Sebastian Thrun: Flying Cars, Autonomous Vehicles, and Education

00:00:00.000 | The following is a conversation with Sebastian Thrun.

00:00:03.480 | He's one of the greatest roboticists,

00:00:05.280 | computer scientists, and educators of our time.

00:00:08.120 | He led the development of the autonomous vehicles

00:00:10.600 | at Stanford that won the 2005 DARPA Grand Challenge

00:00:14.280 | and placed second in the 2007 DARPA Urban Challenge.

00:00:18.160 | He then led the Google Self-Driving Car Program

00:00:20.960 | which launched the self-driving car revolution.

00:00:24.640 | He taught the popular Stanford course

00:00:26.480 | on artificial intelligence in 2011,

00:00:29.080 | which was one of the first massive open online courses

00:00:32.440 | or MOOCs as they're commonly called.

00:00:35.040 | That experience led him to co-found Udacity,

00:00:37.560 | an online education platform.

00:00:39.840 | If you haven't taken courses on it yet,

00:00:41.600 | I highly recommend it.

00:00:43.320 | Their self-driving car program, for example, is excellent.

00:00:47.120 | He's also the CEO of Kitty Hawk,

00:00:49.880 | a company working on building flying cars

00:00:53.000 | or more technically EVTOLs,

00:00:54.960 | which stands for electric vertical takeoff

00:00:56.880 | and landing aircraft.

00:00:58.640 | He has launched several revolutions

00:01:00.640 | and inspired millions of people,

00:01:02.720 | but also, as many know, he's just a really nice guy.

00:01:06.840 | It was an honor and a pleasure to talk with him.

00:01:10.560 | This is the Artificial Intelligence Podcast.

00:01:12.800 | If you enjoy it, subscribe on YouTube,

00:01:15.160 | give it five stars on Apple Podcasts,

00:01:17.120 | follow it on Spotify, support it on Patreon,

00:01:19.800 | or simply connect with me on Twitter

00:01:21.880 | at Lex Friedman, spelled F-R-I-D-M-A-N.

00:01:25.840 | If you leave a review on Apple Podcasts

00:01:27.840 | or YouTube or Twitter, consider mentioning ideas,

00:01:30.600 | people, topics you find interesting.

00:01:32.880 | It helps guide the future of this podcast.

00:01:35.840 | But in general, I just love comments

00:01:38.200 | with kindness and thoughtfulness in them.

00:01:40.160 | This podcast is a side project for me, as many people know,

00:01:43.640 | but I still put a lot of effort into it.

00:01:45.880 | So the positive words of support

00:01:47.720 | from an amazing community, from you, really help.

00:01:52.220 | I recently started doing ads at the end of the introduction.

00:01:55.240 | I'll do one or two minutes after introducing the episode

00:01:58.120 | and never any ads in the middle

00:01:59.680 | that can break the flow of the conversation.

00:02:01.840 | I hope that works for you

00:02:03.120 | and doesn't hurt the listening experience.

00:02:05.420 | I provide timestamps for the start of the conversation

00:02:08.000 | that you can skip to, but it helps if you listen to the ad

00:02:11.480 | and support this podcast by trying out the product

00:02:13.880 | or service being advertised.

00:02:15.480 | This show is presented by Cash App,

00:02:18.800 | the number one finance app in the App Store.

00:02:21.460 | I personally use Cash App to send money to friends,

00:02:24.080 | but you can also use it to buy, sell,

00:02:25.920 | and deposit Bitcoin in just seconds.

00:02:28.240 | Cash App also has a new investing feature.

00:02:31.080 | You can buy fractions of a stock, say $1 worth,

00:02:34.480 | no matter what the stock price is.

00:02:36.600 | Brokerage services are provided by Cash App Investing,

00:02:39.520 | a subsidiary of Square and member SIPC.

00:02:42.960 | I'm excited to be working with Cash App

00:02:44.720 | to support one of my favorite organizations called FIRST,

00:02:47.920 | best known for their FIRST robotics and Lego competitions.

00:02:51.340 | They educate and inspire hundreds of thousands of students

00:02:54.680 | in over 110 countries

00:02:56.520 | and have a perfect rating on Charity Navigator,

00:02:59.040 | which means the donated money

00:03:00.720 | is used to maximum effectiveness.

00:03:03.120 | When you get Cash App from the App Store or Google Play

00:03:06.080 | and use code LEXPODCAST, you'll get $10

00:03:09.360 | and Cash App will also donate $10 to FIRST,

00:03:12.120 | which again is an organization

00:03:13.960 | that I've personally seen inspire girls and boys

00:03:16.680 | to dream of engineering a better world.

00:03:19.760 | And now, here's my conversation with Sebastian Thrun.

00:03:24.000 | You've mentioned that "The Matrix"

00:03:26.760 | may be your favorite movie.

00:03:29.020 | So let's start with a crazy philosophical question.

00:03:32.220 | Do you think we're living in a simulation?

00:03:34.860 | And in general, do you find the thought experiment

00:03:37.980 | interesting?

00:03:40.060 | - Define simulation, I would say.

00:03:42.280 | Maybe we are, maybe we are not,

00:03:43.780 | but it's completely irrelevant to the way we should act.

00:03:47.220 | Putting aside for a moment,

00:03:49.900 | the fact that it might not have any impact

00:03:52.400 | on how we should act as human beings,

00:03:54.940 | for people studying theoretical physics,

00:03:57.340 | these kinds of questions might be kind of interesting,

00:03:59.620 | looking at the universe as a information processing system.

00:04:03.820 | - Universe is an information processing system.

00:04:05.980 | It's a huge physical, biological, chemical computer.

00:04:08.500 | There's no question.

00:04:09.600 | But I live here and now.

00:04:12.940 | I care about people, I care about us.

00:04:15.660 | What do you think is trying to compute?

00:04:17.700 | - I don't think there's an intention.

00:04:18.860 | I think it's just the world evolves the way it evolves,

00:04:22.060 | and it's beautiful, it's unpredictable,

00:04:25.380 | and I'm really, really grateful to be alive.

00:04:28.060 | - Spoken like a true human.

00:04:29.520 | - Which last time I checked I was.

00:04:32.140 | - Well, in fact, this whole conversation

00:04:35.300 | is just a Turing test to see if indeed you are.

00:04:39.360 | You've also said that one of the first programs

00:04:42.780 | or the first few programs you've written

00:04:45.100 | was, wait for it, TI-57 calculator.

00:04:49.160 | - Yeah.

00:04:50.040 | - Maybe that's early '80s.

00:04:52.060 | I don't wanna date calculators or anything.

00:04:54.380 | - That's early '80s, correct.

00:04:55.580 | - Yeah.

00:04:56.460 | So if you were to place yourself back into that time,

00:04:59.820 | into the mindset you were in,

00:05:02.140 | could you have predicted the evolution of computing,

00:05:05.620 | AI, the internet, technology, in the decades that followed?

00:05:10.620 | - I was super fascinated by Silicon Valley,

00:05:13.180 | which I'd seen on television once and thought,

00:05:15.260 | my God, this is so cool.

00:05:16.420 | They build like DRAMs there and CPUs.

00:05:19.620 | How cool is that?

00:05:20.460 | And as a college student a few years later,

00:05:23.980 | I decided to really study intelligence

00:05:25.860 | and study human beings and found that,

00:05:28.100 | even back then in the '80s and '90s,

00:05:30.540 | that artificial intelligence

00:05:31.580 | is what fascinated me the most.

00:05:33.340 | What's missing is that back in the day,

00:05:35.820 | the computers are really small.

00:05:37.700 | They're like the brains you could build

00:05:39.100 | were not anywhere bigger than a cockroach,

00:05:41.580 | and cockroaches aren't very smart.

00:05:43.740 | So we weren't at the scale yet where we are today.

00:05:46.300 | - Did you dream at that time

00:05:48.700 | to achieve the kind of scale we have today?

00:05:51.060 | Or did that seem possible?

00:05:52.660 | - I always wanted to make robots smart.

00:05:54.340 | I felt it was super cool to build an artificial human.

00:05:57.940 | And the best way to build an artificial human

00:05:59.660 | would have been a robot,

00:06:00.660 | because that's kind of the closest we could do.

00:06:03.060 | Unfortunately, we aren't there yet.

00:06:04.900 | The robots today are still very brittle.

00:06:07.180 | But it's fascinating to study intelligence

00:06:09.420 | from a constructive perspective

00:06:10.700 | where you build something.

00:06:12.020 | - To understand you build,

00:06:15.140 | what do you think it takes to build an intelligent system

00:06:19.500 | and an intelligent robot?

00:06:20.900 | - I think the biggest innovation

00:06:22.140 | that we've seen is machine learning.

00:06:23.780 | And it's the idea that the computers

00:06:26.140 | can basically teach themselves.

00:06:27.700 | Let's give an example.

00:06:29.740 | I'd say everybody pretty much knows how to walk.

00:06:33.100 | And we learn how to walk in the first year or two

00:06:35.340 | of our lives.

00:06:36.860 | But no scientist has ever been able

00:06:38.820 | to write down the rules of human gait.

00:06:41.180 | We don't understand it.

00:06:42.180 | We have it in our brains somehow.

00:06:44.060 | We can practice it.

00:06:45.220 | We understand it.

00:06:46.620 | But we can't articulate it.

00:06:47.820 | We can't pass it on by language.

00:06:50.300 | And that to me is kind of the deficiency

00:06:52.140 | of today's computer programming.

00:06:53.380 | When you program a computer,

00:06:55.300 | they're so insanely dumb

00:06:56.780 | that you have to give them rules for every contingencies.

00:06:59.900 | Very unlike the way people learn

00:07:01.780 | but learn from data and experience,

00:07:03.500 | computers are being instructed.

00:07:05.500 | And because it's so hard to get this instruction set right,

00:07:07.900 | you pay software engineers $200,000 a year.

00:07:11.500 | Now, the most recent innovation,

00:07:13.140 | which has been to make for like 30, 40 years,

00:07:15.860 | is an idea that computers can find their own rules.

00:07:18.500 | So they can learn from falling down and getting up

00:07:20.660 | the same way children can learn

00:07:21.980 | from falling down and getting up.

00:07:23.820 | And that revolution has led to a capability

00:07:26.220 | that's completely unmatched.

00:07:28.700 | Today's computers can watch experts do their jobs,

00:07:31.740 | whether you're a doctor or a lawyer,

00:07:33.860 | pick up the regularities, learn those rules,

00:07:36.900 | and then become as good as the best experts.

00:07:39.380 | - So the dream of in the '80s of expert systems,

00:07:41.700 | for example, had at its core the idea

00:07:44.820 | that humans could boil down their expertise

00:07:47.820 | on a sheet of paper.

00:07:49.340 | So to sort of reduce, sort of be able to explain to machines

00:07:53.220 | how to do something explicitly.

00:07:55.460 | So do you think, what's the use of human expertise

00:07:59.100 | into this whole picture?

00:08:00.060 | Do you think most of the intelligence

00:08:01.740 | will come from machines learning from experience

00:08:04.260 | without human expertise input?

00:08:06.460 | - So the question for me is much more

00:08:08.380 | how do you express expertise?

00:08:10.700 | You can express expertise by writing a book.

00:08:12.940 | You can express expertise by showing someone

00:08:15.140 | what you're doing.

00:08:16.260 | You can express expertise by applying it

00:08:18.380 | by many different ways.

00:08:20.020 | And I think the expert systems was our best attempt in AI

00:08:23.700 | to capture expertise and rules.

00:08:25.980 | But someone sat down and say,

00:08:27.100 | "Here are the rules of human gait.

00:08:28.540 | "Here's when you put your big toe forward

00:08:31.260 | "and your heel backwards and here how you stop stumbling."

00:08:34.700 | And as we now know, the set of rules,

00:08:37.100 | the set of language that we can command

00:08:39.420 | is incredibly limited.

00:08:41.180 | The majority of the human brain doesn't deal with language.

00:08:43.780 | It deals with like subconscious numerical perceptual things

00:08:48.180 | that we don't even, that we're self aware of.

00:08:50.420 | Now, when a AI system watches an expert do their job

00:08:55.740 | and practice their job,

00:08:57.900 | it can pick up things that people can't even put into

00:09:01.140 | writing into books or rules.

00:09:03.020 | And that's where the real power is.

00:09:04.500 | We now have AI systems that, for example,

00:09:07.140 | look over the shoulders of highly paid human doctors

00:09:10.540 | like dermatologists or radiologists,

00:09:12.820 | and they can somehow pick up those skills

00:09:15.340 | that no one can express in words.

00:09:16.980 | - So you were a key person in launching three revolutions,

00:09:22.140 | online education, autonomous vehicles,

00:09:25.140 | and flying cars or VTOLs.

00:09:28.180 | So high level,

00:09:30.220 | and I apologize for all the philosophical questions.

00:09:35.020 | - No apology necessary.

00:09:36.180 | - How do you choose what problems to try and solve?

00:09:40.540 | What drives you to make those solutions a reality?

00:09:43.340 | - I have two desires in life.

00:09:44.820 | I wanna literally make the lives of others better.

00:09:48.500 | Or as we often say,

00:09:50.460 | maybe jokingly, make the world a better place.

00:09:52.900 | I actually believe in this.

00:09:54.940 | It's as funny as it sounds.

00:09:56.340 | And second, I wanna learn.

00:09:59.100 | I wanna get in the skillset.

00:10:00.420 | I don't wanna be in a job I'm good at,

00:10:01.820 | because if I'm in a job that I'm good at,

00:10:04.260 | the chances for me to learn something interesting

00:10:05.820 | is actually minimized.

00:10:06.660 | So I wanna be in a job I'm bad at.

00:10:08.340 | That's really important to me.

00:10:10.180 | So in a build, for example,

00:10:11.540 | what people often call flying cars,

00:10:12.980 | these are electrical, vertical,

00:10:14.860 | takeoff and landing vehicles.

00:10:16.500 | I'm just no expert in any of this.

00:10:19.660 | And it's so much fun to learn on the job

00:10:22.220 | what it actually means to build something like this.

00:10:24.860 | Now I'd say the stuff that I've done lately

00:10:27.500 | after I finished my professorship at Stanford,

00:10:31.060 | they really focused on like what has the maximum impact

00:10:33.740 | on society.

00:10:34.820 | Like transportation is something that has transformed

00:10:37.500 | the 21st or 20th century more than any other invention,

00:10:40.100 | in my opinion, even more than communication.

00:10:42.500 | And cities are different, workers different,

00:10:45.020 | women's rights are different because of transportation.

00:10:47.860 | And yet we still have a very suboptimal

00:10:50.740 | transportation solution where we kill 1.2

00:10:54.660 | or so million people every year in traffic.

00:10:57.620 | It's like the leading cause of death

00:10:58.860 | for young people in many countries,

00:11:01.100 | where we are extremely inefficient resource wise,

00:11:03.580 | just go to your average neighborhood city

00:11:06.780 | and look at the number of parked cars,

00:11:08.300 | that's a travesty in my opinion,

00:11:10.380 | or where we spend endless hours in traffic jams.

00:11:13.820 | And very, very simple innovations like a self-driving car

00:11:16.820 | or what people call a flying car

00:11:18.820 | could completely change this.

00:11:20.220 | And it's there, I mean, the technology is basically there.

00:11:23.260 | You have to close your eyes not to see it.

00:11:25.380 | - So lingering on autonomous vehicles,

00:11:29.380 | a fascinating space, some incredible work

00:11:32.060 | you've done throughout your career there.

00:11:33.580 | So let's start with DARPA, I think,

00:11:38.340 | the DARPA challenge through the desert

00:11:40.340 | and then urban to the streets.

00:11:42.820 | I think that inspired an entire generation of roboticists

00:11:45.700 | and obviously sprung this whole excitement

00:11:49.460 | about this particular kind of four wheeled robots

00:11:52.620 | we called autonomous cars, self-driving cars.

00:11:55.500 | So you led the development of Stanley,

00:11:58.180 | the autonomous car that won the race to the desert,

00:12:01.220 | the DARPA challenge in 2005.

00:12:03.940 | And Junior, the car that finished second

00:12:07.300 | in the DARPA urban challenge,

00:12:09.140 | also did incredibly well in 2007, I think.

00:12:13.340 | What are some painful, inspiring

00:12:16.540 | or enlightening experiences from that time

00:12:19.420 | that stand out to you?

00:12:20.580 | - Oh my God, painful were all these incredibly complicated,

00:12:25.580 | stupid bugs that had to be found.

00:12:30.420 | We had a phase where Stanley,

00:12:33.020 | our car that eventually won the DARPA urban challenge

00:12:36.180 | would every 30 miles just commit suicide

00:12:39.300 | and we didn't know why.

00:12:40.860 | And it ended up to be that in the syncing

00:12:43.500 | of two computer clocks, occasionally a clock went backwards

00:12:47.700 | and that negative time elapsed,

00:12:50.340 | screwed up the entire internal logic,

00:12:51.820 | but it took ages to find this.

00:12:54.380 | It were like bugs like that.

00:12:56.300 | I'd say enlightening is the Stanford team

00:12:59.300 | immediately focused on machine learning and on software,

00:13:02.460 | whereas everybody else seemed to focus

00:13:03.740 | on building better hardware.

00:13:05.180 | Our analysis had been a human being

00:13:07.620 | with an existing rental car can perfectly drive the course.

00:13:10.220 | Why do I have to build a better rental car?

00:13:12.100 | I just should replace the human being.

00:13:14.780 | And the human being to me was a conjunction of three steps.

00:13:18.780 | We had sensors, eyes and ears, mostly eyes.

00:13:22.300 | We had brains in the middle

00:13:23.740 | and then we had actuators, our hands and our feet.

00:13:26.260 | Now the actuators are easy to build.

00:13:28.140 | The sensors are actually also easy to build.

00:13:29.660 | What was missing was the brain.

00:13:30.660 | So we had to build a human brain

00:13:32.540 | and nothing clear then to me

00:13:35.180 | that the human brain is a learning machine.

00:13:36.940 | So why not just train our robot?

00:13:38.180 | So we would build massive machine learning into our machine.

00:13:42.260 | And with that, we're able to not just learn

00:13:44.780 | from human drivers, we had the entire speed control

00:13:47.340 | of the vehicle was copied from human driving,

00:13:49.780 | but also have the robot learn from experience

00:13:51.620 | where it made a mistake and go to recover from it

00:13:53.620 | and learn from it.

00:13:55.580 | - You mentioned the pain point of software and clocks.

00:14:00.580 | Synchronization seems to be a problem

00:14:04.580 | that continues with robotics.

00:14:06.020 | It's a tricky one with drones and so on.

00:14:08.020 | What does it take to build a thing,

00:14:13.380 | a system with so many constraints?

00:14:16.620 | You have a deadline, no time.

00:14:20.300 | You're unsure about anything really.

00:14:22.100 | It's the first time that people really even explore.

00:14:24.980 | It's not even sure that anybody can finish

00:14:26.780 | when we're talking about the race

00:14:28.340 | of the desert the year before nobody finished.

00:14:30.620 | What does it take to scramble and finish a product

00:14:33.420 | that actually, a system that actually works?

00:14:35.780 | - I mean, we were very lucky.

00:14:36.780 | We were a really small team.

00:14:38.300 | The core of the team were four people.

00:14:40.460 | It was four because five couldn't comfortably sit

00:14:43.100 | inside a car, but four could.

00:14:45.300 | And I, as a team leader, my job was to get pizza

00:14:47.700 | for everybody and wash the car and stuff like this

00:14:51.180 | and repair the radiator when it broke

00:14:52.860 | and debug the system.

00:14:55.260 | And we were very kind of open-minded.

00:14:56.900 | We had like no egos involved in this.

00:14:58.420 | We just wanted to see how far we can get.

00:15:00.820 | What we did really, really well was time management.

00:15:03.300 | We were done with everything a month before the race.

00:15:06.220 | And we froze the entire software a month before the race.

00:15:08.740 | And it turned out, looking at other teams,

00:15:11.420 | every other team complained if they had just one more week,

00:15:14.100 | they would have won.

00:15:15.420 | And we decided, we're not gonna fall into that mistake.

00:15:18.740 | We're gonna be early.

00:15:19.860 | And we had an entire month to shake the system.

00:15:22.700 | And we actually found two or three minor bugs

00:15:24.940 | in the last month that we had to fix.

00:15:27.060 | And we were completely prepared when the race occurred.

00:15:30.020 | - Okay, so first of all,

00:15:30.860 | that's such an incredibly rare achievement

00:15:34.620 | in terms of being able to be done on time or ahead of time.

00:15:38.980 | What do you, how do you do that in your future work?

00:15:43.060 | What advice do you have in general?

00:15:44.780 | Because it seems to be so rare,

00:15:46.340 | especially in highly innovative projects like this.

00:15:49.300 | People work till the last second.

00:15:50.860 | - Well, the nice thing about the Dark Background Challenge

00:15:52.540 | is that the problem was incredibly well-defined.

00:15:55.300 | We were able for a while

00:15:56.540 | to drive the old Dark Background Challenge course,

00:15:58.780 | which had been used the year before.

00:16:00.780 | And then at some reason, we were kicked out of the region.

00:16:04.020 | So we had to go to different deserts, the Snorren Desert,

00:16:06.300 | and we were able to drive desert trails

00:16:08.860 | just at the same time.

00:16:10.580 | So there was never any debate about,

00:16:12.180 | like, what is actually the problem?

00:16:13.220 | We didn't sit down and say,

00:16:14.340 | "Hey, should we build a car or a plane?"

00:16:16.660 | We had to build a car.

00:16:18.260 | That made it very, very easy.

00:16:20.380 | Then I studied my own life and life of others

00:16:23.780 | and realized that the typical mistake that people make

00:16:26.340 | is that there's this kind of crazy bug left

00:16:29.580 | that they haven't found yet.

00:16:31.220 | And it's just, they regret it.

00:16:34.340 | And the bug would have been trivial to fix.

00:16:36.140 | They just haven't fixed it yet.

00:16:37.740 | They didn't want to fall into that trap.

00:16:39.580 | So I built a testing team.

00:16:41.060 | We had a testing team that built a testing booklet

00:16:43.740 | of 160 pages of tests we had to go through

00:16:46.740 | just to make sure we shake out the system appropriately.

00:16:49.100 | - Wow.

00:16:49.940 | - And the testing team was with us all the time

00:16:51.820 | and dictated to us,

00:16:53.020 | today we do railroad crossings,

00:16:55.540 | tomorrow we do, we practice the start of the event.

00:16:58.460 | And in all of these, we thought,

00:17:00.660 | "Oh my God, this long solve trivial."

00:17:02.260 | And then we tested it out.

00:17:03.180 | "Oh my God, it doesn't do a railroad crossing, why not?

00:17:05.100 | "Oh my God, it mistakes the rails for metal barriers.

00:17:09.340 | "Shit, we have to fix this."

00:17:10.620 | - Yes.

00:17:11.620 | - So it was really a continuous focus

00:17:14.460 | on improving the weakest part of the system.

00:17:16.380 | And as long as you focus

00:17:18.540 | on improving the weakest part of the system,

00:17:20.540 | you eventually build a really great system.

00:17:23.100 | - Let me just pause in that.

00:17:24.780 | To me as an engineer, it's just super exciting

00:17:27.100 | that you were thinking like that,

00:17:28.300 | especially at that stage as brilliant

00:17:30.460 | that testing was such a core part of it.

00:17:33.380 | It may be to linger on the point of leadership.

00:17:35.740 | I think it's one of the first times you were really a leader

00:17:41.700 | and you've led many very successful teams since then.

00:17:45.420 | What does it take to be a good leader?

00:17:48.460 | - I would say most of all, just take credit

00:17:50.820 | for the work of others.

00:17:54.220 | That's very convenient, turns out,

00:17:57.580 | 'cause I can't do all these things myself.

00:18:00.220 | I'm an engineer at heart, so I care about engineering.

00:18:03.740 | So I don't know what the chicken and the egg is,

00:18:06.140 | but as a kid, I loved computers

00:18:07.860 | because you could tell them to do something

00:18:09.540 | and they actually did it.

00:18:10.700 | It was very cool.

00:18:11.540 | And you could like in the middle of the night,

00:18:12.740 | wake up at one in the morning and switch on your computer.

00:18:15.180 | And what you told you to yesterday, it would still do.

00:18:18.140 | That was really cool.

00:18:19.380 | Unfortunately, that didn't quite work with people.

00:18:21.300 | So you go to people and tell them what to do

00:18:22.820 | and they don't do it and they hate you for it.

00:18:25.780 | Or you do it today and then they go a day later

00:18:28.060 | and they stop doing it.

00:18:29.140 | So you have to...

00:18:30.220 | So then the question really became,

00:18:31.460 | how can you put yourself in the brain of people

00:18:34.100 | as opposed to computers?

00:18:35.100 | And in terms of computers, that's super dumb.

00:18:37.340 | That's so dumb.

00:18:38.220 | If people were as dumb as computers,

00:18:39.620 | I wouldn't wanna work with them.

00:18:41.260 | But people are smart and people are emotional

00:18:43.580 | and people have pride and people have aspirations.

00:18:45.900 | So how can I connect to that?

00:18:48.860 | And that's the thing where most of our leadership just fails

00:18:52.540 | because many, many engineers turn manager

00:18:56.220 | believe they can treat their team just the same

00:18:58.340 | way they can treat your computer

00:18:59.300 | and it just doesn't work this way.

00:19:00.420 | It's just really bad.

00:19:02.300 | So how can I connect to people?

00:19:05.100 | And it turns out as a college professor,

00:19:07.620 | the wonderful thing you do all the time

00:19:09.940 | is to empower other people.

00:19:10.980 | Like your job is to make your students look great.

00:19:14.700 | That's all you do.

00:19:15.540 | You're the best coach.

00:19:16.900 | And it turns out if you do a fantastic job

00:19:18.740 | with making your students look great,

00:19:20.820 | they actually love you and their parents love you.

00:19:22.700 | And they give you all the credit

00:19:23.780 | for stuff you don't deserve.

00:19:24.700 | Turns out all my students were smarter than me.

00:19:27.180 | All the great stuff invented at Stanford

00:19:28.740 | was their stuff, not my stuff.

00:19:30.020 | And they give me credit and say,

00:19:31.140 | oh, Sebastian, we're just making them

00:19:33.620 | feel good about themselves.

00:19:35.100 | So the question really is, can you take a team of people

00:19:38.020 | and what does it take to make them,

00:19:40.380 | to connect to what they actually want in life

00:19:43.340 | and turn this into productive action?

00:19:45.740 | It turns out every human being that I know

00:19:48.500 | has incredibly good intentions.

00:19:50.100 | I've really rarely met a person with bad intentions.

00:19:53.220 | I believe every person wants to contribute.

00:19:55.900 | I think every person I've met wants to help others.

00:19:59.420 | It's amazing how much of a urge we have

00:20:01.860 | not to just help ourselves, but to help others.

00:20:04.420 | So how can we empower people

00:20:05.940 | and give them the right framework

00:20:07.700 | that they can accomplish this?

00:20:09.340 | In moments when it works, it's magical

00:20:12.420 | because you'd see the confluence of people

00:20:17.180 | being able to make the world a better place

00:20:19.180 | and just having enormous confidence and pride out of this.

00:20:22.900 | And that's when my environment works the best.

00:20:27.220 | These are moments where I can disappear for a month

00:20:29.460 | and come back and things still work.

00:20:31.620 | It's very hard to accomplish,

00:20:32.780 | but when it works, it's amazing.

00:20:35.100 | - So I agree with you very much.

00:20:37.260 | It's not often heard that most people in the world

00:20:42.020 | have good intentions.

00:20:43.500 | At the core, their intentions are good

00:20:45.900 | and they're good people.

00:20:47.380 | That's a beautiful message.

00:20:48.860 | It's not often heard.

00:20:50.180 | - We make this mistake,

00:20:51.180 | and this is a friend of mine, Alex Voda,

00:20:53.460 | talking this, that we judge ourselves by our intentions

00:20:57.700 | and others by their actions.

00:20:59.140 | And I think that the biggest skill,

00:21:01.860 | I mean, here in Silicon Valley, we're full of engineers

00:21:03.540 | who have very little empathy

00:21:05.140 | and are kind of befuddled by why it doesn't work for them.

00:21:09.180 | The biggest skill I think that people should acquire

00:21:13.060 | is to put themselves into the position of the other

00:21:16.860 | and listen, and listen to what the other has to say.

00:21:19.980 | And they'd be shocked how similar they are to themselves.

00:21:23.380 | And they might even be shocked

00:21:24.580 | how their own actions don't reflect their intentions.

00:21:28.540 | I often have conversations with engineers

00:21:30.900 | where I say, "Look, hey, I love you.

00:21:32.540 | "You're doing a great job.

00:21:33.380 | "And by the way, what you just did has the following effect.

00:21:37.220 | "Are you aware of that?"

00:21:38.820 | And then people would say, "Oh my God, not I wasn't

00:21:41.180 | "because my intention was that."

00:21:43.100 | And I say, "Yeah, I trust your intention.

00:21:44.900 | "You're a good human being.

00:21:46.260 | "But just to help you in the future,

00:21:48.340 | "if you keep expressing it that way,

00:21:51.220 | "then people will just hate you."

00:21:53.380 | And I've had many instances where people say,

00:21:55.140 | "Oh my God, thank you for telling me this

00:21:56.460 | "because it wasn't my intention to look like an idiot.

00:21:59.180 | "It wasn't my intention to help other people.

00:22:00.580 | "I just didn't know how to do it."

00:22:02.460 | - Very simple, by the way.

00:22:03.980 | There's a book, Dale Carnegie, 1936,

00:22:07.420 | "How to Make Friends and How to Influence Others."

00:22:10.460 | Has the entire Bible.

00:22:11.500 | You just read it and you're done

00:22:12.780 | and you apply it every day.

00:22:14.020 | And I wish I was good enough to apply it every day.

00:22:16.820 | But it's just simple things, right?

00:22:18.940 | Like be positive, remember people's names, smile,

00:22:22.620 | and eventually have empathy.

00:22:24.540 | Really think that the person that you hate

00:22:27.460 | and you think is an idiot is actually just like yourself.

00:22:30.500 | It's a person who's struggling, who means well,

00:22:33.260 | and who might need help.

00:22:34.220 | And guess what?

00:22:35.060 | You need help.

00:22:36.620 | I've recently spoken with Stephen Schwarzman.

00:22:39.980 | I'm not sure if you know who that is.

00:22:41.660 | - I do.

00:22:42.500 | It's on my list.

00:22:45.220 | (laughing)

00:22:46.040 | - On the list.

00:22:46.880 | (laughing)

00:22:47.720 | But he said, sort of to expand on what you're saying,

00:22:52.720 | that one of the biggest things you can do

00:22:56.020 | is hear people when they tell you what their problem is

00:23:00.060 | and then help them with that problem.

00:23:02.380 | He says it's surprising how few people

00:23:06.020 | actually listen to what troubles others.

00:23:08.700 | - Yeah.

00:23:09.540 | - And because it's right there in front of you

00:23:12.620 | and you can benefit the world the most.

00:23:15.220 | And in fact, yourself and everybody around you

00:23:18.020 | by just hearing the problems and solving them.

00:23:20.860 | - I mean, that's my little history of engineering.

00:23:23.980 | That is, while I was engineering with computers,

00:23:28.260 | I didn't care at all what the computer's problems were.

00:23:32.380 | I just told them what to do and they do it.

00:23:34.820 | And it just doesn't work this way with people.

00:23:37.580 | - It doesn't work with me.

00:23:38.500 | If you come to me and say, "Do A," I do the opposite.

00:23:41.260 | (laughing)

00:23:43.620 | - But let's return to the comfortable world of engineering.

00:23:47.140 | And can you tell me in broad strokes in how you see it?

00:23:52.140 | Because you're at the core of starting it,

00:23:53.860 | the core of driving it,

00:23:55.100 | the technical evolution of autonomous vehicles

00:23:58.020 | from the first DARPA Grand Challenge

00:24:00.420 | to the incredible success we see with the program

00:24:03.660 | you started with Google Self-Driving Car

00:24:05.420 | and Waymo and the entire industry that sprung up

00:24:08.340 | of different kinds of approaches, debates, and so on.

00:24:11.220 | - Well, the idea of self-driving car goes back to the '80s.

00:24:14.180 | There was a team in Germany, another team in Carnegie Mellon

00:24:16.140 | that did some very pioneering work.

00:24:18.700 | But back in the day, I'd say the computers were so deficient

00:24:21.780 | that even the best professors and engineers in the world

00:24:25.900 | basically stood no chance.

00:24:27.300 | It then folded into a phase where the US government

00:24:31.220 | spent at least half a billion dollars

00:24:33.340 | that I could count on research projects.

00:24:36.180 | But the way the procurement works,

00:24:37.980 | a successful stack of paper describing lots of stuff

00:24:42.820 | that no one's ever gonna read

00:24:43.900 | was a successful product of a research project.

00:24:47.660 | So we trained our researchers to produce lots of paper.

00:24:50.540 | That all changed with the DARPA Grand Challenge.

00:24:54.340 | And I really gotta credit the ingenious people at DARPA

00:24:58.500 | and the US government and Congress

00:25:00.420 | that took a complete new funding model

00:25:02.220 | where they said, "Let's not fund effort,

00:25:04.100 | "let's fund outcomes."

00:25:05.580 | And it sounds very trivial, but there was no tax code

00:25:08.820 | that allowed the use of congressional tax money for a price.

00:25:13.700 | It was all effort-based.

00:25:15.100 | So if you put in 100 hours in, you could charge 100 hours.

00:25:17.460 | If you put in 1,000 hours in, you could build 1,000 hours.

00:25:20.660 | By changing the focus and say, "We're making the price.

00:25:22.780 | "We don't pay you for development,

00:25:23.940 | "we pay you for the accomplishment."

00:25:26.340 | They drew in, they automatically drew out

00:25:28.940 | all these contractors who are used to the drug

00:25:31.700 | of getting money per hour.

00:25:33.380 | And they drew in a whole bunch of new people.

00:25:35.700 | And these people are mostly crazy people.

00:25:37.620 | They were people who had a car and a computer

00:25:40.740 | and they wanted to make a million bucks.

00:25:42.420 | The million bucks was the original price money,

00:25:43.900 | it was then doubled.

00:25:45.460 | And they felt, "If I put my computer in my car

00:25:47.980 | "and program it, I can be rich."

00:25:50.860 | And that was so awesome.

00:25:52.060 | Like half the teams, there was a team that was surfer dudes

00:25:55.500 | and they had like two surfboards on their vehicle

00:25:58.540 | and brought like these fashion girls,

00:26:00.460 | super cute girls, like twin sisters.

00:26:02.820 | And you could tell these guys were not your common

00:26:06.700 | fails babe bandit who gets all these big multi-million

00:26:10.860 | and billion dollar countries from the US government.

00:26:13.540 | And there was a great reset.

00:26:15.260 | Universities moved in.

00:26:18.580 | I was very fortunate at Stanford that I'd just received tenure

00:26:21.820 | so I couldn't get fired no matter what I do,

00:26:23.380 | otherwise I wouldn't have done it.

00:26:25.140 | And I had enough money to finance this thing.

00:26:28.260 | And I was able to attract a lot of money from third parties.

00:26:31.180 | And even car companies moved in.

00:26:32.540 | They kind of moved in very quietly

00:26:34.060 | because they were super scared to be embarrassed

00:26:36.580 | that their car would flip over.

00:26:38.580 | But Ford was there and Volkswagen was there

00:26:40.700 | and a few others and GM was there.

00:26:43.380 | So it kind of reset the entire landscape of people.

00:26:46.340 | And if you look at who's a big name

00:26:48.220 | in self-driving cars today,

00:26:49.500 | these were mostly people who participated

00:26:51.340 | in those challenges.

00:26:52.300 | - Okay, that's incredible.

00:26:54.260 | Can you just comment quickly on your sense of lessons learned

00:26:59.060 | from that kind of funding model

00:27:01.220 | and the research that's going on in academia

00:27:04.380 | in terms of producing papers?

00:27:06.100 | Is there something to be learned and scaled up bigger?

00:27:09.860 | These having these kinds of grand challenges

00:27:11.700 | that could improve outcomes?

00:27:14.540 | - So I'm a big believer in focusing

00:27:16.300 | on kind of an end-to-end system.

00:27:19.620 | I'm a really big believer in systems building.

00:27:21.900 | I've always built systems in my academic career,

00:27:23.660 | even though I do a lot of math and abstract stuff,

00:27:27.020 | but it's all derived from the idea

00:27:28.100 | of let's solve a real problem.

00:27:29.620 | And it's very hard for me to be an academic

00:27:33.780 | and say, let me solve a component of a problem.

00:27:35.740 | Like with someone, there's fields like non-monetary logic

00:27:38.620 | or AI planning systems where people believe

00:27:41.740 | that a certain style of problem solving

00:27:44.260 | is the ultimate end objective.

00:27:47.220 | And I would always turn it around and say,

00:27:49.500 | hey, what problem would my grandmother care about

00:27:52.580 | that doesn't understand computer technology

00:27:54.620 | and doesn't want to understand?

00:27:56.460 | And how could I make her love what I do?

00:27:58.460 | Because only then do I have an impact on the world.

00:28:01.300 | I can easily impress my colleagues.

00:28:02.900 | That is much easier,

00:28:04.700 | but impressing my grandmother is very, very hard.

00:28:07.580 | So I would always thought if I can build a self-driving car

00:28:10.700 | and my grandmother can use it

00:28:12.820 | even after she loses her driving privileges

00:28:14.660 | or children can use it,

00:28:16.100 | or we save maybe a million lives a year,

00:28:20.500 | that would be very impressive.

00:28:22.420 | And there's so many problems like these.

00:28:23.900 | Like there's a problem with curing cancer,

00:28:25.300 | or whatever it is, live twice as long.

00:28:27.780 | Once a problem is defined,

00:28:29.580 | of course I can't solve it in its entirety.

00:28:31.420 | Like it takes sometimes tens of thousands of people

00:28:34.180 | to find a solution.

00:28:35.340 | There's no way you can fund an army of 10,000 at Stanford.

00:28:39.340 | So you're gonna build a prototype.

00:28:41.060 | Let's build a meaningful prototype.

00:28:42.500 | And the Dark Background Challenge was beautiful

00:28:43.900 | because it told me what this prototype had to do.

00:28:46.340 | I didn't have to think about what it had to do.

00:28:47.660 | I just had to read the rules.

00:28:48.820 | And it was really, really beautiful.

00:28:51.060 | - And it's most beautiful, you think,

00:28:52.820 | what academia could aspire to is to build a prototype

00:28:56.220 | that's the systems level that solves,

00:29:00.260 | gives you an inkling that this problem

00:29:02.140 | could be solved with this prototype.

00:29:03.540 | - First of all, I wanna emphasize what academia really is.

00:29:06.540 | And I think people misunderstand it.

00:29:08.580 | First and foremost,

00:29:10.340 | academia is a way to educate young people.

00:29:13.340 | First and foremost, a professor is an educator.

00:29:15.420 | No matter where you are at,

00:29:17.060 | a small suburban college,

00:29:18.580 | or whether you are a Harvard or Stanford professor.

00:29:21.980 | That's not the way most people think of themselves

00:29:25.020 | in academia because we have this kind of competition

00:29:28.020 | going on for citations and publication.

00:29:31.460 | That's a measurable thing,

00:29:32.860 | but that is secondary to the primary purpose

00:29:35.500 | of educating people to think.

00:29:37.820 | Now, in terms of research,

00:29:40.020 | most of the great science,

00:29:42.900 | the great research comes out of universities.

00:29:45.540 | You can trace almost everything back,

00:29:46.980 | including Google to universities.

00:29:48.860 | So there's nothing really fundamentally broken here.

00:29:52.180 | It's a good system.

00:29:53.420 | And I think America has the finest university system

00:29:55.980 | on the planet.

00:29:56.820 | We can talk about reach and how to reach people

00:30:00.620 | outside the system.

00:30:01.500 | It's a different topic,

00:30:02.340 | but the system itself is a good system.

00:30:04.820 | If I had one wish, I would say,

00:30:07.020 | it'd be really great if there was more debate

00:30:10.860 | about what the great big problems are in society.

00:30:15.940 | And focus on those.

00:30:18.780 | And most of them are interdisciplinary.

00:30:21.620 | Unfortunately, it's very easy to fall

00:30:24.660 | into a interdisciplinary viewpoint

00:30:28.180 | where your problem is dictated

00:30:30.460 | by what your closest colleagues believe the problem is.

00:30:33.700 | It's very hard to break out and say,

00:30:35.300 | well, there's an entire new field of problems.

00:30:37.940 | So to give an example,

00:30:39.860 | prior to me working on self-driving cars,

00:30:41.660 | I was a roboticist and a machine learning expert.

00:30:44.660 | And I wrote books on robotics,

00:30:46.860 | something called probabilistic robotics.

00:30:48.540 | It's a very methods driven kind of viewpoint of the world.

00:30:51.540 | I built robots that acted in museums as tour guides,

00:30:54.060 | that like led children around.

00:30:55.660 | It is something that at the time was moderately challenging.

00:31:00.060 | When I started working on cars,

00:31:02.260 | several colleagues told me,

00:31:03.700 | "Sebastian, you're destroying your career

00:31:06.060 | "because in our field of robotics,

00:31:08.140 | "cars are looked like as a gimmick

00:31:10.380 | "and they're not expressive enough.

00:31:11.740 | "They can only push this bottle

00:31:14.140 | "and the brakes, there's no dexterity,

00:31:16.420 | "there's no complexity, it's just too simple."

00:31:19.540 | And no one came to me and said,

00:31:21.180 | "Wow, if you solve that problem,

00:31:22.700 | "you can save a million lives."

00:31:25.020 | Among all robotic problems that I've seen in my life,

00:31:27.260 | I would say the self-driving car, transportation,

00:31:29.780 | is the one that has the most hope for society.

00:31:32.100 | So how come the robotics community

00:31:33.540 | wasn't all over the place?

00:31:35.140 | And it was because we focused on methods and solutions

00:31:37.940 | and not on problems.

00:31:39.900 | Like if you go around today and ask your grandmother,

00:31:42.460 | what bugs you, what really makes you upset?

00:31:45.260 | I challenge any academic to do this

00:31:48.780 | and then realize how far your research

00:31:51.820 | is probably away from that today.

00:31:53.860 | At the very least, that's a good thing

00:31:56.820 | for academics to deliberate on.

00:31:59.300 | The other thing that's really nice in Silicon Valley

00:32:00.900 | is Silicon Valley is full of smart people outside academia.

00:32:04.380 | So there's the Larry Pages and Mark Zuckerbergs

00:32:06.340 | in the world who are anywhere as smart or smarter

00:32:09.020 | than the best academics I've met in my life.

00:32:11.460 | And what they do is they are at a different level.

00:32:15.420 | They build the systems,

00:32:16.740 | they build the customer-facing systems,

00:32:19.340 | they build things that people can use

00:32:21.980 | without technical education.

00:32:23.820 | And they are inspired by research,

00:32:25.860 | they're inspired by scientists.

00:32:27.540 | They hire the best PhDs from the best universities

00:32:30.340 | for a reason.

00:32:31.180 | So I think there's kind of vertical integration

00:32:35.140 | between the real product, the real impact,

00:32:37.780 | and the real thought, the real ideas.

00:32:39.860 | That's actually working surprisingly well

00:32:41.300 | in Silicon Valley.

00:32:42.780 | It did not work as well in other places in this nation.

00:32:44.900 | So when I worked at Carnegie Mellon,

00:32:46.700 | we had the world's finest computer science university,

00:32:49.860 | but there wasn't those people in Pittsburgh

00:32:52.780 | that would be able to take these very fine

00:32:55.140 | computer science ideas and turn them into

00:32:57.140 | massively impactful products.

00:33:00.620 | That symbiosis seemed to exist pretty much only

00:33:03.900 | in Silicon Valley and maybe a bit in Boston and Austin.

00:33:06.620 | - Yeah, with Stanford.

00:33:07.980 | That's really interesting.

00:33:11.100 | So if we look a little bit further on

00:33:14.060 | from the DARPA Grand Challenge

00:33:17.180 | and the launch of the Google self-driving car,

00:33:20.100 | what do you see as the state,

00:33:22.060 | the challenges of autonomous vehicles as they are now?

00:33:25.900 | Is actually achieving that huge scale

00:33:29.220 | and having a huge impact on society?

00:33:31.700 | - I'm extremely proud of what has been accomplished.

00:33:35.300 | And again, I'm taking a lot of credit for the work that I do.

00:33:37.620 | (laughing)

00:33:38.460 | And I'm actually very optimistic

00:33:40.220 | and people have been kind of worrying,

00:33:42.380 | is it too fast?

00:33:43.220 | Is it too slow?

00:33:44.060 | Why is it not there yet?

00:33:44.900 | And so on.

00:33:45.900 | It is actually quite an interesting hard problem.

00:33:48.860 | And in that a self-driving car,

00:33:51.660 | to build one that manages 90% of the problems

00:33:55.340 | and count them every day driving is easy.

00:33:57.260 | We can literally do this over a weekend.

00:33:59.500 | To do 99% might take a month.

00:34:02.100 | Then there's 1% left.

00:34:03.260 | So 1% would mean that you still

00:34:05.660 | have a fatal accident every week.

00:34:08.140 | Very unacceptable.

00:34:09.020 | So now you work on this 1%

00:34:10.940 | and the 99% of that,

00:34:12.900 | the remaining 1% is actually still relatively easy.

00:34:15.780 | But now you're down to like a hundredth of 1%

00:34:18.220 | and it's still completely unacceptable in terms of safety.

00:34:21.580 | So the variety of things you encounter are just enormous.

00:34:24.260 | And that gives me enormous respect for human beings

00:34:26.500 | that we're able to deal with the couch on the highway,

00:34:30.220 | right?

00:34:31.060 | Or the deer in the headlight

00:34:31.900 | or the blown tire that we've never been trained for.

00:34:34.940 | And all of a sudden I have to handle it

00:34:35.980 | in an emergency situation

00:34:37.140 | and often do very, very successfully.

00:34:38.780 | It's amazing from that perspective

00:34:40.700 | how safe driving actually is

00:34:42.220 | given how many millions of miles we drive

00:34:44.860 | every year in this country.

00:34:46.220 | We are now at a point

00:34:48.500 | where I believe the technology is there.

00:34:50.060 | And I've seen it.

00:34:51.620 | I've seen it in Waymo.

00:34:52.540 | I've seen it in Aptiv.

00:34:53.580 | I've seen it in Cruise

00:34:55.460 | and in a number of companies and Voyage

00:34:58.300 | where vehicles not driving around

00:35:00.980 | and basically flawlessly are able to drive people around

00:35:04.260 | in limited scenarios.

00:35:06.100 | In fact, you can go to Vegas today

00:35:08.020 | and order a summon a lift.

00:35:09.940 | And if you got the right setting of your app

00:35:13.540 | you'd be picked up by a driverless car.

00:35:15.820 | Now there's still safety drivers in there

00:35:18.100 | but that's a fantastic way to kind of learn

00:35:21.340 | what the limits of technology today.

00:35:22.980 | And there's still some glitches

00:35:24.740 | but the glitches have become very, very rare.

00:35:26.580 | I think the next step is gonna be to down cost it

00:35:29.700 | to harden it.

00:35:31.300 | The entrapment, the sensors

00:35:33.780 | are not quite an automotive grade standard yet.

00:35:36.220 | And then to really build the business models

00:35:37.820 | to really kind of go somewhere and make the business case.

00:35:40.980 | And the business case is hard work.

00:35:42.580 | It's not just, oh my God, we have this capability

00:35:44.580 | people are just gonna buy it.

00:35:45.540 | You have to make it affordable.

00:35:46.740 | You have to give people the,

00:35:50.220 | find the social acceptance of people.

00:35:52.300 | None of the teams yet has been able to

00:35:54.460 | or gutsy enough to drive around

00:35:56.180 | without a person inside the car.

00:35:59.260 | And that's the next magical hurdle.

00:36:01.300 | We'll be able to send these vehicles around

00:36:03.820 | completely empty in traffic.

00:36:05.780 | And I think, I mean, I wait every day,

00:36:08.140 | wait for the news that Waymo has just done this.

00:36:10.700 | - So, you know, it's interesting you mentioned gutsy.

00:36:14.700 | Let me ask some maybe unanswerable question,

00:36:19.700 | maybe edgy questions, but in terms of

00:36:24.460 | how much risk is required, some guts

00:36:28.700 | in terms of leadership style,

00:36:30.340 | it would be good to contrast approaches.

00:36:32.580 | And I don't think anyone knows what's right.

00:36:34.660 | But if we compare Tesla and Waymo, for example,

00:36:38.540 | Elon Musk and the Waymo team,

00:36:41.420 | there's slight differences in approach.

00:36:45.660 | So on the Elon side, there's more,

00:36:49.540 | I don't know what the right word to use,

00:36:50.860 | but aggression in terms of innovation.

00:36:53.900 | And on Waymo side, there's more sort of cautious,

00:36:58.900 | safety focused approach to the problem.

00:37:03.460 | What do you think it takes?

00:37:06.180 | What leadership at which moment is right?

00:37:09.140 | Which approach is right?

00:37:10.660 | - Look, I don't sit in either of those teams.

00:37:13.860 | So I'm unable to even verify,

00:37:16.220 | like somebody says, correct.

00:37:17.980 | In the end of the day, every innovator in that space

00:37:21.660 | will face a fundamental dilemma.

00:37:23.140 | And I would say you could put aerospace titans

00:37:27.140 | into the same bucket, which is you have to balance

00:37:30.220 | public safety with your drive to innovate.

00:37:34.300 | And this country in particular, in the States,

00:37:36.780 | has a hundred plus year history

00:37:38.340 | of doing this very successfully.

00:37:40.620 | Air travel is what a hundred times as safe per mile

00:37:43.900 | than ground travel, than cars.

00:37:46.620 | And there's a reason for it,

00:37:48.500 | because people have found ways to be very methodological

00:37:52.900 | about ensuring public safety,

00:37:55.140 | while still being able to make progress

00:37:56.940 | on important aspects, for example,

00:37:59.060 | like airline noise and fuel consumption.

00:38:01.780 | So I think that those practices are proven

00:38:06.180 | and they actually work.

00:38:07.900 | We live in a world safer than ever before.

00:38:09.900 | And yes, there will always be the provision

00:38:11.900 | that something goes wrong.

00:38:12.740 | There's always the possibility that someone makes a mistake

00:38:15.260 | or there's an unexpected failure.

00:38:17.140 | We can never guarantee to a hundred percent absolute safety

00:38:20.980 | rather than just not doing it.

00:38:23.340 | But I think I'm very proud of the history of the United States

00:38:27.100 | I mean, we've dealt with much more dangerous technology

00:38:30.140 | like nuclear energy and kept that safe too.

00:38:32.740 | We have nuclear weapons and we keep those safe.

00:38:36.420 | So we have methods and procedures

00:38:39.460 | that really balance these two things very, very successfully.

00:38:42.980 | - You've mentioned a lot of great autonomous vehicle

00:38:44.900 | companies that are taking sort of the level four, level five

00:38:48.780 | they jump in full autonomy with a safety driver

00:38:51.900 | and take that kind of approach.

00:38:53.140 | And also through simulation and so on.

00:38:55.780 | There's also the approach that Tesla Autopilot is doing

00:38:59.580 | which is kind of incrementally taking a level two vehicle

00:39:03.740 | and using machine learning and learning

00:39:05.860 | from the driving of human beings and trying to creep up

00:39:10.580 | trying to incrementally improve the system

00:39:12.380 | until it's able to achieve level four autonomy.

00:39:15.580 | So perfect autonomy in certain kind of geographical regions.

00:39:19.820 | What are your thoughts on these contrasting approaches?

00:39:23.180 | - Well, so first of all, I'm a very proud Tesla owner

00:39:25.620 | and I literally use the Autopilot every day

00:39:27.900 | and it literally has kept me safe.

00:39:29.620 | It is a beautiful technology specifically

00:39:33.980 | for highway driving when I'm slightly tired

00:39:37.660 | because then it turns me into a much safer driver

00:39:42.260 | and that I'm a hundred percent confident that's the case.

00:39:46.100 | In terms of the right approach,

00:39:47.700 | I think the biggest change I've seen

00:39:49.900 | since I ran the Waymo team is this thing called deep learning.

00:39:54.500 | Deep learning was not a hot topic when I started Waymo

00:39:58.020 | or Google self-driving cars.

00:39:59.420 | It was there, in fact, we started Google Brain

00:40:01.740 | at the same time in Google X.

00:40:02.820 | So I invested in deep learning, but people didn't talk

00:40:05.980 | about it, it wasn't a hot topic.

00:40:07.860 | And now it is.

00:40:08.700 | There's a shift of emphasis

00:40:10.380 | from a more geometric perspective

00:40:12.460 | where you use geometric sensors.

00:40:14.340 | They give you a full 3D view

00:40:15.740 | and you do a geometric reasoning about,

00:40:17.300 | oh, this box over here might be a car,

00:40:19.660 | towards a more human-like, oh, let's just learn about it.

00:40:24.180 | This looks like the thing I've seen 10,000 times before.

00:40:26.540 | So maybe it's the same thing, machine learning perspective.

00:40:30.300 | And that has really put, I think,

00:40:32.180 | all these approaches on steroids.

00:40:34.740 | At Udacity, we teach a course in self-driving cars.

00:40:38.740 | In fact, I think we've graduated over 20,000 or so people

00:40:43.820 | on self-driving car skills.

00:40:45.020 | So every self-driving car team

00:40:47.060 | in the world now uses our engineers.

00:40:49.260 | And in this course, the very first homework assignment

00:40:51.900 | is to do lane finding on images.

00:40:54.900 | And lane finding images for laymen,

00:40:56.940 | what this means is you put a camera into your car

00:40:59.060 | or you open your eyes

00:40:59.900 | and you wouldn't know where the lane is, right?

00:41:01.820 | So you can stay inside the lane with your car.

00:41:04.980 | Humans can do this super easily.

00:41:06.540 | You just look and you know where the lane is,

00:41:08.100 | just intuitively.

00:41:10.180 | For machines, for a long time, it was super hard

00:41:12.180 | because people would write these kind of crazy rules.

00:41:14.660 | If there's like one lane markers

00:41:16.100 | and here's what white really means,

00:41:17.660 | this is not quite white enough.

00:41:19.100 | So let's, oh, it's not white.

00:41:20.340 | Or maybe the sun is shining.

00:41:21.420 | So when the sun shines and this is white

00:41:23.500 | and this is a straight line,

00:41:24.660 | or maybe it's not quite a straight line

00:41:25.700 | because the road is curved.

00:41:27.260 | And do we know that there's really six feet

00:41:29.260 | between lane markings or not, or 12 feet, whatever it is.

00:41:32.220 | And now, what the students are doing,

00:41:36.260 | they would take machine learning.

00:41:37.380 | So instead of like writing these crazy rules

00:41:39.620 | for the lane marker, it's just say,

00:41:40.820 | hey, let's take an hour of driving

00:41:42.700 | and label it and tell the vehicle

00:41:44.420 | this is actually the lane by hand.

00:41:45.780 | And then these are examples

00:41:47.380 | and have the machine find its own rules

00:41:49.420 | what lane markings are.

00:41:51.420 | And within 24 hours, now every student

00:41:53.820 | that's never done any programming before in this space

00:41:56.060 | can write a perfect lane finder

00:41:58.340 | as good as the best commercial lane finders.

00:42:00.900 | And that's completely amazing to me.

00:42:02.780 | We've seen progress using machine learning

00:42:05.540 | that completely dwarfs anything

00:42:08.180 | that I saw 10 years ago.

00:42:10.940 | - Yeah, and just as a side note,

00:42:12.820 | the self-driving car nanodegree,

00:42:15.220 | the fact that you launched that many years ago now,

00:42:18.900 | maybe four years ago.

00:42:20.100 | - Three years ago.

00:42:20.940 | - Three years ago is incredible.

00:42:22.060 | That's a great example of system level thinking.

00:42:24.740 | So just taking an entire course that teaches you

00:42:27.580 | how to solve the entire problem,

00:42:29.260 | I definitely recommend people.

00:42:31.220 | - It's become super popular

00:42:32.380 | and it's become actually incredibly high quality.

00:42:34.220 | We've been with Mercedes

00:42:35.260 | and various other companies in that space.

00:42:38.060 | And we find that engineers from Tesla and Waymo

00:42:40.620 | are taking it to bay.

00:42:41.980 | The insight was that two things,

00:42:45.500 | one is existing universities will be very slow to move

00:42:49.220 | because the department lies

00:42:50.500 | and there's no department for self-driving cars.

00:42:52.380 | So between McGee and EE and computer science,

00:42:56.260 | getting those folks together into one room

00:42:57.700 | is really, really hard.

00:42:59.660 | And every professor listening here will know,

00:43:01.260 | they'll probably agree to that.

00:43:02.940 | And secondly, even if all the great universities

00:43:06.460 | did this, which none so far

00:43:08.340 | has developed a curriculum in this field,

00:43:11.060 | it is just a few thousand students that can partake

00:43:13.660 | because all the great universities are super selective.

00:43:16.180 | So how about people in India?

00:43:18.060 | How about people in China or in the Middle East

00:43:20.580 | or Indonesia or Africa?

00:43:23.380 | Why should those be excluded

00:43:25.100 | from the skill of building self-driving cars?

00:43:27.180 | Are they any dumber than we are?

00:43:28.380 | Are they any less privileged?

00:43:30.140 | And the answer is, we should just give everybody the skill

00:43:34.780 | to build a self-driving car.

00:43:35.820 | Because if we do this,

00:43:37.380 | then we have like a thousand self-driving car startups.

00:43:40.300 | And if 10% succeed, that's like a hundred,

00:43:42.900 | that means a hundred countries now

00:43:44.140 | will have self-driving cars and be safer.

00:43:46.740 | - It's kind of interesting to imagine,

00:43:48.940 | impossible to quantify, but the number,

00:43:52.500 | the, you know, over a period of several decades,

00:43:54.980 | the impact that has, like a single course,

00:43:57.900 | like a ripple effect of society.

00:43:59.780 | I just recently talked to Andrew

00:44:02.900 | and who was creator of Cosmos.

00:44:05.340 | So it's a show, it's interesting to think about

00:44:08.180 | how many scientists that show launched.

00:44:10.700 | And so it's really, in terms of impact,

00:44:15.580 | I can't imagine a better course

00:44:17.180 | than the self-driving car course.

00:44:18.620 | That's, you know, there's other more specific disciplines

00:44:21.820 | like deep learning and so on that Udacity is also teaching,

00:44:24.100 | but self-driving cars, it's really,

00:44:25.900 | really interesting course.

00:44:26.860 | - Yeah, and it came at the right moment.

00:44:28.420 | It came at a time when there were a bunch of acqui-hires.

00:44:31.700 | Acqui-hire is acquisition of a company,

00:44:34.180 | not for its technology or its products or business,

00:44:36.380 | but for its people.

00:44:38.300 | So acqui-hire means maybe the company of 70 people,

00:44:40.620 | they have no product yet, but they're super smart people

00:44:43.180 | and they pay a certain amount of money.

00:44:44.340 | So I took acqui-hires like GM Cruise and Uber and others

00:44:48.420 | and did the math and said, hey, how many people are there

00:44:52.220 | and how much money was paid?

00:44:53.780 | And as a lower bound, I estimated the value

00:44:56.980 | of a self-driving car engineer in these acquisitions

00:45:00.420 | to be at least $10 million, right?

00:45:02.220 | So think about this, you get yourself a skill

00:45:05.060 | and you team up and build a company

00:45:06.700 | and your worth now is $10 million.

00:45:09.820 | I mean, that's kind of cool.

00:45:10.820 | I mean, what other thing could you do in life

00:45:13.420 | to be worth $10 million within a year?

00:45:15.940 | - Yeah, amazing.

00:45:17.620 | But to come back for a moment onto deep learning

00:45:21.020 | and its application in autonomous vehicles,

00:45:23.760 | what are your thoughts on Elon Musk's statement,

00:45:28.500 | provocative statement perhaps, that lighter is a crutch?

00:45:31.100 | So this geometric way of thinking about the world

00:45:34.060 | may be holding us back if what we should instead be doing

00:45:38.980 | in this robotics, in this particular space

00:45:41.020 | of autonomous vehicles is using camera as a primary sensor

00:45:46.020 | and using computer vision and machine learning

00:45:48.260 | as the primary way to--

00:45:49.780 | - Look, I have two comments.

00:45:50.620 | I think first of all, we all know that people can drive cars

00:45:55.420 | without lighters in their hands because we only have eyes

00:45:59.020 | and we mostly just use eyes for driving.

00:46:02.140 | Maybe we use some other perception about our bodies,

00:46:04.620 | accelerations, occasionally our ears,

00:46:06.720 | certainly not our noses. (laughs)

00:46:10.740 | So the existence proof is there

00:46:12.500 | that eyes must be sufficient.

00:46:14.640 | In fact, we could even drive a car

00:46:17.980 | if someone put a camera out

00:46:19.480 | and then gave us the camera image with no latency,

00:46:23.460 | we would be able to drive a car that way, the same way.

00:46:26.380 | So a camera is also sufficient.

00:46:28.780 | Secondly, I really love the idea that in the Western world,

00:46:31.860 | we have many, many different people

00:46:33.620 | trying different hypotheses.

00:46:35.700 | It's almost like an ant hill.

00:46:37.340 | An ant hill tries to forge for food, right?

00:46:39.580 | You can sit there as two ants

00:46:40.820 | and agree what the perfect path is

00:46:42.580 | and then every single ant marches

00:46:44.060 | for the most likely location of food is

00:46:46.380 | or you can have them just spread out.

00:46:47.980 | And I promise you the spread out solution will be better

00:46:50.500 | because if the disgusting philosophical,

00:46:54.020 | intellectual ants get it wrong

00:46:55.620 | and they're all moving the wrong direction,

00:46:56.980 | they're gonna waste the day

00:46:58.340 | and then they're gonna discuss again for another week.

00:47:00.540 | Whereas if all these ants go in a random direction,

00:47:02.500 | someone's gonna succeed

00:47:03.540 | and they're gonna come back and claim victory

00:47:05.620 | and get the Nobel prize or whatever the ant equivalent is.

00:47:08.700 | And then they all march in the same direction.

00:47:10.580 | And that's great about society.

00:47:11.860 | That's great about the Western society.

00:47:13.220 | We're not plan-based, we're not central-based.

00:47:15.540 | We don't have a Soviet Union style central government

00:47:19.140 | that tells us where to forge.

00:47:21.020 | We just forge.

00:47:21.860 | We start in C-Corp.

00:47:23.140 | We get investor money, go out and try it out.

00:47:25.860 | And who knows who's gonna win?

00:47:27.820 | (laughs)

00:47:28.740 | - I like it.

00:47:29.580 | When you look at the long-term vision

00:47:33.500 | of autonomous vehicles,

00:47:35.220 | do you see machine learning

00:47:36.940 | as fundamentally being able to solve most of the problems?

00:47:39.620 | So learning from experience.

00:47:42.300 | - I'd say we should be very clear

00:47:44.260 | about what machine learning is and is not.

00:47:46.100 | And I think there's a lot of confusion.

00:47:48.180 | What it is today is a technology

00:47:50.900 | that can go through large databases

00:47:54.700 | of repetitive patterns and find those patterns.

00:47:59.700 | So an example, we did a study at Stanford two years ago

00:48:03.580 | where we applied machine learning

00:48:05.460 | to detecting skin cancer in images.

00:48:07.900 | And we harvested or built a data set

00:48:10.780 | of 129,000 skin photo shots

00:48:15.100 | that all had been biopsied for what the actual situation was.

00:48:19.460 | And those included melanomas and carcinomas,

00:48:22.700 | also included rashes and other skin conditions, lesions.

00:48:26.460 | And then we had a network find those patterns

00:48:30.740 | and it was by and large able to then detect skin cancer

00:48:34.540 | with an iPhone as accurately

00:48:36.700 | as the best board-certified Stanford-level dermatologist.

00:48:41.380 | We proved that.

00:48:42.780 | Now this thing was great in this one thing,

00:48:45.940 | finding skin cancer, but it couldn't drive a car.

00:48:48.540 | So the difference to human intelligence

00:48:51.580 | is we do all these many, many things

00:48:53.260 | and we can often learn from a very small data set

00:48:56.700 | of experiences, whereas machines still need

00:48:59.580 | very large data sets and things that would be very repetitive.

00:49:03.340 | Now that's still super impactful

00:49:04.660 | because almost everything we do is repetitive.

00:49:06.420 | So that's gonna really transform human labor,

00:49:09.980 | but it's not this almighty general intelligence.

00:49:13.100 | We're really far away from a system

00:49:15.300 | that would exhibit general intelligence.

00:49:17.300 | To that end, I actually commiserate the naming a little bit

00:49:21.300 | because artificial intelligence, if you believe Hollywood,

00:49:24.420 | is immediately mixed into the idea of human suppression

00:49:27.300 | and machine superiority.

00:49:30.340 | I don't think that we're gonna see this in my lifetime.

00:49:32.940 | I don't think human suppression is a good idea.

00:49:35.460 | I don't see it coming.

00:49:37.420 | I don't see the technology being there.

00:49:39.700 | What I see instead is a very pointed,

00:49:42.300 | focused pattern recognition technology

00:49:44.300 | that's able to extract patterns from large data sets.

00:49:48.420 | And in doing so, it can be super impactful, super impactful.

00:49:53.420 | Let's take the impact of artificial intelligence

00:49:55.900 | on human work.

00:49:57.620 | We all know that it takes something like 10,000 hours

00:50:00.540 | to become an expert.

00:50:01.540 | If you're gonna be a doctor or a lawyer

00:50:04.260 | or even a really good driver,

00:50:06.300 | it takes a certain amount of time to become experts.

00:50:09.500 | Machines now are able and have been shown

00:50:12.220 | to observe people become experts and observe experts

00:50:16.700 | and then extract those rules from experts

00:50:18.460 | in some interesting way that could go from law to sales

00:50:23.100 | to driving cars to diagnosing cancer

00:50:28.100 | and then giving that capability to people

00:50:31.460 | who are completely new in their job.

00:50:33.060 | We now can, and that's been done.

00:50:35.460 | It's been done commercially in many, many instantiations.

00:50:38.460 | So that means we can use machine learning

00:50:40.780 | to make people expert on the very first day of their work.

00:50:45.500 | Like think about the impact.

00:50:46.540 | If your doctor is still in their first 10,000 hours,

00:50:51.020 | you have a doctor who's not quite an expert yet.

00:50:53.780 | Who would not want a doctor who is the world's best expert?

00:50:57.340 | And now we can leverage machines

00:50:59.180 | to really eradicate error in decision-making,

00:51:02.700 | error in lack of expertise for human doctors.

00:51:06.180 | That could save your life.

00:51:08.340 | - If we can link on that for a little bit,

00:51:10.300 | in which way do you hope machines in the medical field

00:51:14.780 | could help assist doctors?

00:51:16.340 | You mentioned this sort of accelerating the learning curve

00:51:21.300 | or people, if they start a job,

00:51:24.500 | or in the first 10,000 hours can be assisted by machines.

00:51:27.340 | How do you envision that assistance looking?

00:51:29.740 | - So we built this app for an iPhone

00:51:32.340 | that can detect and classify and diagnose skin cancer.

00:51:36.300 | And we proved two years ago that it does pretty much

00:51:40.540 | as good or better than the best human doctors.

00:51:42.220 | So let me tell you a story.

00:51:43.580 | So there's a friend of mine, let's call him Ben.

00:51:45.460 | Ben is a very famous venture capitalist.

00:51:47.660 | He goes to his doctor and the doctor looks at a mole

00:51:50.700 | and says, "Hey, that mole is probably harmless."

00:51:55.340 | And for some very funny reason,

00:51:58.660 | he pulls out that phone with our app.

00:52:00.420 | He's a collaborator in our study.

00:52:02.620 | And the app says, "No, no, no, no, this is a melanoma."

00:52:06.260 | And for background, melanomas are,

00:52:08.660 | and skin cancer is the most common cancer in this country.

00:52:12.380 | Melanomas can go from stage zero to stage four

00:52:16.620 | within less than a year.

00:52:18.100 | Stage zero means you can basically cut it out yourself

00:52:20.860 | with a kitchen knife and be safe.

00:52:23.180 | And stage four means your chances of living

00:52:25.300 | for five more years are less than 20%.

00:52:27.980 | So it's a very serious, serious, serious condition.

00:52:31.180 | So this doctor who took out the iPhone

00:52:36.140 | looked at the iPhone and was a little bit puzzled.

00:52:37.700 | He said, "I mean, but just to be safe,

00:52:39.700 | let's cut it out and biopsy it."

00:52:41.580 | That's the technical term for it.

00:52:43.220 | Let's get an in-depth diagnostics

00:52:45.380 | that is more than just looking at it.

00:52:47.660 | And it came back as cancerous, as a melanoma.

00:52:50.740 | And it was then removed.

00:52:52.180 | And my friend Ben, I was hiking with him

00:52:54.900 | and we were talking about AI.

00:52:56.220 | And I said, I told him I do this book on skin cancer.

00:52:58.820 | And he said, "Oh, funny.

00:53:00.660 | My doctor just had an iPhone that found my cancer."

00:53:03.740 | (both laughing)

00:53:04.580 | Wow.

00:53:05.420 | So I was like completely intrigued.

00:53:06.860 | I didn't even know about this.

00:53:08.140 | So here's a person, I mean,

00:53:09.420 | this is a real human life, right?

00:53:11.260 | - Yes.

00:53:12.100 | - Now who doesn't know somebody

00:53:12.940 | who has been affected by cancer?

00:53:13.980 | Cancer is cause of death number two.

00:53:16.100 | Cancer is this kind of disease that is mean

00:53:19.420 | in the following way.

00:53:21.020 | Most cancers can actually be cured relatively easily

00:53:24.460 | if we catch them early.

00:53:25.820 | And the reason why we don't tend to catch them early

00:53:28.300 | is because they have no symptoms.

00:53:30.540 | Like your very first symptom of a gallbladder cancer

00:53:33.820 | or a pancreas cancer might be a headache.

00:53:37.020 | And when you finally go to your doctor

00:53:38.620 | because of these headaches or your back pain

00:53:41.580 | and you're being imaged, it's usually stage four plus.

00:53:45.860 | And that's the time when the occurring chances

00:53:48.180 | might be dropped to a single digit percentage.

00:53:50.820 | So if we could leverage AI to inspect your body

00:53:54.500 | on a regular basis without even a doctor in the room,

00:53:58.060 | maybe when you take a shower or what have you,

00:54:00.300 | I know this sounds creepy,

00:54:01.420 | but then we might be able to save millions

00:54:03.740 | and millions of lives.

00:54:06.260 | - You've mentioned there's a concern that people have

00:54:09.420 | about near-term impacts of AI in terms of job loss.

00:54:12.780 | So you've mentioned being able to assist doctors,

00:54:15.460 | being able to assist people in their jobs.

00:54:17.860 | Do you have a worry of people losing their jobs

00:54:21.060 | or the economy being affected by the improvements in AI?

00:54:25.420 | - Yeah, anybody concerned about job losses,

00:54:27.660 | please come to Gerdasady.com.

00:54:30.020 | We teach contemporary tech skills

00:54:32.300 | and we have our kind of implicit job promise.

00:54:36.740 | We often, when we measure,

00:54:38.700 | we spend way over 50% of our graduates in new jobs

00:54:41.860 | and they're very satisfied about it.

00:54:43.780 | And it costs almost nothing,

00:54:44.860 | costs like 1,500 max or something like that.

00:54:47.180 | - And I saw there's a cool new program

00:54:48.980 | that you agree with the US government,

00:54:51.180 | guaranteeing that you will help give scholarships

00:54:54.940 | that educate people in this kind of situation.

00:54:57.900 | - Yeah, we're working with the US government

00:55:00.020 | on the idea of basically rebuilding the American dream.

00:55:03.900 | So Udacity has just dedicated 100,000 scholarships

00:55:07.460 | for citizens of America for various levels of courses

00:55:12.100 | that eventually will get you a job.

00:55:15.620 | And those courses are all somewhat related

00:55:18.740 | to the tech sector because the tech sector

00:55:20.500 | is kind of the hottest sector right now.

00:55:22.100 | And they range from inter-level digital marketing

00:55:24.980 | to very advanced self-driving car engineering.

00:55:28.100 | And we're doing this with the White House

00:55:29.460 | because we think it's bipartisan.

00:55:30.900 | It's an issue that if you wanna really make America great,

00:55:35.900 | being able to be a part of the solution

00:55:40.060 | and live the American dream requires us

00:55:43.020 | to be proactive about our education and our skillset.

00:55:45.780 | It's just the way it is today.

00:55:47.700 | And it's always been this way.

00:55:48.700 | And we always had this American dream

00:55:49.940 | to send our kids to college.

00:55:51.140 | And now the American dream has to be

00:55:53.260 | to send ourselves to college.

00:55:54.660 | We can do this very, very, very efficiently

00:55:58.220 | and we can squeeze in the evenings and things to online.

00:56:01.780 | - At all ages.

00:56:03.140 | - All ages.

00:56:03.980 | So our learners go from age 11 to age 80.

00:56:08.980 | I just traveled Germany and the guy

00:56:14.060 | in the train compartment next to me was one of my students.

00:56:17.940 | It's like, wow, that's amazing.

00:56:19.820 | Think about impact.

00:56:21.300 | We've become the educator of choice for now,

00:56:24.020 | I believe officially six countries or five countries.

00:56:26.500 | Mostly in the Middle East, like Saudi Arabia and in Egypt.

00:56:30.100 | In Egypt, we just had a cohort graduate

00:56:33.420 | where we had 1100 high school students

00:56:37.300 | that went through programming skills proficient

00:56:40.340 | at the level of a computer science undergrad.

00:56:42.940 | And we had a 95% graduation rate

00:56:45.220 | even though everything's online.

00:56:46.260 | It's kind of tough, but we kind of trying to figure out

00:56:48.260 | how to make this effective.

00:56:50.140 | The vision is very, very simple.

00:56:52.540 | The vision is education ought to be a basic human right.

00:56:57.540 | It cannot be locked up behind ivory tower walls

00:57:02.300 | only for the rich people,

00:57:03.620 | for the parents who might be bribe themselves

00:57:05.620 | into the system and only for young people

00:57:08.380 | and only for people from the right demographics

00:57:10.460 | and the right geography and possibly even the right race.

00:57:14.220 | It has to be opened up to everybody.

00:57:15.820 | If we are truthful to the human mission,

00:57:18.700 | if we are truthful to our values,

00:57:20.620 | we're gonna open up education to everybody in the world.

00:57:23.420 | So Udacity's pledge of 100,000 scholarships,

00:57:27.180 | I think is the biggest pledge of scholarships ever

00:57:29.180 | in terms of numbers.

00:57:30.740 | And we're working, as I said, with the White House

00:57:32.980 | and with very accomplished CEOs

00:57:35.460 | like Tim Cook from Apple and others

00:57:37.460 | to really bring education to everywhere in the world.

00:57:40.100 | - Not to ask you to pick the favorite of your children,

00:57:44.580 | but at this point-- - Oh, that's Jasper.

00:57:46.620 | (laughing)

00:57:47.460 | I only have one that I know of.

00:57:49.700 | - Okay, good.

00:57:50.540 | In this particular moment, what nano degree,

00:57:55.780 | what set of courses are you most excited about at Udacity?

00:58:00.020 | Or is that too impossible to pick?

00:58:01.980 | - I've been super excited about something

00:58:03.780 | we haven't launched yet and we're building,

00:58:05.460 | which is when we talk to our partner companies,

00:58:09.100 | we have now a very strong footing in the enterprise world.

00:58:12.700 | And also to our students,

00:58:14.580 | we've kind of always focused on these hard skills

00:58:17.260 | like the programming skills or math skills

00:58:19.740 | or building skills or design skills.

00:58:22.180 | And a very common ask is soft skills.

00:58:25.180 | Like how do you behave in your work?

00:58:26.860 | How do you develop empathy?

00:58:28.260 | How do you work in a team?

00:58:29.540 | What are the very basics of management?

00:58:32.380 | How do you do time management?

00:58:33.700 | How do you advance your career

00:58:36.260 | in the context of a broader community?

00:58:39.260 | And that's something that we haven't done very well

00:58:41.740 | at Udacity and I would say most universities

00:58:43.860 | are doing very poorly as well

00:58:45.180 | because we are so obsessed with individual test scores

00:58:47.900 | and pay so little attention to teamwork in education.

00:58:52.620 | So that's something I see us moving into as a company

00:58:55.500 | because I'm excited about this.

00:58:56.940 | And I think, look, we can teach people tech skills

00:59:00.100 | and they're gonna be great.

00:59:00.940 | But if we teach people empathy,

00:59:02.700 | that's gonna have the same impact.

00:59:04.940 | - Maybe harder than self-driving cars, but--

00:59:08.100 | - I don't think so.

00:59:08.940 | I think the rules are really simple.

00:59:11.300 | You just have to want to engage.

00:59:14.380 | It's weird, we literally went in school in K through 12,

00:59:18.180 | we teach kids like get the highest math score.

00:59:20.460 | And if you are a rational human being,

00:59:22.860 | you might evolve from this education,

00:59:24.860 | say having the best math score

00:59:26.740 | and the best English scores, make me the best leader.

00:59:29.620 | And it turns out not to be the case.

00:59:31.060 | It's actually really wrong

00:59:32.860 | because making, first of all, in terms of math scores,

00:59:35.820 | I think it's perfectly fine to hire somebody

00:59:37.620 | with great math skills.

00:59:38.500 | You don't have to do it yourself.

00:59:40.620 | You can hire someone with great empathy for you,

00:59:42.740 | that's much harder, but it can always hire someone

00:59:44.980 | with great math skills.

00:59:46.340 | But we live in an affluent world

00:59:48.980 | where we constantly deal with other people.

00:59:51.020 | And that's a beauty.

00:59:51.900 | It's not a nuisance, it's a beauty.

00:59:53.340 | So if we somewhat develop that muscle

00:59:55.940 | that we can do that well and empower others

00:59:59.700 | in the workplace, I think we're gonna be super successful.

01:00:02.900 | - And I know many fellow roboticists

01:00:05.780 | and computer scientists that I will insist

01:00:08.660 | to take this course.

01:00:09.820 | (laughing)

01:00:11.020 | - Not to be named here.

01:00:12.180 | - Not to be named.

01:00:13.780 | Many, many years ago, 1903,

01:00:17.980 | the Wright brothers flew in Kitty Hawk for the first time.

01:00:22.660 | And you've launched a company of the same name, Kitty Hawk,

01:00:26.980 | with the dream of building flying cars, EVtols.

01:00:31.980 | So at the big picture, what are the big challenges

01:00:35.780 | of making this thing that actually have inspired

01:00:38.700 | generations of people about what the future looks like?

01:00:41.780 | What does it take?

01:00:42.620 | What are the biggest challenges?

01:00:43.700 | - So flying cars has always been a dream.

01:00:47.260 | Every boy, every girl wants to fly.

01:00:49.740 | Let's be honest.

01:00:50.580 | - Yes.

01:00:51.420 | - And let's go back in our history

01:00:52.380 | of your dreaming of flying.

01:00:53.820 | I think my, honestly, my single most remembered

01:00:56.540 | childhood dream has been a dream

01:00:58.340 | where I was sitting on a pillow and I could fly.

01:01:00.780 | I was like five years old.

01:01:02.060 | I remember like maybe three dreams of my childhood,

01:01:04.180 | but that's the one I remember most vividly.

01:01:06.420 | And then Peter Thiel famously said,

01:01:09.380 | "They promised us flying cars

01:01:10.660 | and they gave us 140 characters,"

01:01:12.780 | pointing at Twitter at the time,

01:01:15.220 | limited message size to 140 characters.

01:01:18.380 | So we're coming back now to really go

01:01:20.220 | for this super impactful stuff like flying cars.

01:01:23.220 | And to be precise, they're not really cars.

01:01:25.900 | They don't have wheels.

01:01:27.180 | They're actually much closer to a helicopter

01:01:28.580 | than anything else.

01:01:29.620 | They take off vertically and they fly horizontally,

01:01:32.060 | but they have important differences.

01:01:34.380 | One difference is that they are much quieter.

01:01:37.740 | We just released a vehicle called Project Heaviside

01:01:41.620 | that can fly over you as low as a helicopter.

01:01:43.540 | And you basically can't hear it.

01:01:45.300 | It's like 38 decibels.

01:01:46.740 | It's like, if you were inside the library,

01:01:49.300 | you might be able to hear it,

01:01:50.260 | but anywhere outdoors, your ambient noise is higher.

01:01:53.020 | Secondly, they're much more affordable.

01:01:57.060 | They're much more affordable than helicopters.

01:01:59.020 | And the reason is helicopters are expensive

01:02:01.940 | for many reasons.

01:02:03.060 | There's lots of single point of figures in a helicopter.

01:02:07.020 | There's a bolt between the blades

01:02:09.140 | that's called Jesus bolt.

01:02:10.780 | And the reason why it's called Jesus bolt

01:02:12.420 | is that if this bolt breaks, you will die.

01:02:16.380 | There is no second solution in helicopter flight.

01:02:19.500 | Whereas we have these distributed mechanism.

01:02:21.500 | When you go from gasoline to electric,

01:02:23.740 | you can now have many, many, many small motors

01:02:25.820 | as opposed to one big motor.

01:02:27.260 | And that means if you lose one of those motors,

01:02:28.780 | not a big deal.

01:02:29.620 | Heaviside, if it loses a motor, has eight of those,

01:02:32.820 | if it loses one of those eight motors,

01:02:34.020 | so it's seven left,

01:02:35.180 | it can take off just like before

01:02:37.260 | and land just like before.

01:02:38.820 | We are now also moving into a technology

01:02:42.020 | that doesn't require a commercial pilot

01:02:44.140 | because in some level,

01:02:45.500 | flight is actually easier than ground transportation.

01:02:48.980 | Like in self-driving cars,

01:02:50.700 | the world is full of like children and bicycles

01:02:54.500 | and other cars and mailboxes and curbs

01:02:56.620 | and shrubs and what have you,

01:02:58.420 | all these things you have to avoid.

01:03:00.500 | When you go above the buildings and tree lines,

01:03:03.740 | there's nothing there.

01:03:04.620 | I mean, you can do the test right now,

01:03:06.100 | look outside and count the number of things you see flying.

01:03:09.420 | I'd be shocked if you could see more than two things.

01:03:11.500 | It's probably just zero.

01:03:12.860 | In the Bay Area, the most I've ever seen was six.

01:03:16.940 | And maybe it's 15 or 20, but not 10,000.

01:03:20.420 | So the sky is very ample and very empty and very free.

01:03:23.980 | So the vision is, can we build a socially acceptable

01:03:27.820 | mass transit solution for daily transportation

01:03:32.340 | that is affordable?

01:03:34.260 | And we have an existence proof.

01:03:36.300 | Heaviside can fly 100 miles in range

01:03:39.780 | with still 30% electric reserves.

01:03:43.260 | It can fly up to like 180 miles an hour.

01:03:46.060 | We know that that solution at scale

01:03:48.860 | would make your ground transportation

01:03:51.380 | 10 times as fast as a car

01:03:53.780 | based on US census or statistics data,

01:03:57.340 | which means we would take your 300 hours of daily,

01:04:00.860 | of yearly commute down to 30 hours

01:04:02.980 | and give you 270 hours back.

01:04:05.180 | Who wouldn't want, I mean, who doesn't hate traffic?

01:04:07.660 | Like I hate, give me the person who doesn't hate traffic.

01:04:10.780 | I hate traffic.

01:04:11.620 | Every time I'm in traffic, I hate it.

01:04:13.620 | And if we could free the world from traffic,

01:04:17.540 | we have technology, we can free the world from traffic.

01:04:20.020 | We have the technology.

01:04:21.300 | It's there, we have an existence proof.

01:04:23.020 | It's not a technological problem anymore.

01:04:25.420 | - Do you think there is a future where tens of thousands,

01:04:29.300 | maybe hundreds of thousands of both delivery drones

01:04:34.300 | and flying cars of this kind, EV towers fill the sky?

01:04:39.340 | - I absolutely believe this.

01:04:40.900 | And there's obviously the societal acceptance

01:04:43.540 | is a major question.

01:04:45.420 | And of course, safety is.

01:04:46.900 | I believe in safety,

01:04:48.020 | we're gonna exceed ground transportation safety

01:04:50.300 | as has happened for aviation already, commercial aviation.

01:04:54.460 | And in terms of acceptance,

01:04:56.580 | I think one of the key things is noise.

01:04:58.260 | That's why we are focusing relentlessly on noise

01:05:00.900 | and we built perhaps the quietest electric VTOL vehicle

01:05:05.580 | ever built.

01:05:06.420 | The nice thing about the sky is it's three dimensional.

01:05:09.700 | So any mathematician will immediately recognize

01:05:12.460 | the difference between 1D of like a regular highway

01:05:14.900 | to 3D of a sky.

01:05:16.180 | But to make it clear for the layman,

01:05:19.300 | say you wanna make 100 vertical lanes

01:05:22.700 | of highway 101 in San Francisco,

01:05:24.980 | because you believe building 100 vertical lanes

01:05:27.180 | is the right solution.

01:05:28.820 | Imagine how much it would cost

01:05:30.140 | to stack 100 vertical lanes physically onto 101.

01:05:33.380 | That would be prohibitive.

01:05:34.300 | That would be consuming the world's GDP for an entire year

01:05:37.740 | just for one highway.

01:05:39.180 | It's amazingly expensive.

01:05:41.220 | In the sky, it would just be a recompilation

01:05:43.660 | of a piece of software

01:05:44.540 | because all these lanes are virtual.

01:05:46.500 | That means any vehicle that is in conflict

01:05:49.180 | with another vehicle would just go to different altitudes

01:05:51.780 | and then the conflict is gone.

01:05:53.260 | And if you don't believe this,

01:05:55.300 | that's exactly how commercial aviation works.

01:05:58.500 | When you fly from New York to San Francisco,

01:06:01.380 | another plane flies from San Francisco to New York,

01:06:04.140 | they're at different altitudes

01:06:05.180 | so they don't hit each other.

01:06:06.660 | It's a solved problem for the jet space

01:06:10.300 | and it will be a solved problem for the urban space.

01:06:12.660 | There's companies like Google, Bing and Amazon

01:06:15.300 | working on very innovative solutions

01:06:16.980 | how do we have space management.

01:06:18.500 | They use exactly the same principles as we use today

01:06:21.580 | to route today's jets.

01:06:23.220 | There's nothing hard about this.

01:06:25.860 | - Do you envision autonomy being a key part of it

01:06:28.940 | so that the flying vehicles are either semi-autonomous

01:06:33.940 | or fully autonomous?

01:06:36.860 | - 100% autonomous.

01:06:37.820 | You don't want idiots like me flying in the sky.

01:06:40.420 | I promise you.

01:06:41.900 | And if you have 10,000,

01:06:43.180 | watch the movie "The Fifth Element"

01:06:45.980 | to get a fee for what would happen

01:06:48.180 | if it's not autonomous.

01:06:49.420 | - And a centralized, that's a really interesting idea

01:06:51.660 | of a centralized sort of management system

01:06:55.260 | for lanes and so on.

01:06:56.300 | So actually just being able to have

01:06:58.780 | similar as we have in the current commercial aviation

01:07:02.980 | but scale it up to much more vehicles.

01:07:05.540 | That's a really interesting optimization problem.

01:07:07.660 | - It is mathematically very, very straightforward.

01:07:11.060 | Like the gap we leave between jets is gargantuous.

01:07:13.500 | And part of the reason is there isn't that many jets.

01:07:16.380 | So it just feels like a good solution.

01:07:18.820 | Today, when you get vectored by air traffic control,

01:07:22.380 | someone talks to you, right?

01:07:23.900 | So an ATC controller might have up to maybe 20 planes

01:07:26.980 | on the same frequency.

01:07:28.140 | And then they talk to you, you have to talk back.

01:07:30.340 | And that feels right because there isn't more than 20 planes

01:07:32.700 | around anyhow, so you can talk to everybody.

01:07:34.980 | But if there's 20,000 things around,

01:07:36.740 | you can't talk to everybody anymore.

01:07:37.980 | So we have to do something that's called digital,

01:07:40.260 | like text messaging.

01:07:41.540 | Like we do have solutions.

01:07:43.060 | Like we have what, four or five billion smartphones

01:07:45.540 | in the world now, right?

01:07:46.460 | And they're all connected.

01:07:47.740 | And somehow we solve the scale problem for smartphones.

01:07:50.740 | We know where they all are.

01:07:51.940 | They can talk to somebody and they're very reliable.

01:07:54.860 | They're amazingly reliable.

01:07:56.460 | We could use the same system,

01:07:58.620 | the same scale for air traffic control.

01:08:01.060 | So instead of me as a pilot talking to a human being

01:08:04.060 | in the middle of the conversation,

01:08:06.260 | receiving a new frequency, like how ancient is that?

01:08:09.660 | We could digitize this stuff

01:08:11.260 | and digitally transmit the right flight coordinates.

01:08:15.260 | And that solution will automatically scale

01:08:18.060 | to 10,000 vehicles.

01:08:20.060 | - We talked about empathy a little bit.

01:08:22.180 | Do you think we'll one day build an AI system

01:08:25.780 | that a human being can love and that loves that human back?

01:08:30.140 | Like in the movie "Her."

01:08:31.340 | - Look, I'm a pragmatist.

01:08:33.980 | For me, AI is a tool.

01:08:35.620 | It's like a shovel.

01:08:37.020 | And the ethics of using the shovel are always

01:08:40.800 | with us, the people.

01:08:41.860 | And it has to be this way.

01:08:44.060 | In terms of emotions, I would hate to come into my kitchen

01:08:49.060 | and see that my refrigerator spoiled all my food,

01:08:54.220 | then have it explained to me that it fell in love

01:08:56.500 | with the dishwasher and I wasn't as nice as the dishwasher.

01:08:59.660 | So as a result, it neglected me.

01:09:02.180 | That would just be a bad experience

01:09:05.140 | and it would be a bad product.

01:09:07.060 | I would probably not recommend this refrigerator

01:09:09.540 | to my friends.

01:09:10.420 | And that's where I draw the line.

01:09:12.900 | I think to me, technology has to be reliable.

01:09:16.620 | It has to be predictable.

01:09:17.700 | I want my car to work.

01:09:19.860 | I don't want to fall in love with my car.

01:09:22.860 | I just want it to work.

01:09:24.620 | I want it to compliment me, not to replace me.

01:09:27.180 | I have very unique human properties

01:09:30.620 | and I want the machines to make me,

01:09:33.420 | turn me into a superhuman.

01:09:35.700 | Like I'm already a superhuman today,

01:09:37.820 | thanks to the machines that surround me.

01:09:39.260 | And I give you examples.

01:09:40.780 | I can run across the Atlantic near the speed of sound

01:09:45.700 | at 36,000 feet today.

01:09:48.460 | That's kind of amazing.

01:09:49.580 | My voice now carries me all the way to Australia

01:09:53.620 | using a smartphone today.

01:09:56.580 | And it's not the speed of sound, which would take hours.

01:10:00.060 | It's the speed of light.

01:10:01.300 | My voice travels at the speed of light.

01:10:03.820 | How cool is that?

01:10:04.660 | That makes me superhuman.

01:10:06.300 | I would even argue my flushing toilet makes me superhuman.

01:10:10.500 | Just think of the time before flushing toilets.

01:10:13.780 | And maybe you have a very old person in your family

01:10:16.460 | that you can ask about this

01:10:18.460 | or take a trip to rural India to experience it.

01:10:22.100 | It makes me superhuman.

01:10:25.780 | So to me, what technology does, it compliments me.

01:10:28.860 | It makes me stronger.

01:10:30.900 | Therefore, words like love and compassion have very little,

01:10:34.900 | have very little interest in this for machines.

01:10:38.580 | I have interest in people.

01:10:40.700 | - You don't think, first of all, beautifully put,

01:10:44.260 | beautifully argued,

01:10:45.660 | but do you think love has use in our tools, compassion?

01:10:50.420 | - I think love is a beautiful human concept.

01:10:53.260 | And if you think of what love really is,

01:10:55.380 | love is a means to convey safety, to convey trust.

01:11:00.380 | I think trust has a huge need in technology as well,

01:11:07.420 | not just people.

01:11:09.140 | We want to trust our technology

01:11:11.220 | the same way we, or in a similar way we trust people.

01:11:14.260 | In human interaction, standards have emerged

01:11:19.340 | and feelings, emotions have emerged,

01:11:21.740 | maybe genetically, maybe biologically,

01:11:23.900 | that are able to convey sense of trust, sense of safety,

01:11:26.540 | sense of passion, of love, of dedication

01:11:28.860 | that makes the human fabric.

01:11:30.780 | And I'm a big slacker for love.

01:11:33.700 | I want to be loved.

01:11:34.580 | I want to be trusted.

01:11:35.420 | I want to be admired.

01:11:36.820 | All these wonderful things.

01:11:38.820 | And because all of us, we have this beautiful system,

01:11:42.140 | I wouldn't just blindly copy this to the machines.

01:11:44.780 | Here's why.

01:11:46.140 | When you look at, say, transportation,

01:11:49.300 | you could have observed that up to the end

01:11:53.260 | of the 19th century, almost all transportation

01:11:56.180 | used any number of legs,

01:11:58.060 | from one leg to two legs to a thousand legs.

01:12:01.660 | And you could have concluded that is the right way

01:12:03.740 | to move about the environment.

01:12:06.780 | We've made the exception of birds, who use flapping wings.

01:12:08.940 | In fact, there were many people in aviation

01:12:10.820 | that flapped wings to their arms and jumped from cliffs.

01:12:13.700 | Most of them didn't survive.

01:12:15.100 | Then the interesting thing is that the technology solutions

01:12:19.900 | are very different.

01:12:21.580 | Like, in technology, it's really easy to build a wheel.

01:12:23.900 | In biology, it's super hard to build a wheel.

01:12:25.700 | There's very few perpetually rotating things in biology,

01:12:30.100 | and they usually run cells and things.

01:12:34.180 | In engineering, we can build wheels,

01:12:37.180 | and those wheels gave rise to cars.

01:12:39.980 | Similar wheels gave rise to aviation.

01:12:44.380 | Like, there's no thing that flies

01:12:46.700 | that wouldn't have something that rotates,

01:12:48.820 | like a jet engine or helicopter blades.

01:12:52.420 | So the solutions have used very different physical laws

01:12:55.500 | than nature, and that's great.

01:12:58.060 | So for me to be too much focused on,

01:13:00.100 | oh, this is how nature does it, let's just replicate it,

01:13:03.340 | if we really believed that the solution

01:13:05.380 | to the agricultural revolution was a humanoid robot,

01:13:08.700 | it would still be waiting today.

01:13:10.940 | - Again, beautifully put.

01:13:12.540 | You said that you don't take yourself too seriously.

01:13:15.860 | - Did I say that?

01:13:16.700 | You want me to say that?

01:13:19.180 | - Maybe I did.

01:13:20.020 | - You're not taking me seriously.

01:13:20.980 | - I'm not, yeah, that's right.

01:13:22.860 | - Good, you're right, I don't wanna.

01:13:24.460 | - I just made that up.

01:13:25.740 | But you have a humor and a lightness about life

01:13:29.100 | that I think is beautiful and inspiring to a lot of people.

01:13:33.500 | Where does that come from?

01:13:35.020 | The smile, the humor, the lightness amidst all the chaos

01:13:40.020 | of the hard work that you're in, where does that come from?

01:13:43.660 | - I just love my life.

01:13:44.540 | I love the people around me.

01:13:46.100 | I'm just so glad to be alive.

01:13:49.740 | Like, I'm, what, 52, hard to believe.

01:13:53.620 | People say 52 is a new 51, so now I feel better.

01:13:56.220 | (Lex laughing)

01:13:58.540 | But in looking around the world,

01:14:02.340 | looking, just go back 200, 300 years.

01:14:05.180 | Humanity is, what, 300,000 years old?

01:14:09.340 | But for the first 300,000 years minus the last 100,

01:14:13.980 | our life expectancy would have been

01:14:17.060 | plus or minus 30 years, roughly, give or take.

01:14:20.260 | So I would be long dead now.

01:14:22.620 | That makes me just enjoy every single day of my life

01:14:26.860 | because I don't deserve this.

01:14:28.260 | Why am I born today when so many of my ancestors

01:14:32.460 | died of horrible deaths, like famines, massive wars

01:14:37.460 | that ravaged Europe for the last 1,000 years,

01:14:41.860 | mystically disappeared after World War II

01:14:44.540 | when the Americans and the Allies did something amazing

01:14:47.580 | to my country that didn't deserve it,

01:14:49.780 | the country of Germany.

01:14:51.460 | This is so amazing.

01:14:52.620 | And then when you're alive and feel this every day,

01:14:56.940 | then it's just so amazing what we can accomplish,

01:15:01.940 | what we can do.

01:15:03.460 | We live in a world that is so incredibly

01:15:06.380 | vastly changing every day.

01:15:08.700 | Almost everything that we cherish,

01:15:10.980 | from your smartphone to your flushing toilet,

01:15:14.540 | to all these basic inventions,

01:15:16.180 | your new clothes you're wearing, your watch, your plane,

01:15:19.580 | penicillin, I don't know, anesthesia for surgery,

01:15:25.780 | penicillin, have been invented in the last 150 years.

01:15:30.020 | So in the last 150 years, something magical happened.

01:15:32.380 | And I would trace it back to Gutenberg

01:15:34.340 | and the printing press that has been able

01:15:35.900 | to disseminate information more efficiently than before,

01:15:38.780 | that all of a sudden we're able to invent agriculture

01:15:42.860 | and nitrogen fertilization that made agriculture

01:15:45.940 | so much more potent that we didn't have to work

01:15:48.020 | in the farms anymore and we could start reading and writing

01:15:50.140 | and we could become all these wonderful things

01:15:52.300 | we are today, from airline pilot to massage therapist

01:15:54.740 | to software engineer.

01:15:56.780 | It's just amazing.

01:15:57.620 | Like living in that time is such a blessing.

01:16:00.620 | We should sometimes really think about this.

01:16:03.340 | Steven Pinker, who is a very famous author and philosopher

01:16:07.340 | whom I really adore, wrote a great book called

01:16:09.420 | "Enlightenment Now" and that's maybe the one book

01:16:11.140 | I would recommend.

01:16:11.980 | And he asked the question if there was only

01:16:14.500 | a single article written in the 20th century,

01:16:17.140 | only one article, what would it be?

01:16:19.100 | What's the most important innovation,

01:16:21.140 | the most important thing that happened?

01:16:23.060 | And he would say this article would credit

01:16:24.700 | a guy named Carl Bosch.

01:16:27.020 | And I challenge anybody, have you ever heard

01:16:29.460 | of the name Carl Bosch?

01:16:31.180 | I hadn't, okay.

01:16:32.940 | There's a Bosch Corporation in Germany,

01:16:35.420 | but it's not associated with Carl Bosch.

01:16:37.420 | So I looked it up.

01:16:39.860 | Carl Bosch invented nitrogen fertilization.

01:16:42.620 | And in doing so, together with an older invention

01:16:45.540 | of irrigation, was able to increase the yield

01:16:49.180 | per agricultural land by a factor of 26.

01:16:52.860 | So a 2,500% increase in fertility of land.

01:16:57.700 | And that, so Steve Pinker argues,

01:17:00.540 | saved over two billion lives today.

01:17:03.900 | Two billion people who would be dead

01:17:05.700 | if this man hadn't done what he had done, okay?

01:17:08.420 | Think about that impact and what that means to society.

01:17:12.180 | That's the way I look at the world.

01:17:14.180 | I mean, it's just so amazing to be alive

01:17:16.180 | and to be part of this.

01:17:17.020 | And I'm so glad I lived after Carl Bosch and not before.

01:17:21.340 | - I don't think there's a better way to end this, Sebastian.

01:17:24.020 | It's an honor to talk to you,

01:17:25.500 | to have had the chance to learn from you.

01:17:27.380 | Thank you so much for talking to me.

01:17:28.340 | - Thanks for coming on, it's a real pleasure.

01:17:31.020 | - Thank you for listening to this conversation

01:17:32.820 | with Sebastian Thrun.

01:17:34.420 | And thank you to our presenting sponsor, Cash App.

01:17:37.500 | Download it, use code LEXPODCAST,

01:17:40.260 | you'll get $10 and $10 will go to FIRST,

01:17:43.260 | a STEM education nonprofit that inspires

01:17:45.540 | hundreds of thousands of young minds

01:17:47.500 | to learn and to dream of engineering our future.

01:17:50.580 | If you enjoy this podcast, subscribe on YouTube,

01:17:53.380 | get five stars on Apple Podcast, support on Patreon,

01:17:56.660 | or connect with me on Twitter.

01:17:58.860 | And now let me leave you with some words of wisdom

01:18:01.300 | from Sebastian Thrun.

01:18:03.260 | It's important to celebrate your failures

01:18:05.440 | as much as your successes.

01:18:07.740 | If you celebrate your failures really well,

01:18:09.820 | if you say, wow, I failed, I tried, I was wrong,

01:18:13.940 | but I learned something.

01:18:15.620 | Then you realize you have no fear.

01:18:18.300 | And when your fear goes away, you can move the world.

01:18:21.660 | Thank you for listening and hope to see you next time.

01:18:25.500 | (upbeat music)

01:18:28.080 | (upbeat music)

01:18:30.660 | [BLANK_AUDIO]

Sebastian Thrun: Flying Cars, Autonomous Vehicles, and Education | Lex Fridman Podcast #59

Chapters