Sebastian Thrun: Flying Cars, Autonomous Vehicles, and Education

Lex Fridman (02:05.360)

I provide timestamps for the start of the conversation that you can skip to,

Lex Fridman (02:09.240)

but it helps if you listen to the ad and support this podcast

Lex Fridman (02:12.680)

by trying out the product or service being advertised.

Sebastian Thrun (02:16.440)

This show is presented by Cash App, the number one finance app in the App Store.

Lex Fridman (02:21.400)

I personally use Cash App to send money to friends,

Lex Fridman (02:24.000)

but you can also use it to buy, sell, and deposit Bitcoin in just seconds.

Lex Fridman (02:28.160)

Cash App also has a new investing feature.

Sebastian Thrun (02:31.040)

You can buy fractions of a stock, say $1 worth, no matter what the stock price is.

Lex Fridman (02:36.560)

Broker services are provided by Cash App Investing,

Sebastian Thrun (02:39.480)

a subsidiary of Square, and member SIPC.

Lex Fridman (02:42.920)

I'm excited to be working with Cash App

Sebastian Thrun (02:44.640)

to support one of my favorite organizations called FIRST,

Lex Fridman (02:47.840)

best known for their FIRST Robotics and LEGO competitions.

Sebastian Thrun (02:51.280)

They educate and inspire hundreds of thousands of students

Lex Fridman (02:54.640)

in over 110 countries and have a perfect rating on Charity Navigator,

Sebastian Thrun (02:59.000)

which means the donated money is used to maximum effectiveness.

Lex Fridman (03:03.080)

When you get Cash App from the App Store or Google Play

Lex Fridman (03:06.040)

and use code LEGSPODCAST, you'll get $10,

Lex Fridman (03:09.320)

and Cash App will also donate $10 to FIRST,

Sebastian Thrun (03:12.080)

which again is an organization that I've personally seen inspire girls and boys

Lex Fridman (03:16.640)

to dream of engineering a better world.

Lex Fridman (03:19.720)

And now, here's my conversation with Sebastian Thrun.

Lex Fridman (03:24.920)

You mentioned that The Matrix may be your favorite movie.

Lex Fridman (03:28.960)

So let's start with a crazy philosophical question.

Lex Fridman (03:32.160)

Do you think we're living in a simulation?

Lex Fridman (03:34.800)

And in general, do you find the thought experiment interesting?

Lex Fridman (03:40.000)

Define simulation, I would say.

Sebastian Thrun (03:42.240)

Maybe we are, maybe we are not,

Lex Fridman (03:43.720)

but it's completely irrelevant to the way we should act.

Sebastian Thrun (03:47.160)

Putting aside, for a moment,

Sebastian Thrun (03:49.880)

the fact that it might not have any impact on how we should act as human beings,

Sebastian Thrun (03:55.080)

for people studying theoretical physics,

Lex Fridman (03:57.280)

these kinds of questions might be kind of interesting,

Sebastian Thrun (03:59.560)

looking at the universe as an information processing system.

Lex Fridman (04:03.720)

The universe is an information processing system.

Sebastian Thrun (04:05.920)

It's a huge physical, biological, chemical computer, there's no question.

Lex Fridman (04:10.960)

But I live here and now.

Sebastian Thrun (04:12.880)

I care about people, I care about us.

Lex Fridman (04:15.600)

What do you think is trying to compute?

Sebastian Thrun (04:17.600)

I don't think there's an intention.

Lex Fridman (04:18.800)

I think the world evolves the way it evolves.

Lex Fridman (04:22.000)

And it's beautiful, it's unpredictable.

Lex Fridman (04:25.360)

And I'm really, really grateful to be alive.

Sebastian Thrun (04:28.040)

Spoken like a true human.

Lex Fridman (04:30.480)

Which last time I checked, I was.

Sebastian Thrun (04:33.360)

Or that, in fact, this whole conversation is just a touring test

Lex Fridman (04:36.480)

to see if indeed you are.

Sebastian Thrun (04:40.240)

You've also said that one of the first programs,

Lex Fridman (04:42.720)

or the first few programs you've written was a, wait for it, TI57 calculator.

Sebastian Thrun (04:49.080)

Yeah.

Lex Fridman (04:50.000)

Maybe that's early 80s.

Sebastian Thrun (04:52.000)

We don't want to date calculators or anything.

Lex Fridman (04:54.240)

That's early 80s, correct.

Sebastian Thrun (04:55.560)

Yeah.

Lex Fridman (04:56.440)

So if you were to place yourself back into that time, into the mindset you were in,

Sebastian Thrun (05:02.120)

could you have predicted the evolution of computing, AI,

Lex Fridman (05:06.840)

the internet technology in the decades that followed?

Sebastian Thrun (05:10.720)

I was super fascinated by Silicon Valley, which I'd seen on television once

Lex Fridman (05:14.960)

and thought, my god, this is so cool.

Sebastian Thrun (05:16.400)

They build like DRAMs there and CPUs.

Lex Fridman (05:19.600)

How cool is that?

Lex Fridman (05:20.440)

And as a college student a few years later, I decided to really study

Lex Fridman (05:25.240)

intelligence and study human beings.

Lex Fridman (05:26.920)

And found that even back then in the 80s and 90s,

Lex Fridman (05:30.560)

artificial intelligence is what fascinated me the most.

Sebastian Thrun (05:33.440)

What's missing is that back in the day, the computers are really small.

Lex Fridman (05:38.040)

The brains we could build were not anywhere bigger than a cockroach.

Lex Fridman (05:41.560)

And cockroaches aren't very smart.

Lex Fridman (05:43.760)

So we weren't at the scale yet where we are today.

Lex Fridman (05:46.320)

Did you dream at that time to achieve the kind of scale we have today?

Lex Fridman (05:51.040)

Or did that seem possible?

Sebastian Thrun (05:52.680)

I always wanted to make robots smart.

Lex Fridman (05:54.320)

And I felt it was super cool to build an artificial human.

Lex Fridman (05:57.960)

And the best way to build an artificial human was to build a robot,

Lex Fridman (06:00.680)

because that's kind of the closest we could do.

Sebastian Thrun (06:03.080)

Unfortunately, we aren't there yet.

Lex Fridman (06:04.920)

The robots today are still very brittle.

Lex Fridman (06:07.280)

But it's fascinating to study intelligence from a constructive

Lex Fridman (06:10.240)

perspective when you build something.

Sebastian Thrun (06:12.880)

To understand you build, what do you think it takes to build an intelligent

Lex Fridman (06:18.680)

system, an intelligent robot?

Sebastian Thrun (06:20.880)

I think the biggest innovation that we've seen is machine learning.

Lex Fridman (06:23.760)

And it's the idea that the computers can basically teach themselves.

Sebastian Thrun (06:28.600)

Let's give an example.

Lex Fridman (06:29.720)

I'd say everybody pretty much knows how to walk.

Lex Fridman (06:33.080)

And we learn how to walk in the first year or two of our lives.

Lex Fridman (06:36.800)

But no scientist has ever been able to write down the rules of human gait.

Sebastian Thrun (06:41.120)

We don't understand it.

Lex Fridman (06:42.080)

We have it in our brains somehow.

Sebastian Thrun (06:43.960)

We can practice it.

Lex Fridman (06:45.120)

We understand it.

Lex Fridman (06:46.560)

But we can't articulate it.

Lex Fridman (06:47.720)

We can't pass it on by language.

Lex Fridman (06:50.240)

And that, to me, is kind of the deficiency of today's computer programming.

Sebastian Thrun (06:53.320)

When you program a computer, they're so insanely dumb that you have to give them

Sebastian Thrun (06:57.640)

rules for every contingencies.

Lex Fridman (06:59.840)

Very unlike the way people learn from data and experience,

Sebastian Thrun (07:03.440)

computers are being instructed.

Lex Fridman (07:05.440)

And because it's so hard to get this instruction set right,

Sebastian Thrun (07:07.800)

we pay software engineers $200,000 a year.

Lex Fridman (07:11.480)

Now, the most recent innovation, which has been in the make for 30,

Sebastian Thrun (07:14.440)

40 years, is an idea that computers can find their own rules.

Lex Fridman (07:18.480)

So they can learn from falling down and getting up the same way children can

Sebastian Thrun (07:21.720)

learn from falling down and getting up.

Lex Fridman (07:23.840)

And that revolution has led to a capability that's completely unmatched.

Sebastian Thrun (07:28.720)

Today's computers can watch experts do their jobs, whether you're

Lex Fridman (07:32.120)

a doctor or a lawyer, pick up the regularities, learn those rules,

Lex Fridman (07:36.920)

and then become as good as the best experts.

Lex Fridman (07:39.400)

So the dream of in the 80s of expert systems, for example, had at its core

Sebastian Thrun (07:44.400)

the idea that humans could boil down their expertise on a sheet of paper,

Lex Fridman (07:49.360)

so to sort of reduce, sort of be able to explain to machines

Lex Fridman (07:53.280)

how to do something explicitly.

Lex Fridman (07:55.520)

So do you think, what's the use of human expertise into this whole picture?

Lex Fridman (08:00.040)

Do you think most of the intelligence will come from machines learning

Lex Fridman (08:03.240)

from experience without human expertise input?

Lex Fridman (08:06.480)

So the question for me is much more how do you express expertise?

Lex Fridman (08:10.680)

You can express expertise by writing a book.

Sebastian Thrun (08:12.960)

You can express expertise by showing someone what you're doing.

Lex Fridman (08:16.240)

You can express expertise by applying it by many different ways.

Lex Fridman (08:20.000)

And I think the expert systems was our best attempt in AI

Lex Fridman (08:23.680)

to capture expertise and rules.

Lex Fridman (08:25.960)

But someone sat down and said, here are the rules of human gait.

Lex Fridman (08:28.600)

Here's when you put your big toe forward and your heel backwards

Lex Fridman (08:32.600)

and you always stop stumbling.

Lex Fridman (08:34.720)

And as we now know, the set of rules, the set of language that we can command

Sebastian Thrun (08:39.480)

is incredibly limited.

Lex Fridman (08:41.200)

The majority of the human brain doesn't deal with language.

Sebastian Thrun (08:43.760)

It deals with subconscious, numerical, perceptual things

Lex Fridman (08:48.160)

that we don't even self aware of.

Sebastian Thrun (08:51.360)

Now, when an AI system watches an expert do their job and practice their job,

Lex Fridman (08:57.880)

it can pick up things that people can't even put into writing,

Sebastian Thrun (09:01.680)

into books or rules.

Lex Fridman (09:03.200)

And that's where the real power is.

Sebastian Thrun (09:04.520)

We now have AI systems that, for example, look over the shoulders

Lex Fridman (09:08.280)

of highly paid human doctors like dermatologists or radiologists,

Lex Fridman (09:12.840)

and they can somehow pick up those skills that no one can express in words.

Lex Fridman (09:18.440)

So you were a key person in launching three revolutions,

Sebastian Thrun (09:22.200)

online education, autonomous vehicles, and flying cars or VTOLs.

Lex Fridman (09:28.240)

So high level, and I apologize for all the philosophical questions.

Sebastian Thrun (09:34.680)

There's no apology necessary.

Lex Fridman (09:37.400)

How do you choose what problems to try and solve?

Lex Fridman (09:40.640)

What drives you to make those solutions a reality?

Lex Fridman (09:43.400)

I have two desires in life.

Sebastian Thrun (09:44.840)

I want to literally make the lives of others better.

Lex Fridman (09:48.560)

Or as we often say, maybe jokingly, make the world a better place.

Sebastian Thrun (09:52.840)

I actually believe in this.

Lex Fridman (09:54.920)

It's as funny as it sounds.

Lex Fridman (09:57.720)

And second, I want to learn.

Lex Fridman (09:59.160)

I want to get new skills.

Sebastian Thrun (10:00.400)

I don't want to be in a job I'm good at, because if I'm in a job

Lex Fridman (10:02.920)

that I'm good at, the chances for me to learn something interesting

Sebastian Thrun (10:05.840)

is actually minimized.

Lex Fridman (10:06.760)

So I want to be in a job I'm bad at.

Sebastian Thrun (10:09.040)

That's really important to me.

Lex Fridman (10:10.240)

So in a bill, for example, what people often

Sebastian Thrun (10:12.160)

call flying cars, these are electrical, vertical, takeoff,

Lex Fridman (10:15.320)

and landing vehicles.

Sebastian Thrun (10:17.960)

I'm just no expert in any of this.

Lex Fridman (10:19.720)

And it's so much fun to learn on the job what it actually means

Sebastian Thrun (10:23.080)

to build something like this.

Lex Fridman (10:24.920)

Now, I'd say the stuff that I've done lately

Sebastian Thrun (10:27.560)

after I finished my professorship at Stanford,

Lex Fridman (10:31.120)

they really focused on what has the maximum impact on society.

Sebastian Thrun (10:35.520)

Transportation is something that has transformed the 21st

Lex Fridman (10:38.240)

or 20th century more than any other invention,

Sebastian Thrun (10:40.120)

in my opinion, even more than communication.

Lex Fridman (10:42.600)

And cities are different.

Sebastian Thrun (10:43.600)

Workers are different.

Lex Fridman (10:45.080)

Women's rights are different because of transportation.

Lex Fridman (10:47.920)

And yet, we still have a very suboptimal transportation

Lex Fridman (10:51.360)

solution where we kill 1.2 or so million people every year

Sebastian Thrun (10:56.960)

in traffic.

Lex Fridman (10:57.680)

It's like the leading cause of death for young people

Sebastian Thrun (10:59.880)

in many countries, where we are extremely inefficient

Lex Fridman (11:02.880)

resource wise.

Sebastian Thrun (11:03.600)

Just go to your average neighborhood city

Lex Fridman (11:06.800)

and look at the number of parked cars.

Sebastian Thrun (11:08.320)

That's a travesty, in my opinion.

Lex Fridman (11:10.400)

Or where we spend endless hours in traffic jams.

Lex Fridman (11:13.840)

And very, very simple innovations,

Lex Fridman (11:15.680)

like a self driving car or what people call a flying car,

Sebastian Thrun (11:18.800)

could completely change this.

Lex Fridman (11:20.240)

And it's there.

Sebastian Thrun (11:21.000)

I mean, the technology is basically there.

Lex Fridman (11:23.280)

You have to close your eyes not to see it.

Lex Fridman (11:26.920)

So lingering on autonomous vehicles, a fascinating space,

Lex Fridman (11:30.720)

some incredible work you've done throughout your career there.

Lex Fridman (11:33.560)

So let's start with DARPA, I think, the DARPA challenge,

Lex Fridman (11:39.440)

through the desert and then urban to the streets.

Sebastian Thrun (11:42.840)

I think that inspired an entire generation of roboticists

Lex Fridman (11:45.720)

and obviously sprung this whole excitement

Sebastian Thrun (11:49.520)

about this particular kind of four wheeled robots

Lex Fridman (11:52.680)

we called autonomous cars, self driving cars.

Lex Fridman (11:55.520)

So you led the development of Stanley, the autonomous car

Lex Fridman (11:58.960)

that won the race to the desert, the DARPA challenge in 2005.

Lex Fridman (12:03.920)

And Junior, the car that finished second

Lex Fridman (12:07.400)

in the DARPA urban challenge, also did incredibly well

Sebastian Thrun (12:11.040)

in 2007, I think.

Lex Fridman (12:14.360)

What are some painful, inspiring, or enlightening

Lex Fridman (12:17.360)

experiences from that time that stand out to you?

Lex Fridman (12:20.560)

Oh my god.

Sebastian Thrun (12:22.640)

Painful were all these incredibly complicated,

Lex Fridman (12:28.160)

stupid bugs that had to be found.

Sebastian Thrun (12:30.440)

We had a phase where Stanley, our car that eventually

Lex Fridman (12:35.040)

won the DARPA grand challenge, would every 30 miles

Sebastian Thrun (12:38.120)

just commit suicide.

Lex Fridman (12:39.320)

And we didn't know why.

Lex Fridman (12:40.840)

And it ended up to be that in the sinking of two computer

Lex Fridman (12:44.360)

clocks, occasionally a clock went backwards

Lex Fridman (12:47.720)

and that negative time elapsed, screwed up

Lex Fridman (12:50.880)

the entire internal logic.

Lex Fridman (12:51.880)

But it took ages to find this.

Lex Fridman (12:54.360)

There were bugs like that.

Sebastian Thrun (12:56.280)

I'd say enlightening is the Stanford team immediately

Lex Fridman (12:59.840)

focused on machine learning and on software,

Sebastian Thrun (13:02.360)

whereas everybody else seemed to focus on building better hardware.

Lex Fridman (13:05.160)

Our analysis had been a human being with an existing rental

Sebastian Thrun (13:08.640)

car can perfectly drive the course

Lex Fridman (13:10.240)

but why do I have to build a better rental car?

Sebastian Thrun (13:12.160)

I just should replace the human being.

Lex Fridman (13:15.080)

And the human being, to me, was a conjunction of three steps.

Sebastian Thrun (13:18.840)

We had sensors, eyes and ears, mostly eyes.

Lex Fridman (13:22.360)

We had brains in the middle.

Lex Fridman (13:23.800)

And then we had actuators, our hands and our feet.

Lex Fridman (13:26.360)

Now, the actuators are easy to build.

Sebastian Thrun (13:28.200)

The sensors are actually also easy to build.

Lex Fridman (13:29.720)

What was missing was the brain.

Lex Fridman (13:30.960)

So we had to build a human brain.

Lex Fridman (13:32.640)

And nothing clearer than to me that the human brain

Sebastian Thrun (13:36.040)

is a learning machine.

Lex Fridman (13:37.000)

So why not just train our robot?

Lex Fridman (13:38.240)

So we would build massive machine learning

Lex Fridman (13:40.720)

into our machine.

Lex Fridman (13:42.320)

And with that, we were able to not just learn

Lex Fridman (13:44.840)

from human drivers.

Sebastian Thrun (13:45.680)

We had the entire speed control of the vehicle

Lex Fridman (13:47.960)

was copied from human driving.

Lex Fridman (13:49.840)

But also have the robot learn from experience

Lex Fridman (13:51.640)

where it made a mistake and recover from it

Lex Fridman (13:53.680)

and learn from it.

Lex Fridman (13:55.600)

You mentioned the pain point of software and clocks.

Sebastian Thrun (14:00.720)

Synchronization seems to be a problem that

Lex Fridman (14:04.720)

continues with robotics.

Sebastian Thrun (14:06.080)

It's a tricky one with drones and so on.

Lex Fridman (14:09.920)

What does it take to build a thing, a system

Lex Fridman (14:14.520)

with so many constraints?

Lex Fridman (14:16.640)

You have a deadline, no time.

Sebastian Thrun (14:20.320)

You're unsure about anything really.

Lex Fridman (14:22.080)

It's the first time that people really even exploring.

Sebastian Thrun (14:24.960)

It's not even sure that anybody can finish

Lex Fridman (14:26.800)

when we're talking about the race to the desert

Sebastian Thrun (14:28.840)

the year before nobody finish.

Lex Fridman (14:30.640)

What does it take to scramble and finish

Lex Fridman (14:32.760)

a product that actually, a system that actually works?

Lex Fridman (14:35.800)

We were very lucky.

Sebastian Thrun (14:36.760)

We were a really small team.

Lex Fridman (14:38.280)

The core of the team were four people.

Sebastian Thrun (14:40.440)

It was four because five couldn't comfortably sit

Lex Fridman (14:43.080)

inside a car, but four could.

Lex Fridman (14:45.360)

And I, as a team leader, my job was

Lex Fridman (14:47.080)

to get pizza for everybody and wash the car and stuff

Sebastian Thrun (14:50.120)

like this and repair the radiator when it broke

Lex Fridman (14:52.880)

and debug the system.

Lex Fridman (14:55.240)

And we were very open minded.

Lex Fridman (14:56.880)

We had no egos involved.

Sebastian Thrun (14:58.400)

We just wanted to see how far we can get.

Lex Fridman (15:00.840)

What we did really, really well was time management.

Sebastian Thrun (15:03.280)

We were done with everything a month before the race.

Lex Fridman (15:06.280)

And we froze the entire software a month before the race.

Lex Fridman (15:08.760)

And it turned out, looking at other teams,

Lex Fridman (15:11.440)

every other team complained if they had just one more week,

Sebastian Thrun (15:14.120)

they would have won.

Lex Fridman (15:15.440)

And we decided we're not going to fall into that mistake.

Sebastian Thrun (15:18.760)

We're going to be early.

Lex Fridman (15:19.920)

And we had an entire month to shake the system.

Lex Fridman (15:22.720)

And we actually found two or three minor bugs

Lex Fridman (15:24.920)

in the last month that we had to fix.

Lex Fridman (15:27.080)

And we were completely prepared when the race occurred.

Lex Fridman (15:30.000)

Okay, so first of all, that's such an incredibly rare

Sebastian Thrun (15:33.880)

achievement in terms of being able to be done on time

Lex Fridman (15:37.760)

or ahead of time.

Lex Fridman (15:39.000)

What do you, how do you do that in your future work?

Lex Fridman (15:43.080)

What advice do you have in general?

Sebastian Thrun (15:44.760)

Because it seems to be so rare,

Lex Fridman (15:46.360)

especially in highly innovative projects like this.

Sebastian Thrun (15:49.280)

People work till the last second.

Lex Fridman (15:50.840)

Well, the nice thing about the DARPA Grand Challenge

Sebastian Thrun (15:52.560)

is that the problem was incredibly well defined.

Lex Fridman (15:55.320)

We were able for a while to drive

Sebastian Thrun (15:57.160)

the old DARPA Grand Challenge course,

Lex Fridman (15:58.800)

which had been used the year before.

Lex Fridman (16:00.800)

And then at some reason we were kicked out of the region.

Lex Fridman (16:04.040)

So we had to go to a different desert, the Snorran Desert,

Lex Fridman (16:06.320)

and we were able to drive desert trails

Lex Fridman (16:08.880)

just of the same type.

Lex Fridman (16:10.600)

So there was never any debate about like,

Lex Fridman (16:12.320)

what is actually the problem?

Sebastian Thrun (16:13.240)

We didn't sit down and say,

Lex Fridman (16:14.400)

hey, should we build a car or a plane?

Sebastian Thrun (16:16.680)

We had to build a car.

Lex Fridman (16:18.280)

That made it very, very easy.

Sebastian Thrun (16:20.400)

Then I studied my own life and life of others.

Lex Fridman (16:23.800)

And we realized that the typical mistake that people make

Sebastian Thrun (16:26.360)

is that there's this kind of crazy bug left

Lex Fridman (16:29.600)

that they haven't found yet.

Lex Fridman (16:32.200)

And it's just, they regret it.

Lex Fridman (16:34.360)

And that bug would have been trivial to fix.

Sebastian Thrun (16:36.160)

They just haven't fixed it yet.

Lex Fridman (16:37.760)

They didn't want to fall into that trap.

Lex Fridman (16:39.600)

So I built a testing team.

Lex Fridman (16:41.080)

We had a testing team that built a testing booklet

Sebastian Thrun (16:43.760)

of 160 pages of tests we had to go through

Lex Fridman (16:46.800)

just to make sure we shake out the system appropriately.

Lex Fridman (16:49.720)

And the testing team was with us all the time

Lex Fridman (16:51.800)

and dictated to us today, we do railroad crossings.

Sebastian Thrun (16:55.520)

Tomorrow we do, we practice the start of the event.

Lex Fridman (16:58.480)

And in all of these, we thought,

Sebastian Thrun (17:00.680)

oh my God, it's long solved trivial.

Lex Fridman (17:02.240)

And then we tested it out.

Sebastian Thrun (17:03.200)

Oh my God, it doesn't do a railroad crossing.

Lex Fridman (17:04.560)

Why not?

Sebastian Thrun (17:05.400)

Oh my God, it mistakes the rails for metal barriers.

Lex Fridman (17:09.720)

We have to fix this.

Lex Fridman (17:11.600)

So it was really a continuous focus

Lex Fridman (17:14.480)

on improving the weakest part of the system.

Lex Fridman (17:16.360)

And as long as you focus on improving

Lex Fridman (17:19.160)

the weakest part of the system,

Sebastian Thrun (17:20.560)

you eventually build a really great system.

Lex Fridman (17:23.080)

Let me just pause on that, to me as an engineer,

Sebastian Thrun (17:25.880)

it's just super exciting that you were thinking like that,

Lex Fridman (17:28.280)

especially at that stage as brilliant,

Sebastian Thrun (17:30.440)

that testing was such a core part of it.

Lex Fridman (17:33.400)

It may be to linger on the point of leadership.

Sebastian Thrun (17:36.720)

I think it's one of the first times

Lex Fridman (17:39.120)

you were really a leader

Lex Fridman (17:41.960)

and you've led many very successful teams since then.

Lex Fridman (17:46.440)

What does it take to be a good leader?

Sebastian Thrun (17:48.480)

I would say most of all, I just take credit.

Lex Fridman (17:51.000)

I put the work of others, right?

Sebastian Thrun (17:55.320)

That's very convenient turns out

Lex Fridman (17:57.560)

because I can't do all these things myself.

Sebastian Thrun (18:00.200)

I'm an engineer at heart.

Lex Fridman (18:01.120)

So I care about engineering.

Lex Fridman (18:03.760)

So I don't know what the chicken and the egg is,

Lex Fridman (18:06.160)

but as a kid, I loved computers

Sebastian Thrun (18:07.880)

because you could tell them to do something

Lex Fridman (18:09.560)

and they actually did it.

Sebastian Thrun (18:10.720)

It was very cool.

Lex Fridman (18:11.560)

And you could like in the middle of the night,

Sebastian Thrun (18:12.760)

wake up at one in the morning and switch on your computer.

Lex Fridman (18:15.200)

And what he told you to yesterday, it would still do.

Sebastian Thrun (18:18.160)

That was really cool.

Lex Fridman (18:19.400)

Unfortunately, that didn't quite work with people.

Lex Fridman (18:21.320)

So you go to people and tell them what to do

Lex Fridman (18:22.880)

and they don't do it.

Lex Fridman (18:24.360)

And they hate you for it, or you do it today

Lex Fridman (18:26.960)

and then you go a day later and they stop doing it.

Lex Fridman (18:29.040)

So you have to...

Lex Fridman (18:30.240)

So then the question really became,

Lex Fridman (18:31.480)

how can you put yourself in the brain of people

Lex Fridman (18:34.120)

as opposed to computers?

Lex Fridman (18:35.120)

And in terms of computers, it's super dumb.

Lex Fridman (18:37.400)

That's so dumb.

Sebastian Thrun (18:38.240)

If people were as dumb as computers,

Lex Fridman (18:39.640)

I wouldn't want to work with them.

Lex Fridman (18:41.280)

But people are smart and people are emotional

Lex Fridman (18:43.640)

and people have pride and people have aspirations.

Lex Fridman (18:45.920)

So how can I connect to that?

Lex Fridman (18:49.840)

And that's the thing that most of our leadership just fails

Sebastian Thrun (18:52.560)

because many, many engineers turn manager

Lex Fridman (18:56.240)

believe they can treat their team just the same way

Sebastian Thrun (18:58.480)

it can treat your computer.

Lex Fridman (18:59.320)

And it just doesn't work this way.

Sebastian Thrun (19:00.440)

It's just really bad.

Lex Fridman (19:02.320)

So how can I connect to people?

Lex Fridman (19:05.080)

And it turns out as a college professor,

Lex Fridman (19:07.680)

the wonderful thing you do all the time

Sebastian Thrun (19:10.000)

is to empower other people.

Lex Fridman (19:11.000)

Like your job is to make your students look great.

Sebastian Thrun (19:14.720)

That's all you do.

Lex Fridman (19:15.560)

You're the best coach.

Lex Fridman (19:16.920)

And it turns out if you do a fantastic job with making

Lex Fridman (19:19.160)

your students look great, they actually love you

Lex Fridman (19:21.560)

and their parents love you.

Lex Fridman (19:22.720)

And they give you all the credit for stuff you don't deserve.

Sebastian Thrun (19:25.520)

All my students were smarter than me.

Lex Fridman (19:27.200)

All the great stuff invented at Stanford

Sebastian Thrun (19:28.720)

was their stuff, not my stuff.

Lex Fridman (19:30.040)

And they give me credit and say, oh, Sebastian.

Sebastian Thrun (19:32.480)

We're just making them feel good about themselves.

Lex Fridman (19:35.240)

So the question really is, can you take a team of people

Lex Fridman (19:38.040)

and what does it take to make them

Lex Fridman (19:40.400)

to connect to what they actually want in life

Lex Fridman (19:43.360)

and turn this into productive action?

Lex Fridman (19:45.760)

It turns out every human being that I know

Sebastian Thrun (19:48.520)

has incredibly good intentions.

Lex Fridman (19:50.120)

I've really rarely met a person with bad intentions.

Sebastian Thrun (19:54.120)

I believe every person wants to contribute.

Lex Fridman (19:55.920)

I think every person I've met wants to help others.

Sebastian Thrun (19:59.440)

It's amazing how much of an urge we have

Lex Fridman (1:00:02.880)

And I know many fellow robot assistant computer scientists

Sebastian Thrun (1:00:07.220)

that I will insist to take this course.

Lex Fridman (1:00:09.820)

Not to be named here.

Sebastian Thrun (1:00:12.180)

Not to be named.

Lex Fridman (1:00:13.740)

Many, many years ago, 1903,

Sebastian Thrun (1:00:17.940)

the Wright brothers flew in Kitty Hawk for the first time.

Lex Fridman (1:00:22.580)

And you've launched a company of the same name, Kitty Hawk,

Sebastian Thrun (1:00:26.940)

with the dream of building flying cars, eVTOLs.

Lex Fridman (1:00:32.300)

So at the big picture,

Lex Fridman (1:00:34.560)

what are the big challenges of making this thing

Lex Fridman (1:00:36.620)

that actually have inspired generations of people

Lex Fridman (1:00:39.980)

about what the future looks like?

Lex Fridman (1:00:41.740)

What does it take?

Lex Fridman (1:00:42.580)

What are the biggest challenges?

Lex Fridman (1:00:43.660)

So flying cars has always been a dream.

Sebastian Thrun (1:00:47.220)

Every boy, every girl wants to fly.

Lex Fridman (1:00:49.700)

Let's be honest.

Sebastian Thrun (1:00:50.540)

Yes.

Lex Fridman (1:00:51.360)

And let's go back in our history

Sebastian Thrun (1:00:52.340)

of your dreaming of flying.

Lex Fridman (1:00:53.760)

I think honestly, my single most remembered childhood dream

Sebastian Thrun (1:00:57.420)

has been a dream where I was sitting on a pillow

Lex Fridman (1:00:59.420)

and I could fly.

Sebastian Thrun (1:01:00.740)

I was like five years old.

Lex Fridman (1:01:02.020)

I remember like maybe three dreams of my childhood,

Lex Fridman (1:01:04.140)

but that's the one I remember most vividly.

Lex Fridman (1:01:07.540)

And then Peter Thiel famously said,

Sebastian Thrun (1:01:09.400)

they promised us flying cars

Lex Fridman (1:01:10.660)

and they gave us 140 characters pointing as Twitter

Sebastian Thrun (1:01:14.460)

at the time, limited message size to 140 characters.

Lex Fridman (1:01:18.380)

So if you're coming back now to really go

Sebastian Thrun (1:01:20.220)

for these super impactful stuff like flying cars

Lex Fridman (1:01:23.220)

and to be precise, they're not really cars.

Sebastian Thrun (1:01:25.900)

They don't have wheels.

Lex Fridman (1:01:27.140)

They're actually much closer to a helicopter

Sebastian Thrun (1:01:28.580)

than anything else.

Lex Fridman (1:01:29.640)

They take off vertically and they fly horizontally,

Lex Fridman (1:01:32.080)

but they have important differences.

Lex Fridman (1:01:34.380)

One difference is that they are much quieter.

Sebastian Thrun (1:01:37.740)

We just released a vehicle called Project Heaviside

Lex Fridman (1:01:41.580)

that can fly over you as low as a helicopter

Lex Fridman (1:01:43.500)

and you basically can't hear.

Lex Fridman (1:01:45.200)

It's like 38 decibels.

Sebastian Thrun (1:01:46.700)

It's like, if you were inside the library,

Lex Fridman (1:01:49.240)

you might be able to hear it,

Lex Fridman (1:01:50.220)

but anywhere outdoors, your ambient noise is higher.

Lex Fridman (1:01:53.540)

Secondly, they're much more affordable.

Sebastian Thrun (1:01:57.020)

They're much more affordable than helicopters.

Lex Fridman (1:01:58.980)

And the reason is helicopters are expensive

Sebastian Thrun (1:02:01.920)

for many reasons.

Lex Fridman (1:02:04.380)

There's lots of single point of figures in a helicopter.

Sebastian Thrun (1:02:06.980)

There's a bolt between the blades

Lex Fridman (1:02:09.140)

that's caused Jesus bolt.

Lex Fridman (1:02:10.780)

And the reason why it's called Jesus bolt

Lex Fridman (1:02:12.420)

is that if this bolt breaks, you will die.

Sebastian Thrun (1:02:16.380)

There is no second solution in helicopter flight.

Lex Fridman (1:02:19.500)

Whereas we have these distributed mechanism.

Sebastian Thrun (1:02:21.500)

When you go from gasoline to electric,

Lex Fridman (1:02:23.740)

you can now have many, many, many small motors

Sebastian Thrun (1:02:25.820)

as opposed to one big motor.

Lex Fridman (1:02:27.260)

And that means if you lose one of those motors,

Sebastian Thrun (1:02:28.780)

not a big deal.

Lex Fridman (1:02:29.620)

Heaviside, if it loses a motor, has eight of those.

Sebastian Thrun (1:02:32.820)

If it loses one of those eight motors,

Lex Fridman (1:02:34.020)

so it's seven left, it can take off just like before

Lex Fridman (1:02:37.260)

and land just like before.

Lex Fridman (1:02:40.100)

We are now also moving into a technology

Sebastian Thrun (1:02:42.020)

that doesn't require a commercial pilot

Lex Fridman (1:02:44.160)

because in some level,

Sebastian Thrun (1:02:45.500)

flight is actually easier than ground transportation

Lex Fridman (1:02:48.980)

like in self driving cars.

Sebastian Thrun (1:02:51.820)

The world is full of like children and bicycles

Lex Fridman (1:02:54.500)

and other cars and mailboxes and curbs and shrubs

Lex Fridman (1:02:57.580)

and what have you.

Lex Fridman (1:02:58.420)

All these things you have to avoid.

Sebastian Thrun (1:03:00.500)

When you go above the buildings and tree lines,

Lex Fridman (1:03:03.740)

there's nothing there.

Sebastian Thrun (1:03:04.620)

I mean, you can do the test right now,

Lex Fridman (1:03:06.100)

look outside and count the number of things you see flying.

Sebastian Thrun (1:03:09.420)

I'd be shocked if you could see more than two things.

Lex Fridman (1:03:11.500)

It's probably just zero.

Sebastian Thrun (1:03:13.860)

In the Bay Area, the most I've ever seen was six.

Lex Fridman (1:03:16.940)

And maybe it's 15 or 20,

Lex Fridman (1:03:18.820)

but not 10,000.

Lex Fridman (1:03:20.400)

So the sky is very ample and very empty and very free.

Lex Fridman (1:03:24.000)

So the vision is, can we build a socially acceptable

Lex Fridman (1:03:27.820)

mass transit solution for daily transportation

Lex Fridman (1:03:32.360)

that is affordable?

Lex Fridman (1:03:34.280)

And we have an existence proof.

Sebastian Thrun (1:03:36.340)

Heaviside can fly 100 miles in range

Lex Fridman (1:03:39.780)

with still 30% electric reserves.

Sebastian Thrun (1:03:43.260)

It can fly up to like 180 miles an hour.

Lex Fridman (1:03:46.060)

We know that that solution at scale

Sebastian Thrun (1:03:48.900)

would make your ground transportation

Lex Fridman (1:03:51.420)

10 times as fast as a car

Sebastian Thrun (1:03:53.820)

based on use census or statistics data,

Lex Fridman (1:03:57.580)

which means you would take your 300 hours of daily,

Sebastian Thrun (1:04:00.900)

of yearly commute down to 30 hours

Lex Fridman (1:04:03.020)

and give you 270 hours back.

Lex Fridman (1:04:05.180)

Who wouldn't want, I mean, who doesn't hate traffic?

Lex Fridman (1:04:07.700)

Like I hate, give me the person that doesn't hate traffic.

Sebastian Thrun (1:04:10.820)

I hate traffic.

Lex Fridman (1:04:11.660)

Every time I'm in traffic, I hate it.

Lex Fridman (1:04:13.900)

And if we could free the world from traffic,

Lex Fridman (1:04:17.580)

we have technology.

Sebastian Thrun (1:04:18.460)

We can free the world from traffic.

Lex Fridman (1:04:20.060)

We have the technology.

Sebastian Thrun (1:04:21.340)

It's there.

Lex Fridman (1:04:22.180)

We have an existence proof.

Sebastian Thrun (1:04:23.060)

It's not a technological problem anymore.

Lex Fridman (1:04:25.440)

Do you think there is a future where tens of thousands,

Sebastian Thrun (1:04:29.340)

maybe hundreds of thousands of both delivery drones

Lex Fridman (1:04:34.380)

and flying cars of this kind, EV talls fill the sky?

Sebastian Thrun (1:04:39.940)

I absolutely believe this.

Lex Fridman (1:04:40.940)

And there's obviously the societal acceptance

Sebastian Thrun (1:04:43.860)

is a major question.

Lex Fridman (1:04:45.460)

And of course, safety is.

Sebastian Thrun (1:04:46.940)

I believe in safety,

Lex Fridman (1:04:48.060)

we're gonna exceed ground transportation safety

Sebastian Thrun (1:04:50.340)

as has happened for aviation already, commercial aviation.

Lex Fridman (1:04:54.500)

And in terms of acceptance,

Sebastian Thrun (1:04:56.640)

I think one of the key things is noise.

Lex Fridman (1:04:58.320)

That's why we are focusing relentlessly on noise

Lex Fridman (1:05:00.980)

and we build perhaps the quietest electric vehicle

Lex Fridman (1:05:05.660)

ever built.

Sebastian Thrun (1:05:07.640)

The nice thing about the sky is it's three dimensional.

Lex Fridman (1:05:09.760)

So any mathematician will immediately recognize

Sebastian Thrun (1:05:12.520)

the difference between 1D of like a regular highway

Lex Fridman (1:05:14.980)

to 3D of a sky.

Lex Fridman (1:05:17.320)

But to make it clear for the layman,

Lex Fridman (1:05:20.220)

say you wanna make 100 vertical lanes

Sebastian Thrun (1:05:22.740)

of highway 101 in San Francisco,

Lex Fridman (1:05:25.040)

because you believe building 100 vertical lanes

Sebastian Thrun (1:05:27.220)

is the right solution.

Lex Fridman (1:05:28.900)

Imagine how much it would cost to stack 100 vertical lanes

Sebastian Thrun (1:05:31.780)

physically onto 101.

Lex Fridman (1:05:33.420)

That would be prohibitive.

Sebastian Thrun (1:05:34.340)

That would be consuming the world's GDP for an entire year

Lex Fridman (1:05:37.780)

just for one highway.

Sebastian Thrun (1:05:39.260)

It's amazingly expensive.

Lex Fridman (1:05:41.300)

In the sky, it would just be a recompilation

Sebastian Thrun (1:05:43.740)

of a piece of software because all these lanes are virtual.

Lex Fridman (1:05:46.580)

That means any vehicle that is in conflict

Sebastian Thrun (1:05:49.260)

with another vehicle would just go to different altitudes

Lex Fridman (1:05:51.860)

and then the conflict is gone.

Lex Fridman (1:05:53.340)

And if you don't believe this,

Lex Fridman (1:05:55.380)

that's exactly how commercial aviation works.

Sebastian Thrun (1:05:58.580)

When you fly from New York to San Francisco,

Lex Fridman (1:06:01.460)

another plane flies from San Francisco to New York,

Sebastian Thrun (1:06:04.240)

they are different altitudes.

Lex Fridman (1:06:05.300)

So they don't hit each other.

Sebastian Thrun (1:06:06.740)

It's a solved problem for the jet space

Lex Fridman (1:06:10.420)

and it will be a solved problem for the urban space.

Sebastian Thrun (1:06:12.780)

There's companies like Google Wing and Amazon

Lex Fridman (1:06:15.380)

working on very innovative solutions.

Lex Fridman (1:06:17.060)

How do we have space management?

Lex Fridman (1:06:18.580)

They use exactly the same principles as we use today

Sebastian Thrun (1:06:21.660)

to route today's jets.

Lex Fridman (1:06:23.300)

There's nothing hard about this.

Lex Fridman (1:06:25.940)

Do you envision autonomy being a key part of it

Lex Fridman (1:06:29.040)

so that the flying vehicles are either semi autonomous

Lex Fridman (1:06:34.040)

semi autonomous or fully autonomous?

Lex Fridman (1:06:36.920)

100% autonomous.

Sebastian Thrun (1:06:37.880)

You don't want idiots like me flying in the sky,

Lex Fridman (1:06:40.480)

I promise you.

Lex Fridman (1:06:41.960)

And if you have 10,000,

Lex Fridman (1:06:44.280)

watch the movie, The Fifth Element

Sebastian Thrun (1:06:46.040)

to get a feel for what will happen if it's not autonomous.

Lex Fridman (1:06:49.480)

And a centralized, that's a really interesting idea

Sebastian Thrun (1:06:51.720)

of a centralized sort of management system

Lex Fridman (1:06:55.240)

for lanes and so on.

Lex Fridman (1:06:56.320)

So actually just being able to have

Lex Fridman (1:07:00.280)

similar as we have in the current commercial aviation,

Lex Fridman (1:07:03.000)

but scale it up to much, much more vehicles.

Lex Fridman (1:07:05.560)

That's a really interesting optimization problem.

Sebastian Thrun (1:07:07.660)

It is very mathematically, very, very straightforward.

Lex Fridman (1:07:11.080)

Like the gap we leave between jets is gargantuous.

Lex Fridman (1:07:13.520)

And part of the reason is there isn't that many jets.

Lex Fridman (1:07:16.400)

So it just feels like a good solution.

Sebastian Thrun (1:07:18.800)

Today, when you get vectored by air traffic control,

Lex Fridman (1:07:22.380)

someone talks to you, right?

Lex Fridman (1:07:23.900)

So any ATC controller might have up to maybe 20 planes

Lex Fridman (1:07:26.960)

on the same frequency.

Lex Fridman (1:07:28.160)

And then they talk to you, you have to talk back.

Lex Fridman (1:07:30.360)

And it feels right because there isn't more than 20 planes

Sebastian Thrun (1:07:32.720)

around anyhow, so you can talk to everybody.

Lex Fridman (1:07:34.960)

But if there's 20,000 things around,

Sebastian Thrun (1:07:36.760)

you can't talk to everybody anymore.

Lex Fridman (1:07:37.980)

So we have to do something that's called digital,

Sebastian Thrun (1:07:40.260)

like text messaging.

Lex Fridman (1:07:41.520)

Like we do have solutions.

Sebastian Thrun (1:07:43.040)

Like we have what, four or five billion smartphones

Lex Fridman (1:07:45.560)

in the world now, right?

Lex Fridman (1:07:46.440)

And they're all connected.

Lex Fridman (1:07:47.720)

And somehow we solve the scale problem for smartphones.

Sebastian Thrun (1:07:50.720)

We know where they all are.

Lex Fridman (1:07:51.960)

They can talk to somebody and they're very reliable.

Sebastian Thrun (1:07:54.880)

They're amazingly reliable.

Lex Fridman (1:07:56.460)

We could use the same system,

Sebastian Thrun (1:07:58.640)

the same scale for air traffic control.

Lex Fridman (1:08:01.080)

So instead of me as a pilot talking to a human being

Lex Fridman (1:08:04.080)

and in the middle of the conversation

Lex Fridman (1:08:06.280)

receiving a new frequency, like how ancient is that?

Sebastian Thrun (1:08:09.660)

We could digitize this stuff

Lex Fridman (1:08:11.240)

and digitally transmit the right flight coordinates.

Lex Fridman (1:08:15.240)

And that solution will automatically scale

Lex Fridman (1:08:18.060)

to 10,000 vehicles.

Sebastian Thrun (1:08:20.040)

We talked about empathy a little bit.

Lex Fridman (1:08:22.200)

Do you think we will one day build an AI system

Sebastian Thrun (1:08:25.800)

that a human being can love

Lex Fridman (1:08:27.580)

and that loves that human back, like in the movie, Her?

Sebastian Thrun (1:08:31.320)

Look, I'm a pragmatist.

Lex Fridman (1:08:33.960)

For me, AI is a tool.

Sebastian Thrun (1:08:35.600)

It's like a shovel.

Lex Fridman (1:08:36.920)

And the ethics of using the shovel are always

Sebastian Thrun (1:08:40.800)

with us, the people.

Lex Fridman (1:08:41.840)

And it has to be this way.

Sebastian Thrun (1:08:44.040)

In terms of emotions,

Lex Fridman (1:08:47.160)

I would hate to come into my kitchen

Lex Fridman (1:08:49.800)

and see that my refrigerator spoiled all my food,

Lex Fridman (1:08:54.200)

then have it explained to me

Sebastian Thrun (1:08:55.280)

that it fell in love with the dishwasher

Lex Fridman (1:08:57.960)

and it wasn't as nice as the dishwasher.

Lex Fridman (1:08:59.680)

So as a result, it neglected me.

Lex Fridman (1:09:02.160)

That would just be a bad experience

Lex Fridman (1:09:05.120)

and it would be a bad product.

Lex Fridman (1:09:07.040)

I would probably not recommend this refrigerator

Sebastian Thrun (1:09:09.520)

to my friends.

Lex Fridman (1:09:11.720)

And that's where I draw the line.

Sebastian Thrun (1:09:12.880)

I think to me, technology has to be reliable

Lex Fridman (1:09:16.600)

and has to be predictable.

Sebastian Thrun (1:09:17.680)

I want my car to work.

Lex Fridman (1:09:19.840)

I don't want to fall in love with my car.

Sebastian Thrun (1:09:22.840)

I just want it to work.

Lex Fridman (1:09:24.560)

I want it to compliment me, not to replace me.

Sebastian Thrun (1:09:27.160)

I have very unique human properties

Lex Fridman (1:09:30.640)

and I want the machines to make me,

Sebastian Thrun (1:09:33.420)

turn me into a superhuman.

Lex Fridman (1:09:35.680)

Like I'm already a superhuman today,

Sebastian Thrun (1:09:37.800)

thanks to the machines that surround me.

Lex Fridman (1:09:39.280)

And I give you examples.

Sebastian Thrun (1:09:40.780)

I can run across the Atlantic

Lex Fridman (1:09:44.160)

at near the speed of sound at 36,000 feet today.

Sebastian Thrun (1:09:48.480)

That's kind of amazing.

Lex Fridman (1:09:49.560)

I can, my voice now carries me all the way to Australia

Sebastian Thrun (1:09:54.640)

using a smartphone today.

Lex Fridman (1:09:56.600)

And it's not the speed of sound, which would take hours.

Sebastian Thrun (1:10:00.060)

It's the speed of light.

Lex Fridman (1:10:01.300)

My voice travels at the speed of light.

Lex Fridman (1:10:03.820)

How cool is that?

Lex Fridman (1:10:04.660)

That makes me superhuman.

Sebastian Thrun (1:10:06.320)

I would even argue my flushing toilet makes me superhuman.

Lex Fridman (1:10:10.520)

Just think of the time before flushing toilets.

Lex Fridman (1:10:13.800)

And maybe you have a very old person in your family

Lex Fridman (1:10:16.460)

that you can ask about this

Sebastian Thrun (1:10:18.480)

or take a trip to rural India to experience it.

Lex Fridman (1:10:23.400)

It makes me superhuman.

Lex Fridman (1:10:25.840)

So to me, what technology does, it compliments me.

Lex Fridman (1:10:28.900)

It makes me stronger.

Sebastian Thrun (1:10:30.920)

Therefore, words like love and compassion

Lex Fridman (1:10:33.520)

have very little interest in this for machines.

Sebastian Thrun (1:10:38.640)

I have interest in people.

Lex Fridman (1:10:40.720)

You don't think, first of all, beautifully put,

Sebastian Thrun (1:10:44.280)

beautifully argued,

Lex Fridman (1:10:45.680)

but do you think love has use in our tools?

Sebastian Thrun (1:10:49.520)

Compassion.

Lex Fridman (1:10:50.440)

I think love is a beautiful human concept.

Lex Fridman (1:10:53.280)

And if you think of what love really is,

Lex Fridman (1:10:55.420)

love is a means to convey safety, to convey trust.

Sebastian Thrun (1:11:03.240)

I think trust has a huge need in technology as well,

Lex Fridman (1:11:07.440)

not just people.

Sebastian Thrun (1:11:09.160)

We want to trust our technology the same way,

Lex Fridman (1:11:12.600)

in a similar way we trust people.

Sebastian Thrun (1:11:15.960)

In human interaction, standards have emerged

Lex Fridman (1:11:19.360)

and feelings, emotions have emerged,

Sebastian Thrun (1:11:21.760)

maybe genetically, maybe biologically,

Lex Fridman (1:11:23.920)

that are able to convey sense of trust, sense of safety,

Sebastian Thrun (1:11:26.560)

sense of passion, of love, of dedication

Lex Fridman (1:11:28.880)

that makes the human fabric.

Lex Fridman (1:11:30.800)

And I'm a big slacker for love.

Lex Fridman (1:11:33.740)

I want to be loved.

Sebastian Thrun (1:11:34.600)

I want to be trusted.

Lex Fridman (1:11:35.440)

I want to be admired.

Sebastian Thrun (1:11:36.880)

All these wonderful things.

Lex Fridman (1:11:38.880)

And because all of us, we have this beautiful system,

Sebastian Thrun (1:11:42.200)

I wouldn't just blindly copy this to the machines.

Lex Fridman (1:11:44.840)

Here's why.

Sebastian Thrun (1:11:46.200)

When you look at, say, transportation,

Lex Fridman (1:11:49.360)

you could have observed that up to the end

Sebastian Thrun (1:11:53.320)

of the 19th century, almost all transportation used

Lex Fridman (1:11:57.120)

any number of legs, from one leg to two legs

Sebastian Thrun (1:11:59.820)

to a thousand legs.

Lex Fridman (1:12:01.720)

And you could have concluded that is the right way

Sebastian Thrun (1:12:03.840)

to move about the environment.

Lex Fridman (1:12:06.800)

We've been made the exception of birds

Sebastian Thrun (1:12:08.080)

who use flapping wings.

Lex Fridman (1:12:08.960)

In fact, there are many people in aviation

Sebastian Thrun (1:12:10.880)

that flap wings to their arms and jump from cliffs.

Lex Fridman (1:12:13.720)

Most of them didn't survive.

Sebastian Thrun (1:12:16.920)

Then the interesting thing is that the technology solutions

Lex Fridman (1:12:19.880)

are very different.

Sebastian Thrun (1:12:21.600)

Like in technology, it's really easy to build a wheel.

Lex Fridman (1:12:23.880)

In biology, it's super hard to build a wheel.

Sebastian Thrun (1:12:25.680)

There's very few perpetually rotating things in biology

Lex Fridman (1:12:30.080)

and they usually run cells and things.

Sebastian Thrun (1:12:34.180)

In engineering, we can build wheels.

Lex Fridman (1:12:37.200)

And those wheels gave rise to cars.

Sebastian Thrun (1:12:41.020)

Similar wheels gave rise to aviation.

Lex Fridman (1:12:44.360)

Like there's no thing that flies

Sebastian Thrun (1:12:46.680)

that wouldn't have something that rotates,

Lex Fridman (1:12:48.840)

like a jet engine or helicopter blades.

Lex Fridman (1:12:52.400)

So the solutions have used very different physical laws

Lex Fridman (1:12:55.520)

than nature, and that's great.

Lex Fridman (1:12:58.040)

So for me to be too much focused on,

Lex Fridman (1:13:00.080)

oh, this is how nature does it, let's just replicate it.

Sebastian Thrun (1:13:03.340)

If you really believed that the solution

Lex Fridman (1:13:05.400)

to the agricultural evolution was a humanoid robot,

Sebastian Thrun (1:13:08.720)

you would still be waiting today.

Lex Fridman (1:13:10.920)

Again, beautifully put.

Sebastian Thrun (1:13:12.520)

You said that you don't take yourself too seriously.

Lex Fridman (1:13:15.920)

Did I say that?

Lex Fridman (1:13:18.160)

You want me to say that?

Lex Fridman (1:13:19.160)

Maybe.

Sebastian Thrun (1:13:20.000)

You're not taking me seriously.

Lex Fridman (1:13:20.960)

I'm not, that's right.

Sebastian Thrun (1:13:22.880)

Good, you're right, I don't wanna.

Lex Fridman (1:13:24.480)

I just made that up.

Lex Fridman (1:13:25.720)

But you have a humor and a lightness about life

Lex Fridman (1:13:29.120)

that I think is beautiful and inspiring to a lot of people.

Lex Fridman (1:13:33.520)

Where does that come from?

Lex Fridman (1:13:35.040)

The smile, the humor, the lightness

Sebastian Thrun (1:13:38.400)

amidst all the chaos of the hard work that you're in,

Lex Fridman (1:13:42.600)

where does that come from?

Sebastian Thrun (1:13:43.640)

I just love my life.

Lex Fridman (1:13:44.560)

I love the people around me.

Sebastian Thrun (1:13:47.520)

I'm just so glad to be alive.

Lex Fridman (1:13:49.740)

Like I'm, what, 52, hard to believe.

Sebastian Thrun (1:13:53.640)

People say 52 is a new 51, so now I feel better.

Lex Fridman (1:13:56.260)

But in looking around the world,

Sebastian Thrun (1:14:01.260)

looking around the world, just go back 200, 300 years.

Lex Fridman (1:14:06.180)

Humanity is, what, 300,000 years old?

Lex Fridman (1:14:09.360)

But for the first 300,000 years minus the last 100,

Lex Fridman (1:14:13.980)

our life expectancy would have been

Sebastian Thrun (1:14:17.060)

plus or minus 30 years roughly, give or take.

Lex Fridman (1:14:20.260)

So I would be long dead now.

Sebastian Thrun (1:14:24.360)

That makes me just enjoy every single day of my life

Lex Fridman (1:14:26.840)

because I don't deserve this.

Lex Fridman (1:14:28.260)

Why am I born today when so many of my ancestors

Lex Fridman (1:14:32.460)

died of horrible deaths, like famines, massive wars

Sebastian Thrun (1:14:38.820)

that ravaged Europe for the last 1,000 years

Lex Fridman (1:14:41.860)

mystically disappeared after World War II

Sebastian Thrun (1:14:44.520)

when the Americans and the Allies

Lex Fridman (1:14:46.540)

did something amazing to my country

Sebastian Thrun (1:14:48.300)

that didn't deserve it, the country of Germany.

Lex Fridman (1:14:51.460)

This is so amazing.

Lex Fridman (1:14:52.620)

And then when you're alive and feel this every day,

Lex Fridman (1:14:56.960)

then it's just so amazing what we can accomplish,

Lex Fridman (1:15:02.020)

what we can do.

Lex Fridman (1:15:03.500)

We live in a world that is so incredibly,

Sebastian Thrun (1:15:06.380)

vastly changing every day.

Lex Fridman (1:15:08.720)

Almost everything that we cherish from your smartphone

Sebastian Thrun (1:15:12.900)

to your flushing toilet, to all these basic inventions,

Lex Fridman (1:15:16.220)

your new clothes you're wearing, your watch, your plane,

Sebastian Thrun (1:15:19.620)

penicillin, I don't know, anesthesia for surgery,

Lex Fridman (1:15:24.620)

penicillin have been invented in the last 150 years.

Lex Fridman (1:15:29.060)

So in the last 150 years, something magical happened.

Lex Fridman (1:15:31.420)

And I would trace it back to Gutenberg

Lex Fridman (1:15:33.380)

and the printing press that has been able

Lex Fridman (1:15:34.980)

to disseminate information more efficiently than before

Sebastian Thrun (1:15:37.860)

that all of a sudden we were able to invent agriculture

Lex Fridman (1:15:41.860)

and nitrogen fertilization that made agriculture

Lex Fridman (1:15:44.940)

so much more potent that we didn't have to work

Lex Fridman (1:15:47.100)

in the farms anymore and we could start reading and writing

Lex Fridman (1:15:49.180)

and we could become all these wonderful things

Lex Fridman (1:15:51.340)

we are today, from airline pilot to massage therapist

Sebastian Thrun (1:15:53.860)

to software engineer.

Lex Fridman (1:15:56.300)

It's just amazing.

Sebastian Thrun (1:15:57.140)

Like living in that time is such a blessing.

Lex Fridman (1:16:00.180)

We should sometimes really think about this, right?

Sebastian Thrun (1:16:03.940)

Steven Pinker, who is a very famous author and philosopher

Lex Fridman (1:16:06.860)

whom I really adore, wrote a great book called

Sebastian Thrun (1:16:08.980)

Enlightenment Now.

Lex Fridman (1:16:09.820)

And that's maybe the one book I would recommend.

Lex Fridman (1:16:11.420)

And he asks the question,

Lex Fridman (1:16:13.020)

if there was only a single article written

Lex Fridman (1:16:15.180)

in the 20th century, it's only one article, what would it be?

Lex Fridman (1:16:18.580)

What's the most important innovation,

Lex Fridman (1:16:20.620)

the most important thing that happened?

Lex Fridman (1:16:22.580)

And he would say this article would credit

Sebastian Thrun (1:16:24.700)

a guy named Karl Bosch.

Lex Fridman (1:16:27.020)

And I challenge anybody, have you ever heard

Lex Fridman (1:16:29.460)

of the name Karl Foch?

Lex Fridman (1:16:31.180)

I hadn't, okay.

Sebastian Thrun (1:16:32.940)

There's a Bosch Corporation in Germany,

Lex Fridman (1:16:35.420)

but it's not associated with Karl Bosch.

Lex Fridman (1:16:38.420)

So I looked it up.

Lex Fridman (1:16:39.860)

Karl Bosch invented nitrogen fertilization.

Lex Fridman (1:16:42.660)

And in doing so, together with an older invention

Lex Fridman (1:16:45.580)

of irrigation, was able to increase the yields

Sebastian Thrun (1:16:49.220)

per agricultural land by a factor of 26.

Lex Fridman (1:16:52.860)

So a 2,500% increase in fertility of land.

Lex Fridman (1:16:57.700)

And that, so Steve Pinker argues,

Lex Fridman (1:17:00.540)

saved over 2 billion lives today.

Sebastian Thrun (1:17:03.900)

2 billion people who would be dead

Lex Fridman (1:17:05.700)

if this man hadn't done what he had done, okay?

Sebastian Thrun (1:17:08.420)

Think about that impact and what that means to society.

Lex Fridman (1:17:12.180)

That's the way I look at the world.

Sebastian Thrun (1:17:14.180)

I mean, it's so amazing to be alive and to be part of this.

Lex Fridman (1:17:16.940)

And I'm so glad I lived after Karl Bosch and not before.

Sebastian Thrun (1:17:21.300)

I don't think there's a better way to end this, Sebastian.

Lex Fridman (1:17:23.980)

It's an honor to talk to you,

Sebastian Thrun (1:17:25.460)

to have had the chance to learn from you.

Lex Fridman (1:17:27.340)

Thank you so much for talking to me.

Sebastian Thrun (1:17:28.300)

Thanks for coming out.

Lex Fridman (1:17:29.140)

It's been a real pleasure.

Sebastian Thrun (1:17:30.980)

Thank you for listening to this conversation

Lex Fridman (1:17:32.780)

with Sebastian Thrun.

Lex Fridman (1:17:34.380)

And thank you to our presenting sponsor, Cash App.

Lex Fridman (1:17:37.460)

Download it, use code LexPodcast,

Sebastian Thrun (1:17:40.220)

you'll get $10 and $10 will go to FIRST,

Lex Fridman (1:17:43.220)

a STEM education nonprofit that inspires

Sebastian Thrun (1:17:45.500)

hundreds of thousands of young minds

Lex Fridman (1:17:47.460)

to learn and to dream of engineering our future.

Sebastian Thrun (1:17:50.540)

If you enjoy this podcast, subscribe on YouTube,

Lex Fridman (1:17:53.340)

get five stars on Apple Podcast, support it on Patreon,

Sebastian Thrun (1:17:56.620)

or connect with me on Twitter.

Lex Fridman (1:17:58.860)

And now, let me leave you with some words of wisdom

Sebastian Thrun (1:18:01.260)

from Sebastian Thrun.

Lex Fridman (1:18:03.260)

It's important to celebrate your failures

Sebastian Thrun (1:18:05.420)

as much as your successes.

Lex Fridman (1:18:07.700)

If you celebrate your failures really well,

Sebastian Thrun (1:18:09.780)

if you say, wow, I failed, I tried, I was wrong,

Lex Fridman (1:18:13.900)

but I learned something, then you realize you have no fear.

Lex Fridman (1:18:18.260)

And when your fear goes away, you can move the world.

Lex Fridman (1:18:22.460)

Thank you for listening and hope to see you next time.

Sebastian Thrun (20:01.840)

not to just help ourselves, but to help others.

Lex Fridman (20:04.440)

So how can we empower people and give them

Lex Fridman (20:06.480)

the right framework that they can accomplish this?

Lex Fridman (20:10.600)

In moments when it works, it's magical.

Sebastian Thrun (20:12.400)

Because you'd see the confluence of people

Lex Fridman (20:17.160)

being able to make the world a better place

Lex Fridman (20:19.160)

and deriving enormous confidence and pride out of this.

Lex Fridman (20:22.840)

And that's when my environment works the best.

Sebastian Thrun (20:27.160)

These are moments where I can disappear for a month

Lex Fridman (20:29.400)

and come back and things still work.

Sebastian Thrun (20:31.560)

It's very hard to accomplish.

Lex Fridman (20:32.760)

But when it works, it's amazing.

Lex Fridman (20:35.040)

So I agree with you very much.

Lex Fridman (20:37.240)

It's not often heard that most people in the world

Sebastian Thrun (20:42.000)

have good intentions.

Lex Fridman (20:43.520)

At the core, their intentions are good

Lex Fridman (20:45.920)

and they're good people.

Lex Fridman (20:47.400)

That's a beautiful message, it's not often heard.

Sebastian Thrun (20:50.160)

We make this mistake, and this is a friend of mine,

Lex Fridman (20:52.600)

Alex Werder, talking to us, that we judge ourselves

Sebastian Thrun (20:56.400)

by our intentions and others by their actions.

Lex Fridman (21:00.080)

And I think that the biggest skill,

Sebastian Thrun (21:01.880)

I mean, here in Silicon Valley, we follow engineers

Lex Fridman (21:03.560)

who have very little empathy and are kind of befuddled

Sebastian Thrun (21:06.640)

by why it doesn't work for them.

Lex Fridman (21:09.200)

The biggest skill, I think, that people should acquire

Sebastian Thrun (21:13.080)

is to put themselves into the position of the other

Lex Fridman (21:16.880)

and listen, and listen to what the other has to say.

Lex Fridman (21:20.000)

And they'd be shocked how similar they are to themselves.

Lex Fridman (21:23.400)

And they might even be shocked how their own actions

Sebastian Thrun (21:26.160)

don't reflect their intentions.

Lex Fridman (21:28.320)

I often have conversations with engineers

Sebastian Thrun (21:30.920)

where I say, look, hey, I love you, you're doing a great job.

Lex Fridman (21:33.400)

And by the way, what you just did has the following effect.

Lex Fridman (21:37.320)

Are you aware of that?

Lex Fridman (21:38.840)

And then people would say, oh my God, not I wasn't,

Sebastian Thrun (21:41.280)

because my intention was that.

Lex Fridman (21:43.120)

And I say, yeah, I trust your intention.

Sebastian Thrun (21:45.000)

You're a good human being.

Lex Fridman (21:46.360)

But just to help you in the future,

Sebastian Thrun (21:48.480)

if you keep expressing it that way,

Lex Fridman (21:51.320)

then people will just hate you.

Lex Fridman (21:53.400)

And I've had many instances where people say,

Lex Fridman (21:55.240)

oh my God, thank you for telling me this,

Sebastian Thrun (21:56.600)

because it wasn't my intention to look like an idiot.

Lex Fridman (21:59.280)

It wasn't my intention to help other people.

Sebastian Thrun (22:00.720)

I just didn't know how to do it.

Lex Fridman (22:02.480)

Very simple, by the way.

Sebastian Thrun (22:04.000)

There's a book, Dale Carnegie, 1936,

Lex Fridman (22:07.440)

how to make friends and how to influence others.

Sebastian Thrun (22:10.400)

Has the entire Bible, you just read it and you're done

Lex Fridman (22:12.720)

and you apply it every day.

Lex Fridman (22:13.960)

And I wish I was good enough to apply it every day.

Lex Fridman (22:16.760)

But it's just simple things, right?

Sebastian Thrun (22:18.880)

Like be positive, remember people's name, smile,

Lex Fridman (22:22.600)

and eventually have empathy.

Sebastian Thrun (22:24.480)

Really think that the person that you hate

Lex Fridman (22:27.400)

and you think is an idiot,

Sebastian Thrun (22:28.640)

is actually just like yourself.

Lex Fridman (22:30.440)

It's a person who's struggling, who means well,

Lex Fridman (22:33.200)

and who might need help, and guess what, you need help.

Lex Fridman (22:36.560)

I've recently spoken with Stephen Schwarzman.

Sebastian Thrun (22:39.960)

I'm not sure if you know who that is, but.

Lex Fridman (22:41.960)

I do.

Sebastian Thrun (22:42.920)

So, and he said.

Lex Fridman (22:44.320)

It's on my list.

Sebastian Thrun (22:45.160)

On the list.

Lex Fridman (22:47.440)

But he said, sort of to expand on what you're saying,

Sebastian Thrun (22:52.760)

that one of the biggest things you can do

Lex Fridman (22:56.040)

is hear people when they tell you what their problem is

Lex Fridman (23:00.040)

and then help them with that problem.

Lex Fridman (23:02.360)

He says, it's surprising how few people

Sebastian Thrun (23:06.000)

actually listen to what troubles others.

Lex Fridman (23:09.280)

And because it's right there in front of you

Lex Fridman (23:12.600)

and you can benefit the world the most.

Lex Fridman (23:15.240)

And in fact, yourself and everybody around you

Sebastian Thrun (23:18.040)

by just hearing the problems and solving them.

Lex Fridman (23:20.840)

I mean, that's my little history of engineering.

Sebastian Thrun (23:23.960)

That is, while I was engineering with computers,

Lex Fridman (23:28.240)

I didn't care at all what the computer's problems were.

Sebastian Thrun (23:32.400)

I just told them what to do and to do it.

Lex Fridman (23:34.800)

And it just doesn't work this way with people.

Sebastian Thrun (23:37.600)

It doesn't work with me.

Lex Fridman (23:38.480)

If you come to me and say, do A, I do the opposite.

Lex Fridman (23:43.600)

But let's return to the comfortable world of engineering.

Lex Fridman (23:47.160)

And can you tell me in broad strokes in how you see it?

Sebastian Thrun (23:52.160)

Because you're the core of starting it,

Lex Fridman (23:53.840)

the core of driving it,

Sebastian Thrun (23:55.120)

the technical evolution of autonomous vehicles

Lex Fridman (23:58.040)

from the first DARPA Grand Challenge

Sebastian Thrun (24:00.440)

to the incredible success we see with the program

Lex Fridman (24:03.640)

you started with Google self driving car

Lex Fridman (24:05.400)

and Waymo and the entire industry that sprung up

Lex Fridman (24:08.360)

of different kinds of approaches, debates and so on.

Sebastian Thrun (24:11.200)

Well, the idea of self driving car goes back to the 80s.

Lex Fridman (24:14.160)

There was a team in Germany and another team

Sebastian Thrun (24:15.480)

at Carnegie Mellon that did some very pioneering work.

Lex Fridman (24:18.720)

But back in the day, I'd say the computers were so deficient

Sebastian Thrun (24:21.760)

that even the best professors and engineers in the world

Lex Fridman (24:25.880)

basically stood no chance.

Sebastian Thrun (24:28.200)

It then folded into a phase where the US government

Lex Fridman (24:31.200)

spent at least half a billion dollars

Sebastian Thrun (24:33.320)

that I could count on research projects.

Lex Fridman (24:36.160)

But the way the procurement works,

Sebastian Thrun (24:38.920)

a successful stack of paper describing lots of stuff

Lex Fridman (24:42.800)

that no one's ever gonna read

Sebastian Thrun (24:43.880)

was a successful product of a research project.

Lex Fridman (24:47.640)

So we trained our researchers to produce lots of paper.

Sebastian Thrun (24:52.600)

That all changed with the DARPA Grand Challenge.

Lex Fridman (24:54.320)

And I really gotta credit the ingenious people at DARPA

Lex Fridman (24:58.480)

and the US government and Congress

Lex Fridman (25:00.400)

that took a complete new funding model where they said,

Sebastian Thrun (25:03.000)

let's not fund effort, let's fund outcomes.

Lex Fridman (25:05.640)

And it sounds very trivial,

Lex Fridman (25:06.840)

but there was no tax code that allowed

Lex Fridman (25:09.840)

the use of congressional tax money for a price.

Sebastian Thrun (25:13.720)

It was all effort based.

Lex Fridman (25:15.120)

So if you put in a hundred hours in,

Sebastian Thrun (25:16.320)

you could charge a hundred hours.

Lex Fridman (25:17.480)

If you put in a thousand hours in,

Sebastian Thrun (25:18.520)

you could build a thousand hours.

Lex Fridman (25:20.720)

By changing the focus instead of making the price,

Sebastian Thrun (25:22.880)

we don't pay you for development,

Lex Fridman (25:24.040)

we pay for the accomplishment.

Sebastian Thrun (25:26.360)

They drew in, they automatically drew out

Lex Fridman (25:28.960)

all these contractors who are used to the drug

Sebastian Thrun (25:31.720)

of getting money per hour.

Lex Fridman (25:33.400)

And they drew in a whole bunch of new people.

Lex Fridman (25:35.520)

And these people are mostly crazy people.

Lex Fridman (25:37.600)

They were people who had a car and a computer

Lex Fridman (25:40.680)

and they wanted to make a million bucks.

Lex Fridman (25:42.440)

The million bucks was their visual price money,

Sebastian Thrun (25:43.920)

it was then doubled.

Lex Fridman (25:45.440)

And they felt if I put my computer in my car

Lex Fridman (25:48.040)

and program it, I can be rich.

Lex Fridman (25:50.880)

And that was so awesome.

Sebastian Thrun (25:52.080)

Like half the teams, there was a team that was surfer dudes

Lex Fridman (25:55.480)

and they had like two surfboards on their vehicle

Lex Fridman (25:58.520)

and brought like these fashion girls, super cute girls,

Lex Fridman (26:01.560)

like twin sisters.

Lex Fridman (26:03.720)

And you could tell these guys were not your common

Lex Fridman (26:06.400)

beltway bandit who gets all these big multimillion

Lex Fridman (26:10.840)

and billion dollar countries from the US government.

Lex Fridman (26:13.520)

And there was a great reset.

Sebastian Thrun (26:16.280)

Universities moved in.

Lex Fridman (26:18.560)

I was very fortunate at Stanford that I just received tenure

Lex Fridman (26:21.800)

so I couldn't get fired no matter what I do,

Lex Fridman (26:23.360)

otherwise I wouldn't have done it.

Lex Fridman (26:25.120)

And I had enough money to finance this thing

Lex Fridman (26:28.240)

and I was able to attract a lot of money from third parties.

Lex Fridman (26:31.160)

And even car companies moved in.

Lex Fridman (26:32.520)

They kind of moved in very quietly

Sebastian Thrun (26:34.040)

because they were super scared to be embarrassed

Lex Fridman (26:36.600)

that their car would flip over.

Lex Fridman (26:38.560)

But Ford was there and Volkswagen was there

Lex Fridman (26:40.680)

and a few others and GM was there.

Lex Fridman (26:43.360)

So it kind of reset the entire landscape of people.

Lex Fridman (26:46.360)

And if you look at who's a big name

Sebastian Thrun (26:48.200)

in self driving cars today,

Lex Fridman (26:49.480)

these were mostly people who participated

Sebastian Thrun (26:51.320)

in those challenges.

Lex Fridman (26:53.400)

Okay, that's incredible.

Lex Fridman (26:54.280)

Can you just comment quickly on your sense of lessons learned

Lex Fridman (26:59.080)

from that kind of funding model

Lex Fridman (27:01.240)

and the research that's going on in academia

Lex Fridman (27:04.400)

in terms of producing papers,

Sebastian Thrun (27:06.120)

is there something to be learned and scaled up bigger,

Lex Fridman (27:10.200)

having these kinds of grand challenges

Lex Fridman (27:11.720)

that could improve outcomes?

Lex Fridman (27:14.560)

So I'm a big believer in focusing

Sebastian Thrun (27:16.320)

on kind of an end to end system.

Lex Fridman (27:19.680)

I'm a really big believer in systems building.

Sebastian Thrun (27:21.920)

I've always built systems in my academic career,

Lex Fridman (27:23.680)

even though I do a lot of math and abstract stuff,

Lex Fridman (27:27.040)

but it's all derived from the idea

Lex Fridman (27:28.160)

of let's solve a real problem.

Lex Fridman (27:29.680)

And it's very hard for me to be an academic

Lex Fridman (27:33.840)

and say, let me solve a component of a problem.

Sebastian Thrun (27:35.800)

Like with someone there's fields like nonmonetary logic

Lex Fridman (27:38.680)

or AI planning systems where people believe

Sebastian Thrun (27:41.800)

that a certain style of problem solving

Lex Fridman (27:44.320)

is the ultimate end objective.

Lex Fridman (27:47.280)

And I would always turn it around and say,

Lex Fridman (27:49.600)

hey, what problem would my grandmother care about

Sebastian Thrun (27:52.640)

that doesn't understand computer technology

Lex Fridman (27:54.680)

and doesn't wanna understand?

Lex Fridman (27:56.520)

And how could I make her love what I do?

Lex Fridman (27:58.480)

Because only then do I have an impact on the world.

Sebastian Thrun (28:01.320)

I can easily impress my colleagues.

Lex Fridman (28:02.960)

That is much easier,

Lex Fridman (28:04.760)

but impressing my grandmother is very, very hard.

Lex Fridman (28:07.640)

So I would always thought if I can build a self driving car

Lex Fridman (28:10.760)

and my grandmother can use it

Lex Fridman (28:12.880)

even after she loses her driving privileges

Sebastian Thrun (28:14.720)

or children can use it,

Lex Fridman (28:16.160)

or we save maybe a million lives a year,

Sebastian Thrun (28:20.560)

that would be very impressive.

Lex Fridman (28:22.440)

And then there's so many problems like these,

Sebastian Thrun (28:23.920)

like there's a problem with curing cancer,

Lex Fridman (28:25.320)

or whatever it is, live twice as long.

Sebastian Thrun (28:27.800)

Once a problem is defined,

Lex Fridman (28:29.600)

of course I can't solve it in its entirety.

Sebastian Thrun (28:31.440)

Like it takes sometimes tens of thousands of people

Lex Fridman (28:34.200)

to find a solution.

Sebastian Thrun (28:35.360)

There's no way you can fund an army of 10,000 at Stanford.

Lex Fridman (28:39.360)

So you gotta build a prototype.

Sebastian Thrun (28:41.080)

Let's build a meaningful prototype.

Lex Fridman (28:42.480)

And the DARPA Grand Challenge was beautiful

Sebastian Thrun (28:43.920)

because it told me what this prototype had to do.

Lex Fridman (28:46.400)

I didn't have to think about what it had to do,

Sebastian Thrun (28:47.680)

I just had to read the rules.

Lex Fridman (28:48.840)

And that was really beautiful.

Lex Fridman (28:51.080)

And it's most beautiful,

Lex Fridman (28:52.320)

you think what academia could aspire to

Sebastian Thrun (28:54.720)

is to build a prototype that's the systems level,

Lex Fridman (28:58.600)

that solves or gives you an inkling

Sebastian Thrun (29:01.360)

that this problem could be solved with this prototype.

Lex Fridman (29:03.480)

First of all, I wanna emphasize what academia really is.

Lex Fridman (29:06.520)

And I think people misunderstand it.

Lex Fridman (29:08.560)

First and foremost, academia is a way

Sebastian Thrun (29:11.280)

to educate young people.

Lex Fridman (29:13.320)

First and foremost, a professor is an educator.

Sebastian Thrun (29:15.400)

No matter where you are at,

Lex Fridman (29:17.040)

a small suburban college,

Sebastian Thrun (29:18.560)

or whether you are a Harvard or Stanford professor,

Lex Fridman (29:21.960)

that's not the way most people think of themselves

Sebastian Thrun (29:25.000)

in academia because we have this kind of competition

Lex Fridman (29:28.000)

going on for citations and publication.

Sebastian Thrun (29:31.440)

That's a measurable thing,

Lex Fridman (29:32.840)

but that is secondary to the primary purpose

Sebastian Thrun (29:35.440)

of educating people to think.

Lex Fridman (29:37.800)

Now, in terms of research,

Sebastian Thrun (29:39.960)

most of the great science,

Lex Fridman (29:42.880)

the great research comes out of universities.

Sebastian Thrun (29:45.520)

You can trace almost everything back,

Lex Fridman (29:46.960)

including Google, to universities.

Lex Fridman (29:48.840)

So there's nothing really fundamentally broken here.

Lex Fridman (29:52.120)

It's a good system.

Lex Fridman (29:53.400)

And I think America has the finest university system

Lex Fridman (29:55.920)

on the planet.

Sebastian Thrun (29:57.640)

We can talk about reach

Lex Fridman (29:59.320)

and how to reach people outside the system.

Sebastian Thrun (30:01.440)

It's a different topic,

Lex Fridman (30:02.280)

but the system itself is a good system.

Sebastian Thrun (30:04.760)

If I had one wish, I would say it'd be really great

Lex Fridman (30:08.320)

if there was more debate about

Lex Fridman (30:11.760)

what the great big problems are in society

Lex Fridman (30:15.880)

and focus on those.

Lex Fridman (30:18.760)

And most of them are interdisciplinary.

Lex Fridman (30:21.600)

Unfortunately, it's very easy to fall

Sebastian Thrun (30:24.640)

into an interdisciplinary viewpoint

Lex Fridman (30:28.160)

where your problem is dictated

Sebastian Thrun (30:30.440)

by what your closest colleagues believe the problem is.

Lex Fridman (30:33.680)

It's very hard to break out and say,

Sebastian Thrun (30:35.280)

well, there's an entire new field of problems.

Lex Fridman (30:37.920)

So to give an example,

Sebastian Thrun (30:39.840)

prior to me working on self driving cars,

Lex Fridman (30:41.640)

I was a roboticist and a machine learning expert.

Lex Fridman (30:44.640)

And I wrote books on robotics,

Lex Fridman (30:46.840)

something called probabilistic robotics.

Sebastian Thrun (30:48.480)

It's a very methods driven kind of viewpoint of the world.

Lex Fridman (30:51.480)

I built robots that acted in museums as tour guides,

Sebastian Thrun (30:54.000)

that let children around.

Lex Fridman (30:55.600)

It is something that at the time was moderately challenging.

Sebastian Thrun (31:00.000)

When I started working on cars,

Lex Fridman (31:02.240)

several colleagues told me,

Sebastian Thrun (31:03.720)

Sebastian, you're destroying your career

Lex Fridman (31:06.080)

because in our field of robotics,

Sebastian Thrun (31:08.160)

cars are looked like as a gimmick

Lex Fridman (31:10.400)

and they're not expressive enough.

Sebastian Thrun (31:11.760)

They can only push the throttle and the brakes.

Lex Fridman (31:15.080)

There's no dexterity.

Sebastian Thrun (31:16.440)

There's no complexity.

Lex Fridman (31:18.240)

It's just too simple.

Lex Fridman (31:19.480)

And no one came to me and said,

Lex Fridman (31:21.200)

wow, if you solve that problem,

Lex Fridman (31:22.720)

you can save a million lives, right?

Lex Fridman (31:25.000)

Among all robotic problems that I've seen in my life,

Sebastian Thrun (31:27.240)

I would say the self driving car, transportation,

Lex Fridman (31:29.760)

is the one that has the most hope for society.

Lex Fridman (31:32.080)

So how come the robotics community wasn't all over the place?

Lex Fridman (31:35.120)

And it was because we focused on methods and solutions

Lex Fridman (31:37.920)

and not on problems.

Lex Fridman (31:39.880)

Like if you go around today and ask your grandmother,

Lex Fridman (31:42.400)

what bugs you?

Lex Fridman (31:43.240)

What really makes you upset?

Sebastian Thrun (31:45.240)

I challenge any academic to do this

Lex Fridman (31:48.720)

and then realize how far your research

Sebastian Thrun (31:51.800)

is probably away from that today.

Lex Fridman (31:54.840)

At the very least, that's a good thing

Sebastian Thrun (31:56.760)

for academics to deliberate on.

Lex Fridman (31:59.240)

The other thing that's really nice in Silicon Valley is,

Sebastian Thrun (32:01.600)

Silicon Valley is full of smart people outside academia.

Lex Fridman (32:04.360)

So there's the Larry Pages and Mark Zuckerbergs in the world

Sebastian Thrun (32:06.720)

who are anywhere smarter, smarter

Lex Fridman (32:09.000)

than the best academics I've met in my life.

Lex Fridman (32:11.400)

And what they do is they are at a different level.

Lex Fridman (32:15.360)

They build the systems,

Sebastian Thrun (32:16.680)

they build the customer facing systems,

Lex Fridman (32:19.280)

they build things that people can use

Sebastian Thrun (32:21.920)

without technical education.

Lex Fridman (32:23.760)

And they are inspired by research.

Sebastian Thrun (32:25.800)

They're inspired by scientists.

Lex Fridman (32:27.480)

They hire the best PhDs from the best universities

Sebastian Thrun (32:30.280)

for a reason.

Lex Fridman (32:31.960)

So I think this kind of vertical integration

Sebastian Thrun (32:35.080)

between the real product, the real impact

Lex Fridman (32:37.720)

and the real thought, the real ideas,

Sebastian Thrun (32:39.800)

that's actually working surprisingly well in Silicon Valley.

Lex Fridman (32:42.720)

It did not work as well in other places in this nation.

Lex Fridman (32:44.840)

So when I worked at Carnegie Mellon,

Lex Fridman (32:46.640)

we had the world's finest computer science university,

Lex Fridman (32:49.800)

but there wasn't those people in Pittsburgh

Lex Fridman (32:52.720)

that would be able to take these

Sebastian Thrun (32:54.280)

very fine computer science ideas

Lex Fridman (32:56.000)

and turn them into massive, impactful products.

Sebastian Thrun (33:00.560)

That symbiosis seemed to exist

Lex Fridman (33:02.800)

pretty much only in Silicon Valley

Lex Fridman (33:04.600)

and maybe a bit in Boston and Austin.

Lex Fridman (33:06.560)

Yeah, with Stanford, that's really interesting.

Lex Fridman (33:11.040)

So if we look a little bit further on

Lex Fridman (33:14.000)

from the DARPA Grand Challenge

Lex Fridman (33:17.120)

and the launch of the Google self driving car,

Lex Fridman (33:20.000)

what do you see as the state,

Sebastian Thrun (33:22.000)

the challenges of autonomous vehicles as they are now

Lex Fridman (33:25.840)

is actually achieving that huge scale

Lex Fridman (33:29.120)

and having a huge impact on society?

Lex Fridman (33:31.640)

I'm extremely proud of what has been accomplished.

Lex Fridman (33:35.200)

And again, I'm taking a lot of credit for the work of others.

Lex Fridman (33:38.280)

And I'm actually very optimistic.

Lex Fridman (33:40.160)

And people have been kind of worrying,

Lex Fridman (33:42.320)

is it too fast? Is it too slow?

Lex Fridman (33:43.800)

Why is it not there yet? And so on.

Lex Fridman (33:45.840)

It is actually quite an interesting, hard problem.

Lex Fridman (33:48.800)

And in that a self driving car,

Lex Fridman (33:51.640)

to build one that manages 90% of the problems

Sebastian Thrun (33:55.280)

encountered in everyday driving is easy.

Lex Fridman (33:57.200)

We can literally do this over a weekend.

Sebastian Thrun (33:59.440)

To do 99% might take a month.

Lex Fridman (34:02.040)

Then there's 1% left.

Lex Fridman (34:03.200)

So 1% would mean that you still have a fatal accident

Lex Fridman (34:06.920)

every week, very unacceptable.

Lex Fridman (34:08.960)

So now you work on this 1%

Lex Fridman (34:10.920)

and the 99% of that, the remaining 1%

Sebastian Thrun (34:13.640)

is actually still relatively easy,

Lex Fridman (34:15.760)

but now you're down to like a hundredth of 1%.

Lex Fridman (34:18.160)

And it's still completely unacceptable in terms of safety.

Lex Fridman (34:21.560)

So the variety of things you encounter are just enormous.

Lex Fridman (34:24.200)

And that gives me enormous respect for human being

Lex Fridman (34:26.440)

that we're able to deal with the couch on the highway,

Sebastian Thrun (34:30.440)

or the deer in the headlights, or the blown tire

Lex Fridman (34:33.440)

that we've never been trained for.

Lex Fridman (34:34.880)

And all of a sudden have to handle it

Lex Fridman (34:35.960)

in an emergency situation

Lex Fridman (34:37.080)

and often do very, very successfully.

Lex Fridman (34:38.720)

It's amazing from that perspective,

Lex Fridman (34:40.640)

how safe driving actually is given how many millions

Lex Fridman (34:43.640)

of miles we drive every year in this country.

Sebastian Thrun (34:47.600)

We are now at a point where I believe the technology

Lex Fridman (34:49.400)

is there and I've seen it.

Sebastian Thrun (34:51.560)

I've seen it in Waymo, I've seen it in Aptiv,

Lex Fridman (34:53.520)

I've seen it in Cruise and in a number of companies

Lex Fridman (34:56.760)

and in Voyage where vehicles now driving around

Lex Fridman (35:00.920)

and basically flawlessly are able to drive people around

Sebastian Thrun (35:04.360)

in limited scenarios.

Lex Fridman (35:06.040)

In fact, you can go to Vegas today

Lex Fridman (35:07.960)

and order a Summon and Lift.

Lex Fridman (35:09.880)

And if you get the right setting of your app,

Sebastian Thrun (35:13.480)

you'll be picked up by a driverless car.

Lex Fridman (35:15.760)

Now there's still safety drivers in there,

Lex Fridman (35:18.040)

but that's a fantastic way to kind of learn

Lex Fridman (35:21.280)

what the limits are of technology today.

Lex Fridman (35:22.920)

And there's still some glitches,

Lex Fridman (35:24.680)

but the glitches have become very, very rare.

Sebastian Thrun (35:26.520)

I think the next step is gonna be to down cost it,

Lex Fridman (35:29.680)

to harden it, the entrapment, the sensors

Sebastian Thrun (35:33.720)

are not quite an automotive grade standard yet.

Lex Fridman (35:36.120)

And then to really build the business models,

Sebastian Thrun (35:37.760)

to really kind of go somewhere and make the business case.

Lex Fridman (35:40.920)

And the business case is hard work.

Sebastian Thrun (35:42.520)

It's not just, oh my God, we have this capability,

Lex Fridman (35:44.560)

people are just gonna buy it.

Sebastian Thrun (35:45.480)

You have to make it affordable.

Lex Fridman (35:46.680)

You have to find the social acceptance of people.

Sebastian Thrun (35:52.240)

None of the teams yet has been able to or gutsy enough

Lex Fridman (35:55.360)

to drive around without a person inside the car.

Lex Fridman (35:59.240)

And that's the next magical hurdle.

Lex Fridman (36:01.320)

We'll be able to send these vehicles around

Sebastian Thrun (36:03.800)

completely empty in traffic.

Lex Fridman (36:05.760)

And I think, I mean, I wait every day,

Sebastian Thrun (36:08.120)

wait for the news that Waymo has just done this.

Lex Fridman (36:11.840)

So, interesting you mentioned gutsy.

Sebastian Thrun (36:15.080)

Let me ask some maybe unanswerable question,

Lex Fridman (36:20.200)

maybe edgy questions.

Lex Fridman (36:21.480)

But in terms of how much risk is required,

Lex Fridman (36:26.880)

some guts in terms of leadership style,

Sebastian Thrun (36:30.360)

it would be good to contrast approaches.

Lex Fridman (36:32.600)

And I don't think anyone knows what's right.

Lex Fridman (36:34.680)

But if we compare Tesla and Waymo, for example,

Lex Fridman (36:38.560)

Elon Musk and the Waymo team,

Sebastian Thrun (36:43.200)

there's slight differences in approach.

Lex Fridman (36:45.680)

So on the Elon side, there's more,

Sebastian Thrun (36:49.560)

I don't know what the right word to use,

Lex Fridman (36:50.840)

but aggression in terms of innovation.

Lex Fridman (36:53.920)

And on Waymo side, there's more sort of cautious,

Lex Fridman (36:59.800)

safety focused approach to the problem.

Lex Fridman (37:03.480)

What do you think it takes?

Lex Fridman (37:06.200)

What leadership at which moment is right?

Lex Fridman (37:09.160)

Which approach is right?

Lex Fridman (37:11.600)

Look, I don't sit in either of those teams.

Lex Fridman (37:13.880)

So I'm unable to even verify like somebody says correct.

Lex Fridman (37:18.000)

In the end of the day, every innovator in that space

Sebastian Thrun (37:21.240)

will face a fundamental dilemma.

Lex Fridman (37:23.160)

And I would say you could put aerospace titans

Sebastian Thrun (37:27.120)

into the same bucket,

Lex Fridman (37:28.880)

which is you have to balance public safety

Sebastian Thrun (37:31.600)

with your drive to innovate.

Lex Fridman (37:34.280)

And this country in particular in the States

Sebastian Thrun (37:36.760)

has a hundred plus year history

Lex Fridman (37:38.320)

of doing this very successfully.

Sebastian Thrun (37:40.600)

Air travel is what a hundred times a safe per mile

Lex Fridman (37:43.880)

than ground travel, than cars.

Lex Fridman (37:46.600)

And there's a reason for it because people have found ways

Lex Fridman (37:50.320)

to be very methodological about ensuring public safety

Sebastian Thrun (37:55.080)

while still being able to make progress

Lex Fridman (37:56.880)

on important aspects, for example,

Sebastian Thrun (37:59.000)

like air and noise and fuel consumption.

Lex Fridman (38:03.600)

So I think that those practices are proven

Lex Fridman (38:06.120)

and they actually work.

Lex Fridman (38:07.840)

We live in a world safer than ever before.

Lex Fridman (38:09.840)

And yes, there will always be the provision

Lex Fridman (38:11.880)

that something goes wrong.

Sebastian Thrun (38:12.720)

There's always the possibility

Lex Fridman (38:14.040)

that someone makes a mistake

Sebastian Thrun (38:15.240)

or there's an unexpected failure.

Lex Fridman (38:17.120)

We can never guarantee to a hundred percent

Sebastian Thrun (38:19.720)

absolute safety other than just not doing it.

Lex Fridman (38:23.320)

But I think I'm very proud of the history of the United States.

Sebastian Thrun (38:27.080)

I mean, we've dealt with much more dangerous technology

Lex Fridman (38:30.120)

like nuclear energy and kept that safe too.

Sebastian Thrun (38:33.760)

We have nuclear weapons and we keep those safe.

Lex Fridman (38:36.400)

So we have methods and procedures

Sebastian Thrun (38:39.440)

that really balance these two things very, very successfully.

Lex Fridman (38:42.920)

You've mentioned a lot of great autonomous vehicle companies

Sebastian Thrun (38:46.320)

that are taking sort of the level four, level five,

Lex Fridman (38:48.760)

they jump in full autonomy with a safety driver

Lex Fridman (38:51.840)

and take that kind of approach

Lex Fridman (38:53.120)

and also through simulation and so on.

Sebastian Thrun (38:55.760)

There's also the approach that Tesla Autopilot is doing,

Lex Fridman (38:59.560)

which is kind of incrementally taking a level two vehicle

Lex Fridman (39:03.680)

and using machine learning

Lex Fridman (39:04.920)

and learning from the driving of human beings

Lex Fridman (39:08.360)

and trying to creep up,

Lex Fridman (39:10.560)

trying to incrementally improve the system

Sebastian Thrun (39:12.360)

until it's able to achieve level four autonomy.

Lex Fridman (39:15.520)

So perfect autonomy in certain kind of geographical regions.

Lex Fridman (39:19.760)

What are your thoughts on these contrasting approaches?

Lex Fridman (39:23.120)

Well, so first of all, I'm a very proud Tesla owner

Lex Fridman (39:25.560)

and I literally use the Autopilot every day

Lex Fridman (39:27.840)

and it literally has kept me safe.

Sebastian Thrun (39:30.760)

It is a beautiful technology specifically

Lex Fridman (39:33.920)

for highway driving when I'm slightly tired

Sebastian Thrun (39:37.600)

because then it turns me into a much safer driver.

Lex Fridman (39:42.200)

And I'm 100% confident that's the case.

Sebastian Thrun (39:46.520)

In terms of the right approach,

Lex Fridman (39:47.680)

I think the biggest change I've seen

Sebastian Thrun (39:49.880)

since I went to Waymo team is this thing called deep learning.

Lex Fridman (39:54.280)

I think deep learning was not a hot topic

Sebastian Thrun (39:56.320)

when I started Waymo or Google self driving cars.

Lex Fridman (39:59.400)

It was there, in fact, we started Google Brain

Sebastian Thrun (40:01.760)

at the same time in Google X.

Lex Fridman (40:02.840)

So I invested in deep learning,

Lex Fridman (40:04.760)

but people didn't talk about it, it wasn't a hot topic.

Lex Fridman (40:07.840)

And now it is, there's a shift of emphasis

Sebastian Thrun (40:10.360)

from a more geometric perspective

Lex Fridman (40:12.440)

where you use geometric sensors

Sebastian Thrun (40:14.320)

that give you a full 3D view

Lex Fridman (40:15.680)

when you do a geometric reasoning about,

Sebastian Thrun (40:17.280)

oh, this box over here might be a car

Lex Fridman (40:19.640)

towards a more human like, oh, let's just learn about it.

Sebastian Thrun (40:24.160)

This looks like the thing I've seen 10,000 times before.

Lex Fridman (40:26.520)

So maybe it's the same thing, machine learning perspective.

Lex Fridman (40:30.280)

And that has really put, I think,

Lex Fridman (40:32.160)

all these approaches on steroids.

Sebastian Thrun (40:36.000)

At Udacity, we teach a course in self driving cars.

Lex Fridman (40:38.720)

In fact, I think we've graduated over 20,000 or so people

Sebastian Thrun (40:43.800)

on self driving car skills.

Lex Fridman (40:45.000)

So every self driving car team in the world

Sebastian Thrun (40:47.440)

now uses our engineers.

Lex Fridman (40:49.280)

And in this course, the very first homework assignment

Sebastian Thrun (40:51.920)

is to do lane finding on images.

Lex Fridman (40:54.920)

And lane finding images for layman,

Lex Fridman (40:56.960)

what this means is you put a camera into your car

Lex Fridman (40:59.040)

or you open your eyes and you would know where the lane is.

Lex Fridman (41:02.440)

So you can stay inside the lane with your car.

Lex Fridman (41:05.000)

Humans can do this super easily.

Sebastian Thrun (41:06.520)

You just look and you know where the lane is,

Lex Fridman (41:08.120)

just intuitively.

Sebastian Thrun (41:10.200)

For machines, for a long time, it was super hard

Lex Fridman (41:12.240)

because people would write these kind of crazy rules.

Sebastian Thrun (41:14.680)

If there's like wine lane markers

Lex Fridman (41:16.120)

and here's what white really means,

Sebastian Thrun (41:17.680)

this is not quite white enough.

Lex Fridman (41:19.160)

So let's, oh, it's not white.

Sebastian Thrun (41:20.360)

Or maybe the sun is shining.

Lex Fridman (41:21.480)

So when the sun shines and this is white

Lex Fridman (41:23.520)

and this is a straight line,

Lex Fridman (41:24.720)

I mean, it's not quite a straight line

Sebastian Thrun (41:25.760)

because the road is curved.

Lex Fridman (41:27.320)

And do we know that there's really six feet

Sebastian Thrun (41:29.280)

between lane markings or not or 12 feet, whatever it is.

Lex Fridman (41:34.000)

And now what the students are doing,

Sebastian Thrun (41:36.320)

they would take machine learning.

Lex Fridman (41:37.440)

So instead of like writing these crazy rules

Sebastian Thrun (41:39.640)

for the lane marker,

Lex Fridman (41:40.480)

they'll say, hey, let's take an hour of driving

Lex Fridman (41:42.720)

and label it and tell the vehicle,

Lex Fridman (41:44.440)

this is actually the lane by hand.

Lex Fridman (41:45.800)

And then these are examples

Lex Fridman (41:47.360)

and have the machine find its own rules,

Lex Fridman (41:49.400)

what lane markings are.

Lex Fridman (41:51.400)

And within 24 hours, now every student

Sebastian Thrun (41:53.800)

that's never done any programming before in this space

Lex Fridman (41:56.040)

can write a perfect lane finder

Sebastian Thrun (41:58.320)

as good as the best commercial lane finders.

Lex Fridman (42:00.880)

And that's completely amazing to me.

Sebastian Thrun (42:02.760)

We've seen progress using machine learning

Lex Fridman (42:05.520)

that completely dwarfs anything

Sebastian Thrun (42:08.160)

that I saw 10 years ago.

Lex Fridman (42:10.960)

Yeah, and just as a side note,

Sebastian Thrun (42:12.840)

the self driving car nanodegree,

Lex Fridman (42:15.240)

the fact that you launched that many years ago now,

Sebastian Thrun (42:18.960)

maybe four years ago, three years ago is incredible

Lex Fridman (42:22.080)

that that's a great example of system level thinking

Sebastian Thrun (42:24.760)

sort of just taking an entire course

Lex Fridman (42:27.160)

that teaches you how to solve the entire problem.

Sebastian Thrun (42:29.280)

I definitely recommend people.

Lex Fridman (42:31.240)

It's become super popular

Lex Fridman (42:32.480)

and it's become actually incredibly high quality

Lex Fridman (42:34.320)

really with Mercedes and various other companies

Sebastian Thrun (42:37.360)

in that space.

Lex Fridman (42:38.200)

And we find that engineers from Tesla and Waymo

Sebastian Thrun (42:40.600)

are taking it today.

Lex Fridman (42:43.120)

The insight was that two things,

Sebastian Thrun (42:45.520)

one is existing universities will be very slow to move

Lex Fridman (42:49.240)

because they're departmentalized

Lex Fridman (42:50.520)

and there's no department for self driving cars.

Lex Fridman (42:52.360)

So between Mac E and double E and computer science,

Sebastian Thrun (42:56.240)

getting those folks together

Lex Fridman (42:57.240)

into one room is really, really hard.

Lex Fridman (42:59.680)

And every professor listening here will know,

Lex Fridman (43:01.280)

they'll probably agree to that.

Lex Fridman (43:02.960)

And secondly, even if all the great universities

Lex Fridman (43:06.400)

just did this, which none so far has developed

Sebastian Thrun (43:09.120)

a curriculum in this field,

Lex Fridman (43:11.120)

it is just a few thousand students that can partake

Sebastian Thrun (43:13.720)

because all the great universities are super selective.

Lex Fridman (43:16.280)

So how about people in India?

Lex Fridman (43:18.160)

How about people in China or in the Middle East

Lex Fridman (43:20.680)

or Indonesia or Africa?

Lex Fridman (43:23.480)

Why should those be excluded

Lex Fridman (43:25.200)

from the skill of building self driving cars?

Lex Fridman (43:27.280)

Are they any dumber than we are?

Lex Fridman (43:28.480)

Are we any less privileged?

Lex Fridman (43:30.240)

And the answer is we should just give everybody the skill

Lex Fridman (43:34.880)

to build a self driving car.

Sebastian Thrun (43:35.920)

Because if we do this,

Lex Fridman (43:37.440)

then we have like a thousand self driving car startups.

Lex Fridman (43:40.360)

And if 10% succeed, that's like a hundred,

Lex Fridman (43:42.960)

that means hundred countries now

Sebastian Thrun (43:44.200)

will have self driving cars and be safer.

Lex Fridman (43:46.800)

It's kind of interesting to imagine impossible to quantify,

Lex Fridman (43:50.360)

but the number, the, you know,

Lex Fridman (43:53.600)

over a period of several decades,

Sebastian Thrun (43:55.080)

the impact that has like a single course,

Lex Fridman (43:57.960)

like a ripple effect of society.

Sebastian Thrun (44:00.760)

If you, I just recently talked to Andrew

Lex Fridman (44:03.520)

who was creator of Cosmos show.

Sebastian Thrun (44:06.560)

It's interesting to think about

Lex Fridman (44:08.200)

how many scientists that show launched.

Lex Fridman (44:10.720)

And so it's really, in terms of impact,

Lex Fridman (44:15.600)

I can't imagine a better course

Sebastian Thrun (44:17.200)

than the self driving car course.

Lex Fridman (44:18.680)

That's, you know, there's other more specific disciplines

Sebastian Thrun (44:21.840)

like deep learning and so on that Udacity is also teaching,

Lex Fridman (44:24.120)

but self driving cars,

Sebastian Thrun (44:25.160)

it's really, really interesting course.

Lex Fridman (44:26.920)

And then it came at the right moment.

Sebastian Thrun (44:28.440)

It came at a time when there were a bunch of Acqui hires.

Lex Fridman (44:31.720)

Acqui hire is a acquisition of a company,

Sebastian Thrun (44:34.200)

not for its technology or its products or business,

Lex Fridman (44:36.400)

but for its people.

Lex Fridman (44:38.320)

So Acqui hire means maybe that a company of 70 people,

Lex Fridman (44:40.640)

they have no product yet, but they're super smart people

Lex Fridman (44:43.160)

and they pay a certain amount of money.

Lex Fridman (44:44.320)

So I took Acqui hires like GM Cruise and Uber and others,

Lex Fridman (44:48.440)

and did the math and said,

Lex Fridman (44:50.120)

hey, how many people are there and how much money was paid?

Lex Fridman (44:53.760)

And as a lower bound,

Lex Fridman (44:55.640)

I estimated the value of a self driving car engineer

Lex Fridman (44:58.560)

in these acquisitions to be at least $10 million, right?

Lex Fridman (45:02.240)

So think about this, you get yourself a skill

Lex Fridman (45:05.080)

and you team up and build a company

Lex Fridman (45:06.680)

and your worth now is $10 million.

Sebastian Thrun (45:09.800)

I mean, that's kind of cool.

Lex Fridman (45:10.840)

I mean, what other thing could you do in life

Lex Fridman (45:13.440)

to be worth $10 million within a year?

Lex Fridman (45:15.920)

Yeah, amazing.

Lex Fridman (45:17.640)

But to come back for a moment on to deep learning

Lex Fridman (45:21.000)

and its application in autonomous vehicles,

Lex Fridman (45:23.760)

what are your thoughts on Elon Musk's statement,

Lex Fridman (45:28.480)

provocative statement, perhaps that light air is a crutch.

Lex Fridman (45:31.080)

So this geometric way of thinking about the world

Lex Fridman (45:34.000)

may be holding us back if what we should instead be doing

Sebastian Thrun (45:38.920)

in this robotic space,

Lex Fridman (45:39.920)

in this particular space of autonomous vehicles

Sebastian Thrun (45:42.520)

is using camera as a primary sensor

Lex Fridman (45:46.440)

and using computer vision and machine learning

Sebastian Thrun (45:48.200)

as the primary way to...

Lex Fridman (45:49.720)

Look, I have two comments.

Sebastian Thrun (45:50.560)

I think first of all, we all know

Lex Fridman (45:52.240)

that people can drive cars without lighters in their heads

Sebastian Thrun (45:56.880)

because we only have eyes

Lex Fridman (45:59.000)

and we mostly just use eyes for driving.

Sebastian Thrun (46:02.080)

Maybe we use some other perception about our bodies,

Lex Fridman (46:04.560)

accelerations, occasionally our ears,

Sebastian Thrun (46:08.000)

certainly not our noses.

Lex Fridman (46:10.680)

So the existence proof is there,

Sebastian Thrun (46:12.440)

that eyes must be sufficient.

Lex Fridman (46:15.560)

In fact, we could even drive a car

Sebastian Thrun (46:17.920)

if someone put a camera out

Lex Fridman (46:19.440)

and then gave us the camera image with no latency,

Sebastian Thrun (46:23.440)

we would be able to drive a car that way the same way.

Lex Fridman (46:26.360)

So a camera is also sufficient.

Sebastian Thrun (46:28.720)

Secondly, I really love the idea that in the Western world,

Lex Fridman (46:31.840)

we have many, many different people

Sebastian Thrun (46:33.600)

trying different hypotheses.

Lex Fridman (46:35.680)

It's almost like an anthill,

Sebastian Thrun (46:36.840)

like if an anthill tries to forge for food,

Lex Fridman (46:39.560)

you can sit there as two ants

Lex Fridman (46:41.000)

and agree what the perfect path is

Lex Fridman (46:42.560)

and then every single ant marches

Sebastian Thrun (46:44.040)

for the most likely location of food is,

Lex Fridman (46:46.320)

or you can even just spread out.

Lex Fridman (46:47.960)

And I promise you the spread out solution will be better

Lex Fridman (46:50.440)

because if the discussing philosophical,

Sebastian Thrun (46:53.960)

intellectual ants get it wrong

Lex Fridman (46:55.560)

and they're all moving the wrong direction,

Sebastian Thrun (46:56.920)

they're going to waste a day

Lex Fridman (46:58.240)

and then they're going to discuss again for another week.

Sebastian Thrun (47:00.520)

Whereas if all these ants go in a random direction,

Lex Fridman (47:02.480)

someone's going to succeed

Lex Fridman (47:03.520)

and they're going to come back and claim victory

Lex Fridman (47:05.560)

and get the Nobel prize or whatever the ant equivalent is.

Lex Fridman (47:08.520)

And then they all march in the same direction.

Lex Fridman (47:10.520)

And that's great about society.

Sebastian Thrun (47:11.800)

That's great about the Western society.

Lex Fridman (47:13.160)

We're not plan based, we're not central based.

Sebastian Thrun (47:15.480)

We don't have a Soviet Union style central government

Lex Fridman (47:19.120)

that tells us where to forge.

Sebastian Thrun (47:20.960)

We just forge.

Lex Fridman (47:21.800)

We started in C Corp.

Sebastian Thrun (47:24.040)

We get investor money, go out and try it out.

Lex Fridman (47:25.840)

And who knows who's going to win.

Sebastian Thrun (47:28.720)

I like it.

Lex Fridman (47:30.160)

In your, when you look at the longterm vision

Sebastian Thrun (47:33.440)

of autonomous vehicles,

Lex Fridman (47:35.160)

do you see machine learning

Lex Fridman (47:36.920)

as fundamentally being able to solve most of the problems?

Lex Fridman (47:39.600)

So learning from experience.

Sebastian Thrun (47:42.280)

I'd say we should be very clear

Lex Fridman (47:44.200)

about what machine learning is and is not.

Lex Fridman (47:46.080)

And I think there's a lot of confusion.

Lex Fridman (47:48.160)

What it is today is a technology

Sebastian Thrun (47:50.880)

that can go through large databases

Lex Fridman (47:54.680)

of repetitive patterns and find those patterns.

Lex Fridman (48:00.880)

So in example, we did a study at Stanford two years ago

Lex Fridman (48:03.560)

where we applied machine learning

Sebastian Thrun (48:05.440)

to detecting skin cancer in images.

Lex Fridman (48:07.880)

And we harvested or built a data set

Sebastian Thrun (48:10.760)

of 129,000 skin photo shots

Lex Fridman (48:15.080)

that were all had been biopsied

Sebastian Thrun (48:17.000)

for what the actual situation was.

Lex Fridman (48:19.440)

And those included melanomas and carcinomas,

Sebastian Thrun (48:22.680)

also included rashes and other skin conditions, lesions.

Lex Fridman (48:27.200)

And then we had a network find those patterns.

Lex Fridman (48:30.720)

And it was by and large able to then detect skin cancer

Lex Fridman (48:34.520)

with an iPhone as accurately

Sebastian Thrun (48:36.680)

as the best board certified Stanford level dermatologist.

Lex Fridman (48:41.400)

We proved that.

Sebastian Thrun (48:42.800)

Now this thing was great in this one thing

Lex Fridman (48:45.880)

and finding skin cancer, but it couldn't drive a car.

Lex Fridman (48:49.680)

So the difference to human intelligence

Lex Fridman (48:51.600)

is we do all these many, many things

Lex Fridman (48:53.280)

and we can often learn from a very small data set

Lex Fridman (48:56.720)

of experiences.

Sebastian Thrun (48:58.160)

Whereas machines still need very large data sets

Lex Fridman (49:01.120)

and things that will be very repetitive.

Sebastian Thrun (49:03.320)

Now that's still super impactful

Lex Fridman (49:04.680)

because almost everything we do is repetitive.

Lex Fridman (49:06.440)

So that's gonna really transform human labor

Lex Fridman (49:10.000)

but it's not this almighty general intelligence.

Sebastian Thrun (49:13.120)

We're really far away from a system

Lex Fridman (49:15.280)

that will exhibit general intelligence.

Sebastian Thrun (49:18.760)

To that end, I actually commiserate the naming a little bit

Lex Fridman (49:21.320)

because artificial intelligence, if you believe Hollywood

Sebastian Thrun (49:24.440)

is immediately mixed into the idea of human suppression

Lex Fridman (49:27.320)

and machine superiority.

Sebastian Thrun (49:30.360)

I don't think that we're gonna see this in my lifetime.

Lex Fridman (49:32.960)

I don't think human suppression is a good idea.

Sebastian Thrun (49:36.440)

I don't see it coming.

Lex Fridman (49:37.440)

I don't see the technology being there.

Lex Fridman (49:39.720)

What I see instead is a very pointed focused

Lex Fridman (49:42.960)

pattern recognition technology that's able to

Sebastian Thrun (49:45.440)

extract patterns from large data sets.

Lex Fridman (49:48.400)

And in doing so, it can be super impactful.

Sebastian Thrun (49:51.520)

Super impactful.

Lex Fridman (49:53.520)

Let's take the impact of artificial intelligence

Sebastian Thrun (49:55.920)

on human work.

Lex Fridman (49:57.640)

We all know that it takes something like 10,000 hours

Sebastian Thrun (50:00.520)

to become an expert.

Lex Fridman (50:01.520)

If you're gonna be a doctor or a lawyer

Sebastian Thrun (50:03.360)

or even a really good driver,

Lex Fridman (50:05.320)

it takes a certain amount of time to become experts.

Sebastian Thrun (50:08.520)

Machines now are able and have been shown

Lex Fridman (50:11.400)

to observe people become experts and observe experts

Lex Fridman (50:15.640)

and then extract those rules from experts

Lex Fridman (50:17.440)

in some interesting way.

Sebastian Thrun (50:18.680)

They could go from law to sales to driving cars

Lex Fridman (50:25.840)

to diagnosing cancer.

Lex Fridman (50:28.200)

And then giving that capability to people who are

Lex Fridman (50:30.840)

completely new in their job.

Sebastian Thrun (50:32.320)

We now can, and that's been done.

Lex Fridman (50:34.760)

It's been done commercially in many, many instantiations.

Lex Fridman (50:37.800)

So that means we can use machine learning

Lex Fridman (50:40.120)

to make people expert on the very first day of their work.

Sebastian Thrun (50:44.880)

Like think about the impact.

Lex Fridman (50:45.880)

If your doctor is still in their first 10,000 hours,

Sebastian Thrun (50:50.360)

you have a doctor who is not quite an expert yet.

Lex Fridman (50:53.120)

Who would not want a doctor who is the world's best expert?

Lex Fridman (50:56.720)

And now we can leverage machines to really eradicate

Lex Fridman (51:00.400)

the error in decision making,

Sebastian Thrun (51:02.760)

error and lack of expertise for human doctors.

Lex Fridman (51:06.240)

That could save your life.

Sebastian Thrun (51:08.360)

If we can link on that for a little bit,

Lex Fridman (51:10.360)

in which way do you hope machines in the medical field

Lex Fridman (51:14.800)

could help assist doctors?

Lex Fridman (51:16.360)

You mentioned this sort of accelerating the learning curve

Sebastian Thrun (51:21.320)

or people, if they start a job or in the first 10,000 hours

Lex Fridman (51:26.120)

can be assisted by machines.

Lex Fridman (51:27.360)

How do you envision that assistance looking?

Lex Fridman (51:29.720)

So we built this app for an iPhone that can detect

Lex Fridman (51:33.480)

and classify and diagnose skin cancer.

Lex Fridman (51:36.320)

And we proved two years ago that it does pretty much

Sebastian Thrun (51:40.560)

as good or better than the best human doctors.

Lex Fridman (51:42.240)

So let me tell you a story.

Lex Fridman (51:43.600)

So there's a friend of mine, let's call him Ben.

Lex Fridman (51:45.480)

Ben is a very famous venture capitalist.

Sebastian Thrun (51:47.680)

He goes to his doctor and the doctor looks at a mole

Lex Fridman (51:50.720)

and says, hey, that mole is probably harmless.

Lex Fridman (51:55.360)

And for some very funny reason, he pulls out that phone

Lex Fridman (51:59.800)

with our app.

Sebastian Thrun (52:00.640)

He's a collaborator in our study.

Lex Fridman (52:02.640)

And the app says, no, no, no, no, this is a melanoma.

Lex Fridman (52:06.320)

And for background, melanomas are,

Lex Fridman (52:08.720)

and skin cancer is the most common cancer in this country.

Sebastian Thrun (52:12.400)

Melanomas can go from stage zero to stage four

Lex Fridman (52:16.640)

within less than a year.

Sebastian Thrun (52:18.120)

Stage zero means you can basically cut it out yourself

Lex Fridman (52:20.880)

with a kitchen knife and be safe.

Lex Fridman (52:23.200)

And stage four means your chances of living

Lex Fridman (52:25.520)

five more years in less than 20%.

Lex Fridman (52:28.000)

So it's a very serious, serious, serious condition.

Lex Fridman (52:31.160)

So this doctor who took out the iPhone,

Sebastian Thrun (52:36.160)

looked at the iPhone and was a little bit puzzled.

Lex Fridman (52:37.680)

He said, I mean, but just to be safe,

Sebastian Thrun (52:39.720)

let's cut it out and biopsy it.

Lex Fridman (52:41.600)

That's the technical term for let's get

Sebastian Thrun (52:43.560)

an in depth diagnostics that is more than just looking at it.

Lex Fridman (52:47.720)

And it came back as cancerous, as a melanoma.

Lex Fridman (52:50.760)

And it was then removed.

Lex Fridman (52:52.240)

And my friend, Ben, I was hiking with him

Lex Fridman (52:54.960)

and we were talking about AI.

Lex Fridman (52:56.280)

And I told him I do this work on skin cancer.

Lex Fridman (52:58.880)

And he said, oh, funny.

Lex Fridman (53:00.720)

My doctor just had an iPhone that found my cancer.

Lex Fridman (53:05.480)

So I was like completely intrigued.

Lex Fridman (53:06.920)

I didn't even know about this.

Lex Fridman (53:08.200)

So here's a person, I mean, this is a real human life, right?

Lex Fridman (53:11.640)

Like who doesn't know somebody

Sebastian Thrun (53:12.920)

who has been affected by cancer.

Lex Fridman (53:14.000)

Cancer is cause of death number two.

Sebastian Thrun (53:16.160)

Cancer is this kind of disease that is mean

Lex Fridman (53:19.440)

in the following way.

Sebastian Thrun (53:21.080)

Most cancers can actually be cured relatively easily

Lex Fridman (53:24.520)

if we catch them early.

Lex Fridman (53:25.880)

And the reason why we don't tend to catch them early

Lex Fridman (53:28.360)

is because they have no symptoms.

Sebastian Thrun (53:30.600)

Like your very first symptom of a gallbladder cancer

Lex Fridman (53:33.880)

or a pancreas cancer might be a headache.

Lex Fridman (53:37.040)

And when you finally go to your doctor

Lex Fridman (53:38.680)

because of these headaches or your back pain

Lex Fridman (53:41.600)

and you're being imaged, it's usually stage four plus.

Lex Fridman (53:45.880)

And that's the time when the occurring chances

Sebastian Thrun (53:48.200)

might be dropped to a single digit percentage.

Lex Fridman (53:50.880)

So if we could leverage AI to inspect your body

Sebastian Thrun (53:54.560)

on a regular basis without even a doctor in the room,

Lex Fridman (53:58.120)

maybe when you take a shower or what have you,

Sebastian Thrun (54:00.360)

I know this sounds creepy,

Lex Fridman (54:01.480)

but then we might be able to save millions

Lex Fridman (54:03.800)

and millions of lives.

Lex Fridman (54:06.320)

You've mentioned there's a concern that people have

Sebastian Thrun (54:09.520)

about near term impacts of AI in terms of job loss.

Lex Fridman (54:12.880)

So you've mentioned being able to assist doctors,

Sebastian Thrun (54:15.560)

being able to assist people in their jobs.

Lex Fridman (54:17.940)

Do you have a worry of people losing their jobs

Lex Fridman (54:22.260)

or the economy being affected by the improvements in AI?

Lex Fridman (54:25.480)

Yeah, anybody concerned about job losses,

Sebastian Thrun (54:27.680)

please come to Gdacity.com.

Lex Fridman (54:30.040)

We teach contemporary tech skills

Lex Fridman (54:32.320)

and we have a kind of implicit job promise.

Lex Fridman (54:36.680)

We often, when we measure,

Sebastian Thrun (54:38.960)

we spend way over 50% of our graders in new jobs

Lex Fridman (54:41.840)

and they're very satisfied about it.

Lex Fridman (54:43.720)

And it costs almost nothing,

Lex Fridman (54:44.800)

costs like 1,500 max or something like that.

Lex Fridman (54:47.120)

And so there's a cool new program

Lex Fridman (54:48.920)

that you agree with the U.S. government,

Sebastian Thrun (54:51.080)

guaranteeing that you will help us give scholarships

Lex Fridman (54:54.880)

that educate people in this kind of situation.

Sebastian Thrun (54:57.840)

Yeah, we're working with the U.S. government

Lex Fridman (54:59.960)

on the idea of basically rebuilding the American dream.

Lex Fridman (55:03.880)

So Gdacity has just dedicated 100,000 scholarships

Lex Fridman (55:07.440)

for citizens of America for various levels of courses

Sebastian Thrun (55:12.080)

that eventually will get you a job.

Lex Fridman (55:15.560)

And those courses are all somewhat related

Sebastian Thrun (55:18.740)

to the tech sector because the tech sector

Lex Fridman (55:20.460)

is kind of the hottest sector right now.

Lex Fridman (55:22.060)

And they range from interlevel digital marketing

Lex Fridman (55:24.940)

to very advanced self diving car engineering.

Lex Fridman (55:28.060)

And we're doing this with the White House

Lex Fridman (55:29.420)

because we think it's bipartisan.

Sebastian Thrun (55:30.860)

It's an issue that if you wanna really make America great,

Lex Fridman (55:36.020)

being able to be a part of the solution

Lex Fridman (55:40.060)

and live the American dream requires us to be proactive

Lex Fridman (55:43.780)

about our education and our skillset.

Sebastian Thrun (55:45.780)

It's just the way it is today.

Lex Fridman (55:47.700)

And it's always been this way.

Lex Fridman (55:48.700)

And we always had this American dream

Lex Fridman (55:49.940)

to send our kids to college.

Lex Fridman (55:51.140)

And now the American dream has to be

Lex Fridman (55:53.260)

to send ourselves to college.

Sebastian Thrun (55:54.660)

We can do this very, very, very efficiently

Lex Fridman (55:58.220)

and very, very, we can squeeze in in the evenings

Lex Fridman (56:00.900)

and things to online.

Lex Fridman (56:01.820)

So at all ages.

Sebastian Thrun (56:03.140)

All ages.

Lex Fridman (56:03.980)

So our learners go from age 11 to age 80.

Sebastian Thrun (56:08.980)

I just traveled Germany and the guy in the train compartment

Lex Fridman (56:15.180)

next to me was one of my students.

Sebastian Thrun (56:17.500)

It's like, wow, that's amazing.

Lex Fridman (56:19.820)

Think about impact.

Sebastian Thrun (56:21.020)

We've become the educator of choice for now,

Lex Fridman (56:24.020)

I believe officially six countries or five countries.

Sebastian Thrun (56:26.500)

Most in the Middle East, like Saudi Arabia and in Egypt.

Lex Fridman (56:30.080)

In Egypt, we just had a cohort graduate

Sebastian Thrun (56:33.420)

where we had 1100 high school students

Lex Fridman (56:37.280)

that went through programming skills,

Sebastian Thrun (56:39.820)

proficient at the level of a computer science undergrad.

Lex Fridman (56:42.920)

And we had a 95% graduation rate,

Sebastian Thrun (56:45.220)

even though everything's online, it's kind of tough,

Lex Fridman (56:46.900)

but we kind of trying to figure out

Lex Fridman (56:48.260)

how to make this effective.

Lex Fridman (56:50.120)

The vision is very, very simple.

Sebastian Thrun (56:52.540)

The vision is education ought to be a basic human right.

Lex Fridman (56:58.340)

It cannot be locked up behind ivory tower walls

Sebastian Thrun (57:02.320)

only for the rich people, for the parents

Lex Fridman (57:04.420)

who might be bribe themselves into the system.

Lex Fridman (57:06.780)

And only for young people and only for people

Lex Fridman (57:09.260)

from the right demographics and the right geography

Lex Fridman (57:11.740)

and possibly even the right race.

Lex Fridman (57:14.260)

It has to be opened up to everybody.

Sebastian Thrun (57:15.860)

If we are truthful to the human mission,

Lex Fridman (57:18.740)

if we are truthful to our values,

Sebastian Thrun (57:20.660)

we're gonna open up education to everybody in the world.

Lex Fridman (57:23.460)

So Udacity's pledge of 100,000 scholarships,

Sebastian Thrun (57:27.220)

I think is the biggest pledge of scholarships ever

Lex Fridman (57:29.220)

in terms of numbers.

Lex Fridman (57:30.760)

And we're working, as I said, with the White House

Lex Fridman (57:33.020)

and with very accomplished CEOs like Tim Cook

Sebastian Thrun (57:36.100)

from Apple and others to really bring education

Lex Fridman (57:39.020)

to everywhere in the world.

Sebastian Thrun (57:40.980)

Not to ask you to pick the favorite of your children,

Lex Fridman (57:44.620)

but at this point.

Sebastian Thrun (57:45.580)

Oh, that's Jasper.

Lex Fridman (57:46.600)

I only have one that I know of.

Sebastian Thrun (57:49.740)

Okay, good.

Lex Fridman (57:52.700)

In this particular moment, what nano degree,

Lex Fridman (57:55.820)

what set of courses are you most excited about at Udacity

Lex Fridman (58:00.060)

or is that too impossible to pick?

Sebastian Thrun (58:02.020)

I've been super excited about something

Lex Fridman (58:03.820)

we haven't launched yet in the building,

Sebastian Thrun (58:05.500)

which is when we talk to our partner companies,

Lex Fridman (58:09.100)

we have now a very strong footing in the enterprise world.

Lex Fridman (58:12.700)

And also to our students,

Lex Fridman (58:14.580)

we've kind of always focused on these hard skills,

Sebastian Thrun (58:17.260)

like the programming skills or math skills

Lex Fridman (58:19.740)

or building skills or design skills.

Lex Fridman (58:22.180)

And a very common ask is soft skills.

Lex Fridman (58:25.180)

Like how do you behave in your work?

Lex Fridman (58:26.860)

How do you develop empathy?

Lex Fridman (58:28.280)

How do you work on a team?

Lex Fridman (58:30.460)

What are the very basics of management?

Lex Fridman (58:32.380)

How do you do time management?

Lex Fridman (58:33.700)

How do you advance your career

Lex Fridman (58:36.240)

in the context of a broader community?

Lex Fridman (58:39.260)

And that's something that we haven't done very well

Lex Fridman (58:41.740)

at Udacity and I would say most universities

Sebastian Thrun (58:43.860)

are doing very poorly as well

Lex Fridman (58:45.180)

because we are so obsessed with individual test scores

Lex Fridman (58:47.900)

and pays a little attention to teamwork in education.

Lex Fridman (58:52.620)

So that's something I see us moving into as a company

Sebastian Thrun (58:55.500)

because I'm excited about this.

Lex Fridman (58:56.940)

And I think, look, we can teach people tech skills

Lex Fridman (59:00.100)

and they're gonna be great.

Lex Fridman (59:00.940)

But if you teach people empathy,

Sebastian Thrun (59:02.700)

that's gonna have the same impact.

Lex Fridman (59:04.960)

Maybe harder than self driving cars, but.

Sebastian Thrun (59:08.100)

I don't think so.

Lex Fridman (59:08.940)

I think the rules are really simple.

Sebastian Thrun (59:11.300)

You just have to, you have to want to engage.

Lex Fridman (59:14.380)

It's, we literally went in school and in K through 12,

Sebastian Thrun (59:18.180)

we teach kids like get the highest math score.

Lex Fridman (59:20.460)

And if you are a rational human being,

Sebastian Thrun (59:22.900)

you might evolve from this education say,

Lex Fridman (59:25.620)

having the best math score and the best English scores

Sebastian Thrun (59:28.060)

make me the best leader.

Lex Fridman (59:29.640)

And it turns out not to be that case.

Sebastian Thrun (59:31.060)

It's actually really wrong because making the,

Lex Fridman (59:34.340)

first of all, in terms of math scores,

Sebastian Thrun (59:35.820)

I think it's perfectly fine to hire somebody

Lex Fridman (59:37.620)

with great math skills.

Sebastian Thrun (59:38.500)

You don't have to do it yourself.

Lex Fridman (59:40.620)

You can hire someone with good empathy for you.

Sebastian Thrun (59:42.740)

That's much harder,

Lex Fridman (59:43.860)

but you can always hire someone with great math skills.

Lex Fridman (59:46.340)

But we live in an affluent world

Lex Fridman (59:48.940)

where we constantly deal with other people.

Lex Fridman (59:51.000)

And that's a beauty.

Lex Fridman (59:51.880)

It's not a nuisance.

Sebastian Thrun (59:52.760)

It's a beauty.

Lex Fridman (59:53.600)

So if we somehow develop that muscle

Sebastian Thrun (59:55.940)

that we can do that well and empower others

Lex Fridman (59:59.700)

in the workplace, I think we're gonna be super successful.