful with,

Time: 9687.59

I've noticed this because I fall in love with every,

Time: 9689.48

like this mug,

Time: 9690.56

everything I fall in love with things in this world.

Time: 9694.05

So, like, you have to be really careful

Time: 9695.55

because a girl comes up to you and says

Time: 9701.08

she loves DUSTY HUSKY,

Time: 9703.929

that doesn't necessarily mean to marry her tonight.

Time: 9706.74

- Yes.

Time: 9707.573

And I like the way you said that out loud

Time: 9709.1

so that you heard it,

Time: 9710

you doesn't mean you need to marry her tonight, right?

Time: 9712.62

- [Lex] Exactly.

Time: 9713.453

- [Andrew] Right.

Time: 9714.447

- But I mean, but people are amazing,

Time: 9716.14

and people are beautiful and that's,

Time: 9718.19

so I'm fully embraced that,

Time: 9720.457

but also you have to be careful with relationships.

Time: 9722.97

And at the same time,

Time: 9723.912

like I mentioned to you offline, I don't,

Time: 9727.84

there's something about me

Time: 9728.93

that appreciates swinging for the fences and not dating,

Time: 9733.68

like doing serial dating, or dating around.

Time: 9735.503

- Like you're a one guy, one girl kind of guy.

Time: 9737.61

- [Lex] Yeah.

Time: 9738.443

- And you said that.

Time: 9739.276

- And it's tricky

Time: 9741.1

because you want to be careful with that kind of stuff.

Time: 9743.9

Especially now there's a growing platform

Time: 9746.086

that have a ridiculous amount of female interests

Time: 9749.19

of a certain kind.

Time: 9751.4

But I'm looking for deep connection,

Time: 9753.347

and I'm looking by sending home alone,

Time: 9756.163

and every once in a while talking to Stanford professors.

Time: 9760.699

- Perfect solution. - On a podcast.

Time: 9762.44

- Perfect solution.

Time: 9763.273

- Is going to workout great.

Time: 9764.29

- It's well,

Time: 9765.34

it's part of,

Time: 9767.3

that constitutes machine learning of sorts.

Time: 9769.58

- Yeah, of sorts.

Time: 9771.13

- I do,

Time: 9771.963

you mentioned what has now become a quite extensive

Time: 9775.89

and expansive public platform, which is incredible.

Time: 9779.3

I mean, the number of people out,

Time: 9781.3

first time I saw your podcast,

Time: 9782.39

I noticed the suit,

Time: 9783.57

I was like, he respects his audience, which was great,

Time: 9785.5

but I also thought this is amazing.

Time: 9788.055

People are showing up for science and engineering

Time: 9790.79

and technology information and those discussions

Time: 9792.85

and other sorts of discussions.

Time: 9794.16

Now, I do want to talk for a moment about the podcast.

Time: 9798.14

So my two questions about the podcast are,

Time: 9801.76

when you started it, did you have a plan?

Time: 9804.34

And regardless of what that answer is,

Time: 9806.988

do you know where you're taking it,

Time: 9809.87

or would you like to leave us?

Time: 9811.758

I do believe in an element of surprise is always fun.

Time: 9815.25

But what about the podcast?

Time: 9816.44

Do you enjoy the podcast?

Time: 9817.71

I mean, your audience certainly includes me,

Time: 9820.38

really enjoys the podcast.

Time: 9821.76

It's incredible.

Time: 9822.67

- So I love talking to people,

Time: 9826.59

and there's something about microphones

Time: 9830.23

that really bring out the best in people.

Time: 9831.95

Like you don't get a chance to talk like this.

Time: 9834.53

If you and I were just hanging out,

Time: 9836.13

we would have a very different conversation

Time: 9837.724

in the amount of focus we allocate to each other.

Time: 9841.91

We would be having fun talking about other stuff

Time: 9844.42

and doing other things.

Time: 9845.726

There'll be a lot of distraction.

Time: 9847.52

There would be some phone use and all that kind of stuff.

Time: 9851.07

But here we're 100% focused on each other

Time: 9854.02

and focus on the idea.

Time: 9856.05

And like sometimes playing with ideas

Time: 9858.24

that we both don't know the answer to,

Time: 9861.13

like a question we don't know the answer to.

Time: 9863.12

We're both like fumbling with it,

Time: 9864.4

trying to figure out,

Time: 9865.53

trying to get some insights

Time: 9867.15

at something we haven't really figured out before

Time: 9869.56

and together arriving at that.

Time: 9871.31

I think that's magical.

Time: 9872.44

I don't know why we need microphones for that,

Time: 9874.24

but we somehow do.

Time: 9875.28

- It feels like doing science.

Time: 9876.78

- It feels like doing science for me, definitely.

Time: 9878.76

That's exactly it.

Time: 9880.222

And I'm really glad you said that

Time: 9882.36

because I don't actually often say this,

Time: 9885.553

but that's exactly what I felt like.

Time: 9888.124

I wanted to talk to friends and colleagues at MIT

Time: 9893.97

to do real science together.

Time: 9896.46

That's how I felt about it.

Time: 9897.83

Like to really talk through problems

Time: 9899.83

that are actually interesting,

Time: 9903.15

as opposed to like incremental work

Time: 9906.1

that we're currently working for

Time: 9909.51

for a particular conference.

Time: 9910.89

So really asking questions like,

Time: 9912.32

what are we doing?

Time: 9914.17

Like, where's this headed to?

Time: 9916.03

Like, what are the big,

Time: 9917.24

is this really going to help us solve,

Time: 9919.89

in the case of AI, solve intelligence?

Time: 9922.83

Like, is this even working on intelligence?

Time: 9924.835

There's a certain sense,

Time: 9926.41

which is why I initially called it artificial intelligence.

Time: 9930.02

Is like most of us are not working

Time: 9932.94

on artificial intelligence.

Time: 9934.32

You're working on some very specific problem

Time: 9937.72

and a set of techniques,

Time: 9939.207

at the time it's machine learning

Time: 9941.31

to solve this particular problem.

Time: 9943.01

This is not going to take us to a system

Time: 9945.061

that is anywhere close to the generalizability

Time: 9949.25

of the human mind.

Time: 9951.37

Like the kind of stuff the human mind can do

Time: 9952.98

in terms of memory, in terms of cognition,

Time: 9954.7

in terms of reasoning,

Time: 9955.73

common sense reasoning.

Time: 9956.98

This doesn't seem to take us there.

Time: 9958.73

So the initial impulse was,

Time: 9960.57

can I talk to these folks

Time: 9962.541

do science together through conversation?

Time: 9965.239

And I also thought that there was not enough,

Time: 9971.61

I didn't think there was enough good conversations

Time: 9973.389

with world-class minds that I got to meet.

Time: 9977.7

And not the ones with the book,

Time: 9979.85

or like this was just the thing.

Time: 9981.79

Oftentimes you go on this tour when you have a book,

Time: 9984.3

but there's a lot of minds that don't write books.

Time: 9986.95

- And the books constrain the conversation too,

Time: 9988.714

when you're talking about this thing, this book.

Time: 9991.77

- But there's,

Time: 9992.66

I've noticed that,

Time: 9994.03

with people that haven't written a book who are brilliant,

Time: 9997.29

we get to talk about ideas in a new way.

Time: 10000.11

We both haven't actually,

Time: 10002.65

when we raise a question,

Time: 10003.81

we don't know the answer to it once the question is raised.

Time: 10007.13

And we try to arrive there.

Time: 10009.013

Like, I dunno,

Time: 10010.7

I remember asking questions of world-class researchers

Time: 10015.09

in deep learning of,

Time: 10018.26

why do neural networks work as well as they do?

Time: 10022.64

That question is often loosely asked,

Time: 10026.66

but like when you have microphones

Time: 10028.938

and you have to think through it,

Time: 10031.04

and you have 30 minutes to an hour

Time: 10032.5

to think through it together, I think that's science.

Time: 10036.15

I think that's really powerful.

Time: 10037.66

So that was the one goal.

Time: 10039.72

The other one is,

Time: 10043.546

again, don't usually talk about this,

Time: 10045.5

but there's some sense in which I wanted

Time: 10048.54

to have dangerous conversations.

Time: 10052.49

Part of the reasons I wanted to wear a suit

Time: 10055.5

is like, I want it to be fearless.

Time: 10057.941

The reason I don't usually talk about it

Time: 10060.22

is because I feel like I'm not good at conversation.

Time: 10063.04

So it looks like it doesn't match the current skill level.

Time: 10068.28

But I wanted to have really dangerous conversations

Time: 10073.9

that I uniquely would be able to do.

Time: 10078.21

Not completely uniquely,

Time: 10080.32

but like, I'm a huge fan of Joe Rogan,

Time: 10082.5

and I had to ask myself,

Time: 10084.57

what conversations can I do that Joe Rogan can't?

Time: 10087.932

For me, I know I bring this up,

Time: 10092.01

but for me that person I thought about

Time: 10093.8

at the time was Putin.

Time: 10095.69

Like that's why I bring him up.

Time: 10097.57

He's just like with Costello,

Time: 10100.49

he's not just a person.

Time: 10102.11

He's also an idea to me for what I strive for.

Time: 10105.63

Just to have those dangerous conversations.

Time: 10107.9

And the reason I'm uniquely qualified

Time: 10110.67

as both the Russian,

Time: 10111.503

but also there's the judo and the martial arts,

Time: 10114.08

there's a lot of elements that make me have a conversation

Time: 10117.85

he hasn't had before.

Time: 10119.168

And there's a few other people

Time: 10123.2

that I kept in mind, like Don Knuth,

Time: 10126.6

is a computer scientist from Stanford

Time: 10128.968

that I thought is one of the most beautiful minds ever.

Time: 10134.09

And nobody really talked to him,

Time: 10137.32

like really talked to him.

Time: 10139.54

He did a few lectures, which people love,

Time: 10141.42

but really just have a conversation with him.

Time: 10143.327

There's a few people like that.

Time: 10144.85

One of them passed away, John Conway,

Time: 10146.75

that I never got,

Time: 10147.583

we agreed to talk, but he died before we did.

Time: 10150.69

There's a few people like that,

Time: 10152

that I thought like it's such a crime

Time: 10155.73

to not hear those folks.

Time: 10159.45

And I have the unique ability

Time: 10162.52

to know how to purchase a microphone on Amazon

Time: 10166.88

and plug it into a device that records audio

Time: 10169.48

and then publish it, which seems relatively unique.

Time: 10172.09

Like that's not easy in the scientific community.

Time: 10174.75

People knowing how to plug in a microphone.

Time: 10176.74

- No.

Time: 10177.573

They can build Faraday cages,

Time: 10178.96

and two-photon microscopes and bioengineer,

Time: 10182.24

all sorts of things,

Time: 10183.08

but the idea that you could take ideas

Time: 10186.42

and export them into a structure or a pseudo structure

Time: 10189.32

that people would benefit from

Time: 10190.71

seems like a cosmic achievement to them.

Time: 10194.09

- I don't know if it's fear

Time: 10195.53

or just a basically they haven't tried it,

Time: 10198.193

so they haven't learned the skill level.

Time: 10200.37

- But I think they're not trained.

Time: 10201.64

I mean, we could riff on this for awhile,

Time: 10203.41

but I think that,

Time: 10205.5

but it's important.

Time: 10206.37

And maybe we should, which is that it's,

Time: 10208.928

they're not trained to do it.

Time: 10211.09

They're trained to think in specific games

Time: 10212.86

and specific hypotheses,

Time: 10214.3

and many of them don't care to, right?

Time: 10217.979

They became scientists because that's where they felt safe,

Time: 10222.71

and so why would they leave that Haven of safety?

Time: 10225.92

- Well, they also don't necessarily always see

Time: 10227.9

the value in it.

Time: 10228.8

We're all together learning,

Time: 10230.64

you and I are learning the value of this.

Time: 10233.78

I think you're probably having

Time: 10236.01

an exceptionally successful and amazing podcast

Time: 10239.233

that you started just recently.

Time: 10240.95

- Thanks to your encouragement.

Time: 10242.47

- Well, but there's a raw skill there that's,

Time: 10247.02

you're definitely an inspiration to me

Time: 10249.14

in how you did the podcast

Time: 10250.437

in the level of excellence you reach.

Time: 10252.88

But I think you've discovered

Time: 10254.114

that that's also an impactful way to do science.

Time: 10257.08

That podcast.

Time: 10258.18

And I think a lot of scientists

Time: 10260.24

have not yet discovered that.

Time: 10262.053

That this is a,

Time: 10263.25

if they apply same kind of rigor

Time: 10266.9

as they do to academic publication

Time: 10268.82

or to even conference presentations,

Time: 10271.487

and they do that rigor and effort to podcast,

Time: 10276.18

whatever that is,

Time: 10277.013

that could be a five-minute podcast,

Time: 10278.55

a two-hour podcasts,

Time: 10279.61

it could be conversational,

Time: 10280.717

or it can be more like lecture like,

Time: 10282.96

if they apply that effort,

Time: 10284.128

you have the potential to reach over time,

Time: 10286.87

tens of thousands, hundreds of thousands,

Time: 10288.64

millions of people.

Time: 10289.907

And that's really, really powerful.

Time: 10292.51

But yeah, for me giving a platform to a few of those folks,

Time: 10299.44

especially for me personally,

Time: 10300.99

so maybe you can speak to what fields you're drawn to,

Time: 10306.3

but I thought computer scientists

Time: 10311.2

were especially bad at this.

Time: 10313.59

So there's brilliant computer scientists

Time: 10316.33

that I thought it would be amazing to explore their mind,

Time: 10320.32

explore their thinking.

Time: 10322.09

And so that I took that almost as an,

Time: 10325.19

on as an effort.

Time: 10326.58

And at the same time I had other guests in mind,

Time: 10331.21

or people that connect to my own interests.

Time: 10333.94

So the wrestling.

Time: 10336.84

Wrestling, music, football,

Time: 10338.93

both American football and soccer.

Time: 10341.13

I have a few particular people

Time: 10342.65

that I'm really interested in.

Time: 10344.781

Buvaisar Saitiev.

Time: 10346.567

The Saitiev brothers, even Khabib for wrestling,

Time: 10349.68

just to talk to them, 'cause.

Time: 10350.91

- Oh, 'cause you can,

Time: 10351.75

you guys can communicate-

Time: 10353.05

- In Russian and in wrestling, right?

Time: 10356.42

As wrestlers and as Russians.

Time: 10358.88

And so that little,

Time: 10361.79

it's like an opportunity to explore a mind

Time: 10363.763

that I'm able to bring to the world.

Time: 10367.72

And also it,

Time: 10370.46

I feel like it makes me a better person,

Time: 10373.404

just that being that vulnerable

Time: 10375.67

and exploring ideas together.

Time: 10377.41

I don't know, like good conversation.

Time: 10379.85

I don't know how often

Time: 10380.683

you have really good conversation with friends,

Time: 10382.27

but like podcasts are like that.

Time: 10383.991

And it's deeply moving.

Time: 10386.49

- It's the best.

Time: 10387.96

And what you brought through.

Time: 10389.97

I mean, when I saw you sit down with Penrose,

Time: 10392.23

Nobel Prize winning physicist,

Time: 10393.98

and these other folks that,

Time: 10395.06

it's not just 'cause he has a Nobel,

Time: 10396.34

it's what comes out of his mouth is incredible.

Time: 10398.07

And what you were able to hold in that conversation

Time: 10402.94

was so much better,

Time: 10404.44

light years beyond what he had any other interviewer,

Time: 10408.91

I don't want to even call you an interviewer

Time: 10410.14

'cause it's really about conversation.

Time: 10411.65

Light years beyond what anyone else

Time: 10413.24

had been able to engage with him

Time: 10416.76

was such a beacon of what's possible.

Time: 10420

And I know that,

Time: 10420.97

I think that's what people are drawn to.

Time: 10422.48

And there's a certain intimacy,

Time: 10424.45

that certainly to people, our friends, as we are,

Time: 10427.58

and they know each other,

Time: 10428.63

that there's more of that,

Time: 10429.9

but there's an intimacy

Time: 10431.38

in those kinds of private conversations

Time: 10433.23

that are made public.

Time: 10434.99

And.

Time: 10435.92

- Well, that's the, with you,

Time: 10437.94

you're probably starting to realize, and Costello,

Time: 10441.23

is like, part of it,

Time: 10443.51

because you're authentic

Time: 10444.6

and you're putting yourself out there completely,

Time: 10446.726

people are almost not just consuming

Time: 10450.165

the words you're saying,

Time: 10453.33

they also enjoy watching you, Andrew,

Time: 10457.03

struggle with these ideas

Time: 10459.13

or try to communicate these ideas.

Time: 10460.74

They like the flaws,

Time: 10461.75

they like a human being.

Time: 10463.64

- Oh, good, that flaws.

Time: 10464.93

- Well, that's good 'cause I got plenty of those.

Time: 10466.26

- But they like the self-critical aspects,

Time: 10468.7

like where you're very careful,

Time: 10470.36

where you're very self-critical about your flaws.

Time: 10472.796

I mean, in that same way,

Time: 10474.9

it's interesting I think for people

Time: 10476.25

to watch me talk to Penrose,

Time: 10477.99

not just because Penrose is communicating ideas,

Time: 10482.33

but here's this like silly kid trying to explore ideas.

Time: 10487

Like they know this kid.

Time: 10488.396

There's a human connection that is really powerful.

Time: 10491.27

Same, I think with Putin, right?

Time: 10493.61

Like it's not just as a good interview with Putin,

Time: 10497.49

it's also, here's this kid struggling

Time: 10500.2

to talk with one of the most powerful,

Time: 10504.81

some will argue dangerous people in the world.

Time: 10508.3

They love that.

Time: 10509.24

The authenticity that led up to that.

Time: 10512.084

And in return,

Time: 10514.2

I get to connect,

Time: 10515.14

everybody I run to in the street

Time: 10516.84

and all those kinds of things,

Time: 10519.068

there's a depth of connection there,

Time: 10521.06

almost within like a minute or two that's unlike any other.

Time: 10524.35

- Yeah, there's an intimacy

Time: 10525.27

that you've formed with with them.

Time: 10526.88

- Yeah, we've been on this like journey together.

Time: 10529.77

And yeah, I have the same thing with Joe Rogan

Time: 10531.33

before I ever met him, right?

Time: 10532.394

Like I was,

Time: 10533.83

because I was a fan of Joe for so many years,

Time: 10536.107

there's something,

Time: 10538.42

there's a kind of friendship as absurd as it might be

Time: 10542.67

to say in podcasting and listening to podcasts.

Time: 10545.99

- Yeah.

Time: 10546.823

Maybe it fills in a little bit of that

Time: 10548.9

or solves a little bit of that loneliness

Time: 10550.69

that you've been talking about earlier.

Time: 10551.523

- Until the robots are here.

Time: 10553.11

[laughing]

Time: 10554.55

- I have just a couple more questions,

Time: 10556.57

but one of them is on behalf of your audience, which is,

Time: 10561.65

I'm not going to ask you the meaning of the hedgehog,

Time: 10564.51

but I just want to know,

Time: 10566.44

does it have a name?

Time: 10568.5

And you don't have to tell us the name,

Time: 10569.95

but just, does it have a name?

Time: 10571.01

Yes or no?

Time: 10572.035

- Well, there's a name he likes to be referred to as,

Time: 10577.55

and then there's a private name

Time: 10579.33

in the privacy of his own company that we call each other.

Time: 10581.32

No.

Time: 10582.365

[Lex laughing]

Time: 10583.32

I'm not that insane.

Time: 10584.4

No, his name is Hedgy.

Time: 10587.32

He's a hedgehog.

Time: 10588.67

I don't like stuffed animals.

Time: 10591.75

But his story is one of minimalism.

Time: 10595.63

So I gave away everything I own,

Time: 10599.25

now three times in my life.

Time: 10601.15

By everything I mean almost everything,

Time: 10603

kept jeans and shirt and a laptop.

Time: 10606.17

And recently it's also been guitar,

Time: 10609.14

things like that.

Time: 10611.29

But he survived because he was always in,

Time: 10615.36

at least in the first two times was in the laptop bag,

Time: 10618.59

and he just got lucky.

Time: 10620.34

And so I just liked the perseverance of that.

Time: 10623.02

And I first saw him in the,

Time: 10626

the reason I got a stuffed animal,

Time: 10627.34

I don't have other stuffed animals,

Time: 10629.63

is it was in a thrift store,

Time: 10633.01

in this like giant pile of stuffed animals

Time: 10636.05

and he jumped out at me,

Time: 10638.13

because unlike all the rest of them,

Time: 10640.05

he has this intense mean look about him.

Time: 10645.27

That he's just,

Time: 10646.9

he's upset at life,

Time: 10649.64

at the cruelty of life.

Time: 10650.94

And it's just,

Time: 10651.773

especially in the contrast of the other stuffed animals,

Time: 10653.393

they have this dumb smile on their face.

Time: 10655.89

If you look at most stuffed animals,

Time: 10657.344

they have this dumb look on their face.

Time: 10658.75

They're just happy.

Time: 10659.68

Is like "Pleasantville."

Time: 10660.59

- It's what we say in neuroscience,

Time: 10661.7

they have a smooth cortex,

Time: 10663.015

not many form.

Time: 10664.47

- Exactly.

Time: 10665.303

And this,

Time: 10666.136

like Hedgy like saw through all of it.

Time: 10668.16

He was like Dustyesky's man from underground.

Time: 10672.04

I mean, there's a sense

Time: 10672.92

that he saw the darkness of the world and persevered.

Time: 10676.6

So like,

Time: 10677.433

and there's also a famous Russian cartoon,

Time: 10680.667

"Hedgehog in the Fog" that I grew up with,

Time: 10683.95

I connected with.

Time: 10684.95

[Lex laughing]

Time: 10685.81

People who know of that cartoon,

Time: 10687.63

you could see it on YouTube, it's.

Time: 10689.657

- "Hedgehog in the Fog."

Time: 10690.83

- Yeah.

Time: 10691.733

[Lex laughing]

Time: 10693.83

It's just, as you would expect,

Time: 10696.108

especially from like early Soviet cartoons.

Time: 10697.94

It's a hedgehog, like sad,

Time: 10701.1

walking through the fog,

Time: 10702.91

exploring like loneliness and sadness.

Time: 10705.34

It's like, but it's beautiful.

Time: 10706.81

It's like a piece of art,

Time: 10707.87

people should,

Time: 10708.703

even if you don't speak Russian,

Time: 10709.63

you'll see, you'll understand.

Time: 10711.61

- Oh, the moment you said that I was going to ask,

Time: 10713.97

so it's in Russian?

Time: 10714.803

But of course it's in-

Time: 10715.636

- It's in Russian, but it's more,

Time: 10717.37

it's very little speaking in it.

Time: 10719.06

It's almost a,

Time: 10721.18

there's an interesting exploration

Time: 10723.09

of how you make sense of the world

Time: 10727.41

when you see it only vaguely through the fog.

Time: 10732.2

So he's trying to understand the world.

Time: 10735.47

- We have Mickey Mouse,

Time: 10736.757

we have Bugs Bunny.

Time: 10738.77

We have all these crazy animals,

Time: 10741.347

and you have the "Hedgehog in the Fog."

Time: 10743.61

- So there's a certain period,

Time: 10745.55

and this is again,

Time: 10746.727

I don't know what it's attributed to,

Time: 10749.29

but it was really powerful,

Time: 10750.427

which there's a period in Soviet history,

Time: 10752.99

I think probably '70s and '80s where like,

Time: 10758.42

especially kids were treated very seriously.

Time: 10761.33

Like they were treated like they're able to deal

Time: 10764.35

with the weightiness of life.

Time: 10767.85

And that was reflected in the cartoons.

Time: 10770.123

And there was,

Time: 10772.15

it was allowed to have like really artistic content,

Time: 10777.22

not like dumb cartoons

Time: 10778.64

that are trying to get you to like smile and run around,

Time: 10781.01

but like create art.

Time: 10782.35

Like stuff that,

Time: 10783.65

you know how like short cartoons

Time: 10785.35

or short films can win Oscars,

Time: 10786.636

like that's what they're swinging for.

Time: 10788.74

- So what strikes me about this is a little bit

Time: 10791.14

how we were talking about the suit earlier,

Time: 10792.81

it's almost like they treat kids with respect.

Time: 10795.16

- Yeah.

Time: 10795.993

- Like that they have an intelligence

Time: 10798.32

and they honor that intelligence.

Time: 10799.79

- Yeah, they're really just adult in a small body.

Time: 10803.4

Like you want to protect them

Time: 10804.65

from the true cruelty of the world.

Time: 10806.15

- Sure.

Time: 10806.983

- But in terms of their intellectual capacity

Time: 10808.49

or like philosophical capacity,

Time: 10810.29

they are right there with you.

Time: 10811.56

And so that the cartoons reflected that,

Time: 10813.914

the art that they consumed,

Time: 10816.06

education reflected that.

Time: 10817.75

So he represents that.

Time: 10819.15

I mean, there's a sense of,

Time: 10822.62

because it's survived so long

Time: 10824.73

and because I don't like stuffed animals,

Time: 10827.37

that it's like we've been through all of this together

Time: 10830.95

and it's the same,

Time: 10831.93

sharing the moments together as the friendship.

Time: 10834.38

And there's a sense in which,

Time: 10836.25

if all the world turns on you and goes to hell,

Time: 10839.04

at least we got each other.

Time: 10841.22

And he doesn't die,

Time: 10842.43

because he's an inanimate object, so.

Time: 10845.52

- Until you animate him.

Time: 10847.24

- Until you animate him.

Time: 10849.1

And then I probably would want to know

Time: 10850.47

what he was thinking about this whole time.

Time: 10853.52

He's probably really into Taylor Swift

Time: 10855.24

or something like that.

Time: 10856.073

And it's like that I wouldn't even want to know anyway.

Time: 10859.55

- Well, I now feel a connection to Hedgy,

Time: 10861.84

the hedgehog that I certainly didn't have before.

Time: 10864.15

And I think that encapsulates

Time: 10865.4

the kind of possibility of connection

Time: 10869.67

that is possible between human and other object

Time: 10873.66

and through robotics certainly.

Time: 10877.77

There's a saying that I heard when I was a graduate student

Time: 10880

that's just been ringing in my mind

Time: 10882.56

throughout this conversation in such a,

Time: 10885.1

I think appropriate way, which is that,

Time: 10888.61

Lex, you are in a minority of one,

Time: 10891.35

you are truly extraordinary in your ability to encapsulate

Time: 10896.75

so many aspects of science, engineering,

Time: 10900.21

public communication, about so many topics,

Time: 10903.428

martial arts and the emotional depth that you bring to it.

Time: 10907.16

And just the purposefulness,

Time: 10909.09

and I think if it's not clear to people,

Time: 10911.68

it absolutely should be stated.

Time: 10913.93

But I think it's abundantly clear

Time: 10915.49

that just the amount of time

Time: 10917.82

and thinking that you put into things is,

Time: 10921.42

it is the ultimate mark of respect.

Time: 10924.85

So, I'm just extraordinarily grateful for your friendship

Time: 10928.07

and for this conversation.

Time: 10929.19

- I'm proud to be a friend.

Time: 10931.18

And I just wished you showed me

Time: 10932.44

the same kind of respect by wearing a suit

Time: 10934.34

and make your father proud maybe next time.

Time: 10936.596

[Andrew laughing]

Time: 10937.53

- Next time indeed.

Time: 10939.11

Thanks so much my friend.

Time: 10940.49

- Thank you.

Time: 10941.323

Thank you, Andrew.

Time: 10942.39

- Thank you for joining me for my discussion

Time: 10944.57

with Dr. Lex Fridman.

Time: 10946.32

If you're enjoying this podcast and learning from it,

Time: 10949.18

please consider subscribing on YouTube.

Time: 10951.4

As well, you can subscribe to us on Spotify or Apple.

Time: 10955.185

Please leave any questions and comments and suggestions

Time: 10958.25

that you have for future podcast episodes and guests

Time: 10960.88

in the comment section on YouTube.

Time: 10963.27

At Apple, you can also leave us up to a five-star review.

Time: 10966.82

If you'd like to support this podcast,

Time: 10968.49

we have a Patreon.

Time: 10969.67

That's patreon.com/andrewhuberman.

Time: 10972.86

And there you can support us at any level that you like.

Time: 10976.24

Also, please check out our sponsors mentioned

Time: 10978.57

at the beginning of the podcast episode.

Time: 10980.89

That's the best way to support this podcast.

Time: 10983.24

Links to our sponsors can be found in the show notes.

Time: 10986.8

And finally, thank you for your interest in science.

Time: 10990.532

[bright music]

Copyright © 2024. All rights reserved.