back to indexKyle Vogt: Cruise Automation | Lex Fridman Podcast #14
link |
The following is a conversation with Kyle Vogt.
link |
He's the president and the CTO of Cruise Automation,
link |
leading an effort to solve one of the biggest robotics challenges of our time,
link |
vehicle automation. He's a cofounder of two successful companies, Twitch
link |
and Cruise, that have each sold for a billion dollars.
link |
And he's a great example of the innovative spirit that flourishes
link |
in Silicon Valley, and now is facing an interesting and exciting challenge of
link |
matching that spirit with the mass production and the safety centric
link |
culture of a major automaker like General Motors. This conversation is
link |
part of the MIT Artificial General Intelligence series
link |
and the Artificial Intelligence podcast. If you enjoy it,
link |
please subscribe on YouTube, iTunes, or simply connect with me on Twitter
link |
at Lex Friedman, spelled F R I D. And now here's my conversation with Kyle Vogt.
link |
You grew up in Kansas, right? Yeah, and I just saw that picture you had hidden
link |
over there, so I'm a little bit a little bit worried about that now.
link |
Yeah, so in high school in Kansas City, you joined Shawnee Mission
link |
North high school robotics team. Yeah. Now that wasn't your high school.
link |
That's right, that was that was the only high school in the area that had a
link |
like a teacher who was willing to sponsor our first robotics team.
link |
I was gonna troll you a little bit. Jog your memory a little bit.
link |
Yeah, I was trying to look super cool and intense, because you know this
link |
was BattleBots. This is serious business. So we're standing there with a welded
link |
steel frame and looking tough. So go back there. What is that drew you
link |
to robotics? Well, I think I've been trying to figure
link |
this out for a while, but I've always liked building things with Legos. And
link |
when I was really, really young, I wanted the Legos that had motors and
link |
other things. And then, you know, Lego Mindstorms came out, and for the
link |
first time you could program Lego contraptions. And I think things
link |
just sort of snowballed from that. But I remember
link |
seeing, you know, the BattleBots TV show on Comedy Central and thinking that is
link |
the coolest thing in the world. I want to be a part of that.
link |
And not knowing a whole lot about how to build these
link |
200 pound fighting robots. So I sort of obsessively poured over
link |
the internet forums where all the creators for BattleBots would sort of
link |
hang out and talk about, you know, document their build progress and
link |
everything. And I think I read, I must have read like,
link |
you know, tens of thousands of forum posts from basically
link |
everything that was out there on what these people were doing. And eventually
link |
like sort of triangulated how to put some of these things together.
link |
And I ended up doing BattleBots, which was, you know, I was like 13 or 14, which
link |
was pretty awesome. I'm not sure if the show is still
link |
running, but so BattleBots is, there's not an artificial
link |
intelligence component. It's remotely controlled. And it's
link |
almost like a mechanical engineering challenge of building things
link |
that can be broken. They're radio controlled. So,
link |
and I think that they allowed some limited form of autonomy.
link |
But, you know, in a two minute match, you're, in the way these things ran,
link |
you're really doing yourself a disservice by trying to automate it
link |
versus just, you know, do the practical thing, which is drive it yourself.
link |
And there's an entertainment aspect, just going on YouTube, there's like an,
link |
some of them wield an axe, some of them, I mean, there's that fun.
link |
So what drew you to that aspect? Was it the mechanical engineering?
link |
Was it the dream to create like Frankenstein and
link |
sentient being? Or was it just like the Lego,
link |
you like tinkering with stuff? I mean, that was just building something.
link |
I think the idea of, you know, this radio controlled machine that
link |
can do various things, if it has like a weapon or something was pretty
link |
interesting. I agree it doesn't have the same
link |
appeal as, you know, autonomous robots, which I,
link |
which I, you know, sort of gravitated towards later on. But it was definitely
link |
an engineering challenge because everything you did in that
link |
competition was pushing components to their limits. So we would
link |
buy like these $40 DC motors that came out of a
link |
winch, like on the front of a pickup truck or something, and we'd
link |
power the car with those and we'd run them at like double or triple their
link |
rated voltage. So they immediately start overheating,
link |
but for that two minute match you can get,
link |
you know, a significant increase in the power output of those motors
link |
before they burn out. And so you're doing the same thing for your battery packs,
link |
all the materials in the system. And I think there's something, something
link |
intrinsically interesting about just seeing like where things break.
link |
And did you offline see where they break? Did you
link |
take it to the testing point? Like how did you know two minutes? Or was there a
link |
reckless let's just go with it and see? We weren't very good at
link |
BattleBots. We lost all of our matches the first round.
link |
The one I built first, both of them were these wedge shaped robots because
link |
wedge, even though it's sort of boring to look
link |
at, is extremely effective. You drive towards another robot and
link |
the front edge of it gets under them and then they sort of flip over,
link |
kind of like a door stopper. And the first one had a
link |
pneumatic polished stainless steel spike on the front that would shoot out about
link |
eight inches. The purpose of which is what? Pretty,
link |
pretty ineffective actually, but it looks cool.
link |
And was it to help with the lift? No, it was, it was just to try to poke holes
link |
in the other robot. And then the second time I did it, which is the following,
link |
I think maybe 18 months later, we had a, well a titanium axe with a, with a
link |
hardened steel tip on it that was powered by a hydraulic
link |
cylinder which we were activating with liquid CO2, which was,
link |
had its own set of problems. So great, so that's kind of on the hardware side.
link |
I mean at a certain point there must have been born a fascination
link |
on the software side. So what was the first piece of code you've written?
link |
Go back there, see what language was it?
link |
What, what was that? Was it Emacs? Vim? Was it a more
link |
respectable modern IDE? Do you, do you remember any of this?
link |
Yeah, well I remember, I think maybe when I was in
link |
third or fourth grade, the school I was at, elementary school, had a bunch of
link |
Apple II computers and we'd play games on those. And I
link |
remember every once in a while something would,
link |
would, would crash or wouldn't start up correctly and it would dump you out to
link |
what I later learned was like sort of a command prompt.
link |
And my teacher would come over and type, I actually remember this to this day for
link |
some reason, like PR number six or PR pound six, which is
link |
peripheral six, which is the disk drive, which would fire up the disk and load the
link |
program. And I just remember thinking wow, she's
link |
like a hacker, like teach me these, these codes, these error codes, that is what
link |
I called them at the time. But she had no interest in that, so it
link |
wasn't until I think about fifth grade that I had a school where you could
link |
actually go on these Apple IIs and learn to program. And so it was all in basic,
link |
you know, where every line, you know, the line numbers are all number, that every
link |
line is numbered and you have to like leave enough space
link |
between the numbers so that if you want to tweak your code you go back and
link |
the first line was 10 and the second line is 20. Now you have to go back and
link |
insert 15 and if you need to add code in front of
link |
that, you know, 11 or 12 and you hope you don't run out of line numbers and have
link |
to redo the whole thing. And there's go to statements? Yeah, go to
link |
and it's very basic, maybe hence the name, but a lot of fun.
link |
And that was like, that was, you know, that's when, you know,
link |
when you first program you see the magic of it.
link |
It's like, it just, just like this world opens up with, you know, endless
link |
possibilities for the things you could build or
link |
or accomplish with that computer. So you got the bug then, so
link |
even starting with basic and then what C++ throughout,
link |
what did you, was there computer programming, computer science classes in
link |
high school? Not, not where I went, so it was self
link |
taught, but I did a lot of programming. The thing that, you know, sort of
link |
pushed me in the path of eventually working on self driving cars is actually
link |
one of these really long trips driving from my
link |
house in Kansas to, to I think Las Vegas where we did the BattleBots competition
link |
and I had just gotten my, I think my learner's permit or
link |
early driver's permit and so I was driving this,
link |
you know, 10 hour stretch across western Kansas where it's just,
link |
you're going straight on a highway and it is mind numbingly boring. And I
link |
remember thinking even then with my sort of mediocre programming
link |
background that this is something that a computer can do, right? Let's take a
link |
picture of the road, let's find the yellow lane markers and,
link |
you know, steer the wheel. And, you know, later I'd come to realize this had been
link |
done, you know, since, since the 80s or the 70s or even
link |
earlier, but I still wanted to do it and sort of
link |
immediately after that trip switched from sort of BattleBots, which is more
link |
radio controlled machines, to thinking about building,
link |
you know, autonomous vehicles of some scale. Start off with really small
link |
electric ones and then, you know, progress to what we're
link |
doing now. So what was your view of artificial intelligence at that point?
link |
What did you think? So this is before, there's been waves in artificial
link |
intelligence, right? The current wave with deep learning
link |
makes people believe that you can solve in a really rich deep way the computer
link |
vision perception problem, but like in
link |
before the deep learning craze, you know, how do you think about,
link |
how would you even go about building a thing that perceives itself in the
link |
world, localizes itself in the world, moves around the world?
link |
Like when you were younger, I mean, what was your thinking about it? Well,
link |
prior to deep neural networks or convolutional neural
link |
analysis, these modern techniques we have, or at least
link |
ones that are in use today, it was all a heuristic space and so like old school
link |
image processing and I think extracting, you know, yellow lane markers out of an
link |
image of a road is one of the problems that lends itself
link |
reasonably well to those heuristic based methods, you know, like
link |
just do a threshold on the color yellow and then try to
link |
fit some lines to that using a Huff transform or something and then
link |
go from there. Traffic light detection and stop sign detection, red, yellow, green.
link |
And I think you could, I mean, if you wanted to do a full,
link |
I was just trying to make something that would stay in between the lanes on a
link |
highway, but if you wanted to do the full,
link |
the full, you know, set of capabilities needed for a driverless car,
link |
I think you could, and we'd done this at cruise, you know, in the very first days,
link |
you can start off with a really simple, you know, human written heuristic just to
link |
get the scaffolding in place for your system. Traffic light detection,
link |
probably a really simple, you know, color thresholding on day one just to
link |
get the system up and running before you migrate to, you know, a deep
link |
learning based technique or something else.
link |
And, you know, back in when I was doing this, my first one, it was on a Pentium
link |
203, 233 megahertz computer in it and I think I wrote the first version in
link |
basic, which is like an interpreted language. It's
link |
extremely slow because that's the thing I knew at the time.
link |
And so there was no, no chance at all of using,
link |
there was no, no computational power to do any sort of reasonable
link |
deep nets like you have today. So I don't know what kids these days are doing. Are
link |
kids these days, you know, at age 13 using neural networks in
link |
their garage? I mean, that would be awesome.
link |
I get emails all the time from, you know, like 11, 12 year olds
link |
saying I'm having, you know, I'm trying to follow this TensorFlow tutorial
link |
and I'm having this problem. And the general approach
link |
in the deep learning community is of extreme optimism of, as opposed to,
link |
you mentioned like heuristics, you can, you can, you can
link |
separate the autonomous driving problem into modules and try to solve it sort of
link |
rigorously, or you can just do it end to end.
link |
And most people just kind of love the idea that, you know, us humans do it end
link |
to end. We just perceive and act. We should be able to use that, do the
link |
same kind of thing when you're on nets. And that,
link |
that kind of thinking, you don't want to criticize that kind of thinking because
link |
eventually they will be right. Yeah. And so it's exciting and especially
link |
when they're younger to explore that as a really exciting approach. But yeah,
link |
it's, it's changed the, the language,
link |
the kind of stuff you're tinkering with. It's kind of exciting to see when these
link |
teenagers grow up. Yeah. I can only imagine if you,
link |
if your starting point is, you know, Python and TensorFlow at age 13
link |
where you end up, you know, after 10 or 15 years of that,
link |
that's, that's pretty cool. Because of GitHub, because the state tools for
link |
solving most of the major problems in artificial intelligence
link |
are within a few lines of code for most kids.
link |
And that's incredible to think about also on the entrepreneurial side.
link |
And, and on that point, was there any thought about entrepreneurship before
link |
you came to college? Is sort of doing, you're building this
link |
into a thing that impacts the world on a large scale? Yeah. I've always
link |
wanted to start a company. I think that's, you know, just a cool concept of
link |
creating something and exchanging it for value or creating value, I guess.
link |
So in high school, I was, I was trying to build like, you know, servo motor
link |
drivers, little circuit boards and sell them online
link |
or other, other things like that. And certainly knew at some point I wanted to
link |
do a startup, but it wasn't really, I'd say until college, until I felt
link |
I guess the right combination of the environment, the smart people around you
link |
and some free time and a lot of free time at MIT.
link |
So you came to MIT as an undergrad 2004. That's right. And that's when the first
link |
DARPA Grand Challenge was happening. Yeah. The, the timing of that is
link |
beautifully poetic. So how did you get yourself involved in that one?
link |
Originally there wasn't a official entry. Yeah, faculty sponsored thing. And so a
link |
bunch of undergrads, myself included, started meeting and got together and
link |
tried to haggle together some sponsorships. We got a vehicle donated,
link |
a bunch of sensors and tried to put something together. And so we had,
link |
our team was probably mostly freshmen and sophomores, you know, which, which was
link |
not really a fair, fair fight against maybe the,
link |
you know, postdoc and faculty led teams from other schools. But
link |
we, we got something up and running. We had our vehicle drive by wire and
link |
you know, very, very basic control and things. But
link |
on the day of the qualifying, sort of pre qualifying round, the one and
link |
only steering motor that we had purchased,
link |
the thing that we had retrofitted to turn the steering wheel on the truck
link |
died. And so our vehicle was just dead in the water, couldn't steer.
link |
So we didn't make it very far. On the hardware side. So was there a software
link |
component? Was there, like, how did your view of autonomous
link |
vehicles in terms of artificial intelligence
link |
evolve in this moment? I mean, you know, like you said from the 80s has been
link |
autonomous vehicles, but really that was the birth of the modern wave.
link |
The, the thing that captivated everyone's imagination that we can actually do this.
link |
So what, how were you captivated in that way?
link |
So how did your view of autonomous vehicles change at that point?
link |
I'd say at that point in time it was, it was a
link |
curiosity as in, like, is this really possible?
link |
And I think that was generally the spirit and
link |
the purpose of that original DARPA Grand Challenge, which was to just get
link |
a whole bunch of really brilliant people exploring the space and pushing the
link |
limits. And I think, like, to this day that
link |
DARPA Challenge with its, you know, million dollar prize pool
link |
was probably one of the most effective, you know, uses of taxpayer
link |
money dollar for dollar that I've seen, you know, because that,
link |
that small sort of initiative that DARPA put,
link |
put out sort of, in my view, was the catalyst or the tipping point
link |
for this, this whole next wave of autonomous vehicle development. So that
link |
was pretty cool. So let me jump around a little bit on that point.
link |
They also did the Urban Challenge where it was in the city, but it was very
link |
artificial and there's no pedestrians and there's very little human
link |
involvement except a few professional drivers. Yeah.
link |
Do you think there's room, and then there was the Robotics Challenge with
link |
humanoid robots. Right. So in your now role is looking at this,
link |
you're trying to solve one of the, you know, autonomous driving, one of the
link |
harder, more difficult places in San Francisco.
link |
Is there a role for DARPA to step in to also kind of help out,
link |
like, challenge with new ideas, specifically pedestrians and so on, all
link |
these kinds of interesting things? Well, I haven't, I haven't thought about it
link |
from that perspective. Is there anything DARPA could do today to further
link |
accelerate things? And I would say, my instinct is that that's maybe not the
link |
highest and best use of their resources and time,
link |
because, like, kick starting and spinning up the flywheel is, I think, what
link |
what they did in this case for very, very little money. But today this has become,
link |
this has become, like, commercially interesting to very large companies and
link |
the amount of money going into it and the amount of
link |
people, like, going through your class and learning about these things and
link |
developing these skills is just, you know, orders of magnitude
link |
more than it was back then. And so there's enough momentum and inertia
link |
and energy and investment dollars into this space right now that
link |
I don't, I don't, I think they're, I think they're, they can just say mission
link |
accomplished and move on to the next area of technology that needs help.
link |
So then stepping back to MIT, you left MIT during your junior year.
link |
What was that decision like? As I said, I always wanted to do
link |
a company in, or start a company, and this opportunity landed in my lap, which
link |
was a couple guys from Yale were starting a
link |
new company, and I googled them and found that they had
link |
started a company previously and sold it actually on eBay for
link |
about a quarter million bucks, which was a pretty interesting story, but
link |
so I thought to myself, these guys are, you know, rock star entrepreneurs, they've
link |
done this before, they must be driving around in Ferraris
link |
because they sold their company, and, you know, I thought I could learn
link |
a lot from them, so I teamed up with those guys and,
link |
you know, went out during, went out to California during IAP, which is MIT's
link |
month off, on a one way ticket and basically never went back.
link |
We were having so much fun, we felt like we were building something and creating
link |
something, and it was going to be interesting
link |
that, you know, I was just all in and got completely hooked, and that
link |
that business was Justin TV, which is originally a reality show about a guy
link |
named Justin, which morphed into a live video
link |
streaming platform, which then morphed into what is Twitch
link |
today, so that was, that was quite an unexpected journey.
link |
So no regrets? No. Looking back, it was just an obvious, I mean,
link |
one way ticket. I mean, if we just pause on that for a second,
link |
there was no, how did you know these are the right guys, this is the
link |
right decision, you didn't think it was just follow the
link |
heart kind of thing? Well, I didn't know, but, you know, just trying something for a
link |
month during IAP seems pretty low risk, right?
link |
And then, you know, well, maybe I'll take a semester off, MIT's pretty flexible
link |
about that, you can always go back, right? And then after two or three cycles of
link |
that, I eventually threw in the towel, but, you know, I think it's,
link |
I guess in that case I felt like I could always hit the undo button if I had to.
link |
Right. But nevertheless, from when you look
link |
in retrospect, I mean, it seems like a brave decision,
link |
you know, it would be difficult for a lot of people to make. It wasn't as
link |
popular, I'd say that the general, you know, flux of people
link |
out of MIT at the time was mostly into, you know, finance or consulting jobs in
link |
Boston or New York, and very few people were going to
link |
California to start companies, but today I'd say that's,
link |
it's probably inverted, which is just a sign of,
link |
a sign of the times, I guess. Yeah. So there's a story about
link |
midnight of March 18, 2007, where TechCrunch, I guess, announced
link |
Justin.TV earlier than it was supposed to, a few hours.
link |
The site didn't work. I don't know if any of this is true, you can tell me.
link |
And you and one of the folks at Justin.TV,
link |
Emmett Shearer, coded through the night. Can you take me through that experience?
link |
So let me, let me say a few nice things that,
link |
the article I read quoted Justin Kahn said that you were known for bureau
link |
coding through problems and being a creative, quote, creative
link |
genius. So on that night,
link |
what, what was going through your head, or maybe I'd put another way,
link |
how do you solve these problems? What's your approach to solving these kinds of
link |
problems where the line between success and failure seems to be pretty
link |
thin? That's a good question. Well, first of all, that's, that's a
link |
nice of Justin to say that. I think, you know, I would have been
link |
maybe 21 years old then and not very experienced at programming,
link |
but as with, with everything in a startup, you're sort of racing against
link |
the clock. And so our plan was the second we had
link |
this live streaming camera backpack up and running, where Justin could wear it
link |
and no matter where he went in a city, it
link |
would be streaming live video. And this is
link |
even before the iPhones. This is like hard to do back then.
link |
We would launch. And so we thought we were there and the backpack was working
link |
and then we sent out all the emails to launch the,
link |
launch the company and do the press thing. And then, you know,
link |
we weren't quite actually there. And then
link |
we thought, oh, well, you know, they're not going to announce it until
link |
maybe 10 a.m. the next morning. And it's, I don't know, it's 5 p.m. now. So
link |
how many hours do we have left? What is that? Like, you know, 17 hours to go.
link |
And, and that was, that was going to be fine.
link |
Was the problem obvious? Did you understand
link |
what could possibly, like, how complicated was the system at that point?
link |
It was, it was pretty messy. So to get a live video feed that looked decent
link |
working from anywhere in San Francisco, I put together this system where we had
link |
like three or four cell phone data modems and
link |
they were, like, we take the video stream and,
link |
you know, sort of spray it across these three or four modems and then try to
link |
catch all the packets on the other side, you know, with unreliable cell phone
link |
networks. It's pretty low level networking.
link |
Yeah, and putting these, like, you know, sort of
link |
protocols on top of all that to, to reassemble and reorder the packets and
link |
have time buffers and error correction and all that kind of stuff.
link |
And the night before it was just staticky. Every once in a while the image
link |
would, would go to staticky and there would be this horrible,
link |
like, screeching audio noise because the audio was also corrupted.
link |
And this would happen, like, every five to ten minutes or so and it was
link |
a really, you know, off putting to the viewers.
link |
How do you tackle that problem? What was the, uh, you're just freaking out behind a
link |
computer. There's, are there other, other folks working
link |
on this problem? Like, were you behind a whiteboard? Were you doing, uh,
link |
Yeah, it was a little, it was a little, yeah, it's a little lonely because there's four of us
link |
working on the company and only two people really wrote code.
link |
And Emmett wrote the website and the chat system and I wrote the
link |
software for this video streaming device and video server.
link |
And so, you know, it's my sole responsibility to figure that out.
link |
And I think, I think it's those, you know, setting,
link |
setting deadlines, trying to move quickly and everything where you're in that
link |
moment of intense pressure that sometimes people do their
link |
best and most interesting work. And so even though that was a terrible moment,
link |
I look back on it fondly because that's like, you know, that's one of those
link |
character defining moments, I think. So in 2013, October, you founded
link |
Cruise Automation. Yeah. So progressing forward, another
link |
exceptionally successful company was acquired by GM in 16
link |
for $1 billion. But in October 2013, what was on your mind?
link |
What was the plan? How does one seriously start to tackle
link |
one of the hardest robotics, most important impact for robotics
link |
problems of our age? After going through Twitch, Twitch was,
link |
was, and is today, pretty successful. But the, the work was,
link |
the result was entertainment, mostly. Like, the better the product was,
link |
the more we would entertain people and then, you know, make money on the ad
link |
revenues and other things. And that was, that was a good thing. It
link |
felt, felt good to entertain people. But I figured like, you know, what is really
link |
the point of becoming a really good engineer and
link |
developing these skills other than, you know, my own enjoyment? And I
link |
realized I wanted something that scratched more of an existential
link |
itch, like something that, that truly matters. And so I
link |
basically made this list of requirements for a new, if I was going to
link |
do another company, and the one thing I knew in the back of
link |
my head that Twitch took like eight years to become successful.
link |
And so whatever I do, I better be willing to commit, you know, at least 10 years
link |
to something. And when you think about things from that perspective,
link |
you certainly, I think, raise the bar on what you choose to work on. So for me,
link |
the three things were it had to be something where the technology
link |
itself determines the success of the product,
link |
like hard, really juicy technology problems, because that's what
link |
motivates me. And then it had to have a direct and positive impact on society in
link |
some way. So an example would be like, you know,
link |
health care, self driving cars, because they save lives, other things where
link |
there's a clear connection to somehow improving other people's lives.
link |
And the last one is it had to be a big business, because
link |
for the positive impact to matter, it's got to be a large scale.
link |
And I was thinking about that for a while, and I made like, I tried
link |
writing a Gmail clone and looked at some other ideas.
link |
And then it just sort of light bulb went off, like self driving cars, like that
link |
was the most fun I had ever had in college working on that.
link |
And like, well, what's the state of the technology? It's been 10 years.
link |
Maybe times have changed, and maybe now is the time to make this work.
link |
And I poked around and looked at, the only other thing out there really at the
link |
time was the Google self driving car project.
link |
And I thought, surely there's a way to, you know, have an entrepreneur mindset
link |
and sort of solve the minimum viable product here.
link |
And so I just took the plunge right then and there and said, this is something I
link |
know I can commit 10 years to. It's the probably the greatest
link |
applied AI problem of our generation. And if it works, it's going to be both a
link |
huge business and therefore like, probably the most positive impact I can
link |
possibly have on the world. So after that light bulb went off, I went
link |
all in on cruise immediately and got to work.
link |
Did you have an idea how to solve this problem? Which aspect of the problem to
link |
solve? You know, slow, like we just had Oliver
link |
from Voyage here, slow moving retirement communities,
link |
urban driving, highway driving. Did you have, like, did you have a vision of the
link |
city of the future where, you know, the transportation is
link |
largely automated, that kind of thing? Or was it sort of
link |
more fuzzy and gray area than that? My analysis of the situation is that
link |
Google is putting a lot, had been putting a lot of money into that project. They
link |
had a lot more resources. And so, and they still hadn't cracked
link |
the fully driverless car. You know, this is 2013, I guess.
link |
So I thought, what can I do to sort of go from zero to,
link |
you know, significant scale so I can actually solve the real problem, which is
link |
the driverless cars. And I thought, here's the strategy. We'll
link |
start by doing a really simple problem or solving a
link |
really simple problem that creates value for people. So,
link |
eventually ended up deciding on automating highway driving,
link |
which is relatively more straightforward as long as there's a
link |
backup driver there. And, you know, the go to market will be able to retrofit
link |
people's cars and just sell these products directly. And
link |
the idea was, we'll take all the revenue and profits from that and use it
link |
to do the, so sort of reinvest that in research for
link |
doing fully driverless cars. And that was the plan.
link |
The only thing that really changed along the way between then and now is
link |
we never really launched the first product. We had enough interest from
link |
investors and enough of a signal that this was
link |
something that we should be working on, that after about a year of working on
link |
the highway autopilot, we had it working, you know, on a
link |
prototype stage. But we just completely abandoned that
link |
and said, we're going to go all in on driverless cars now is the time.
link |
Can't think of anything that's more exciting and if it works more impactful,
link |
so we're just going to go for it. The idea of retrofit is kind of
link |
interesting. Yeah. Being able to, it's how you achieve scale.
link |
It's a really interesting idea. Is it something that's still in the
link |
in the back of your mind as a possibility?
link |
Not at all. I've come full circle on that one. After
link |
trying to build a retrofit product, and I'll touch on some of the complexities
link |
of that, and then also having been inside an OEM
link |
and seeing how things work and how a vehicle is developed and
link |
validated. When it comes to something that has
link |
safety critical implications like controlling the steering and
link |
other control inputs on your car, it's pretty hard to get there
link |
with a retrofit. Or if you did, even if you did, it creates a whole bunch
link |
of new complications around liability or how did you truly validate
link |
that. Or you know, something in the base vehicle fails and causes your system to
link |
fail, whose fault is it?
link |
Or if the car's anti lock brake systems or other things kick in
link |
or the software has been, it's different in one version of the car you retrofit
link |
versus another and you don't know because
link |
the manufacturer has updated it behind the scenes. There's basically an
link |
infinite list of long tail issues that can get you.
link |
And if you're dealing with a safety critical product, that's not really
link |
acceptable. That's a really convincing summary of why
link |
that's really challenging. But I didn't know all that at the time, so we tried it
link |
anyway. But as a pitch also at the time, it's a
link |
really strong one. Because that's how you achieve scale and that's how you beat
link |
the current, the leader at the time of Google or the only one in the market.
link |
The other big problem we ran into, which is perhaps the biggest problem from a
link |
business model perspective, is we had kind of assumed that we
link |
started with an Audi S4 as the vehicle we
link |
retrofitted with this highway driving capability.
link |
And we had kind of assumed that if we just knock out like three
link |
make and models of vehicles, that'll cover like 80% of the San Francisco
link |
market. Doesn't everyone there drive, I don't know, a BMW or a Honda Civic or
link |
one of these three cars? And then we surveyed our users and we found out that
link |
it's all over the place. We would, to get even a decent number of
link |
units sold, we'd have to support like, you know, 20 or 50 different models.
link |
And each one is a little butterfly that takes time and effort to maintain,
link |
you know, that retrofit integration and custom hardware and all this.
link |
So it was a tough business. So GM manufactures and sells over 9 million
link |
cars a year. And what you with Cruise are trying to do
link |
some of the most cutting edge innovation in terms of applying AI.
link |
And so how do those, you've talked about a little bit before, but it's also just
link |
fascinating to me. We work a lot of automakers,
link |
you know, the difference between the gap between Detroit
link |
and Silicon Valley, let's say, just to be sort of poetic about it, I guess.
link |
How do you close that gap? How do you take GM into the future
link |
where a large part of the fleet will be autonomous, perhaps?
link |
I want to start by acknowledging that GM is made up of,
link |
you know, tens of thousands of really brilliant, motivated people who want to
link |
be a part of the future. And so it's pretty fun to work
link |
within the attitude inside a car company like that is, you
link |
know, embracing this transformation and change
link |
rather than fearing it. And I think that's a testament to
link |
the leadership at GM and that's flown all the way through to everyone you
link |
talk to, even the people in the assembly plants working on these cars.
link |
So that's really great. So starting from that position makes it a lot easier
link |
so then when the people in San Francisco at Cruise
link |
interact with the people at GM, at least we have this common set of values, which
link |
is that we really want this stuff to work
link |
because we think it's important and we think it's the future.
link |
That's not to say, you know, those two cultures don't clash. They absolutely do.
link |
There's different sort of value systems. Like in a
link |
car company, the thing that gets you promoted and sort of the reward
link |
system is following the processes, delivering
link |
the program on time and on budget. So any sort of risk taking
link |
is discouraged in many ways because if a program is late or if you shut down
link |
the plant for a day, it's, you know, you can count the millions of dollars that
link |
burn by pretty quickly. Whereas I think, you know, most Silicon
link |
Valley companies and in Cruise and the methodology
link |
we were employing, especially around the time of the acquisition,
link |
the reward structure is about trying to solve
link |
these complex problems in any way shape or form or coming up with crazy ideas
link |
that, you know, 90% of them won't work. And so meshing that culture
link |
of sort of continuous improvement and experimentation
link |
with one where everything needs to be rigorously defined up front so that
link |
you never slip a deadline or miss a budget
link |
was a pretty big challenge. And that we're over three years in now
link |
after the acquisition and I'd say like, you know, the investment we made in
link |
figuring out how to work together successfully and
link |
who should do what and how we bridge the gaps between these
link |
very different systems and way of doing engineering work
link |
is now one of our greatest assets because I think we have this really
link |
powerful thing. But for a while it was both GM and Cruise were very
link |
steep on the learning curve. Yeah, so I'm sure it was very stressful.
link |
It's really important work because that's how
link |
to revolutionize the transportation, really to revolutionize
link |
any system. You know, you look at the health care system or you look at the
link |
legal system. I have people like Loris come up to me all the time like
link |
everything they're working on can easily be automated.
link |
But then that's not a good feeling. Yeah, well it's not a good feeling but also
link |
there's no way to automate because the entire infrastructure is really,
link |
you know, based is older and it moves very slowly. And so
link |
how do you close the gap between I have an
link |
how can I replace, of course, Loris don't want to be replaced with an app, but you
link |
could replace a lot of aspect when most of the data is still on paper.
link |
And so the same thing was with automotive.
link |
I mean, it's fundamentally software. It's basically hiring software
link |
engineers. It's thinking in a software world.
link |
I mean, I'm pretty sure nobody in Silicon Valley has ever hit a deadline.
link |
So and then on GM. That's probably true, yeah. And GM side is probably the
link |
opposite. Yeah. So that's that culture gap is really fascinating.
link |
So you're optimistic about the future of that?
link |
Yeah, I mean, from what I've seen, it's impressive. And I think like
link |
especially in Silicon Valley, it's easy to write off building cars because,
link |
you know, people have been doing that for over 100 years now in this country. And
link |
so it seems like that's a solved problem, but that doesn't mean it's an easy
link |
problem. And I think it would be easy to sort of
link |
overlook that and think that, you know, we're
link |
Silicon Valley engineers. We can solve any problem, you know, building a car.
link |
It's been done. Therefore, it's, you know, it's not a real
link |
engineering challenge. But after having seen just the sheer
link |
scale and magnitude and industrialization
link |
that occurs inside of an automotive assembly plant, that is a lot of work
link |
that I am very glad that we don't have to reinvent
link |
to make self driving cars work. And so to have, you know, partners who have done
link |
that for 100 years now, these great processes and this huge infrastructure
link |
and supply base that we can tap into is just remarkable
link |
because the scope and surface area of
link |
the problem of deploying fleets of self driving cars is so large
link |
that we're constantly looking for ways to do less
link |
so we can focus on the things that really matter more. And if we had to
link |
figure out how to build and assemble and
link |
you know, build the cars themselves. I mean, we work closely with GM on
link |
that. But if we had to develop all that capability in house as well,
link |
you know, that would just make the problem really intractable, I think.
link |
So yeah, just like your first entry at the MIT DARPA challenge when
link |
there was what the motor that failed, somebody that knows what they're
link |
doing with the motor did it. That would have been nice if we could
link |
focus on the software, not the hardware platform.
link |
Yeah. Right. So from your perspective now,
link |
you know, there's so many ways that autonomous vehicles can impact
link |
society in the next year, five years, ten years.
link |
What do you think is the biggest opportunity to make
link |
money in autonomous driving, sort of make it a
link |
financially viable thing in the near term?
link |
What do you think will be the biggest impact there?
link |
Well, the things that drive the economics for fleets of self driving
link |
cars are, there's sort of a handful of variables. One is,
link |
you know, the cost to build the vehicle itself. So the material cost, how many,
link |
you know, what's the cost of all your sensors plus the cost of the vehicle and
link |
every all the other components on it. Another one is the lifetime of the
link |
vehicle. It's very different if your vehicle
link |
drives 100,000 miles and then it falls apart versus,
link |
you know, two million. And then, you know, if you have a fleet, it's
link |
kind of like an airplane or an airline where
link |
once you produce the vehicle, you want it to be in
link |
operation as many hours a day as possible producing revenue.
link |
And then, you know, the other piece of that is
link |
how are you generating revenue? I think that's kind of what you're asking. And I
link |
think the obvious things today are, you know, the ride sharing business
link |
because that's pretty clear that there's demand for that,
link |
there's existing markets you can tap into and
link |
large urban areas, that kind of thing. Yeah, yeah. And I think that there are
link |
some real benefits to having cars without
link |
drivers compared to sort of the status quo for
link |
people who use ride share services today.
link |
You know, you get privacy, consistency, hopefully significantly improve safety,
link |
all these benefits versus the current product.
link |
But it's a crowded market. And then other opportunities, which you've
link |
seen a lot of activity in the last, really in the last six or twelve months,
link |
is, you know, delivery, whether that's parcels and packages,
link |
food or groceries. Those are all sort of, I think, opportunities that are
link |
pretty ripe for these, you know, once you have this
link |
core technology, which is the fleet of autonomous vehicles, there's all sorts of
link |
different business opportunities you can build on
link |
top of that. But I think the important thing, of course, is that
link |
there's zero monetization opportunity until you actually have that fleet of
link |
very capable driverless cars that are that are as good or better than humans.
link |
And that's sort of where the entire industry is
link |
sort of in this holding pattern right now.
link |
Yeah, they're trying to achieve that baseline. So, but you said sort of
link |
not reliability, consistency. It's kind of interesting, I think I heard you say
link |
somewhere, I'm not sure if that's what you meant, but
link |
you know, I can imagine a situation where you would get an autonomous vehicle
link |
and, you know, when you get into an Uber or Lyft,
link |
you don't get to choose the driver in a sense that you don't get to choose the
link |
personality of the driving. Do you think there's a, there's room
link |
to define the personality of the car the way it drives you in terms of
link |
aggressiveness, for example, in terms of sort of pushing the
link |
bound? One of the biggest challenges of autonomous driving is the
link |
is the trade off between sort of safety and
link |
assertiveness. And do you think there's any room
link |
for the human to take a role in that decision? Sort of accept some of the
link |
liability, I guess. I wouldn't, no, I'd say within
link |
reasonable bounds, as in we're not gonna, I think it'd be
link |
highly unlikely we'd expose any knob that would let you, you know, significantly
link |
increase safety risk. I think that's just not
link |
something we'd be willing to do. But I think driving style or like, you
link |
know, are you going to relax the comfort constraints slightly or things like that,
link |
all of those things make sense and are plausible. I see all those as, you know,
link |
nice optimizations. Once again, we get the core problem solved in these fleets
link |
out there. But the other thing we've sort of
link |
observed is that you have this intuition that if you
link |
sort of slam your foot on the gas right after the light turns green and
link |
aggressively accelerate, you're going to get there faster. But the
link |
actual impact of doing that is pretty small.
link |
You feel like you're getting there faster, but so the same would be
link |
true for AVs. Even if they don't slam their, you know, the pedal to the floor
link |
when the light turns green, they're going to get you there within, you
link |
know, if it's a 15 minute trip, within 30 seconds of what you would have done
link |
otherwise if you were going really aggressively.
link |
So I think there's this sort of self deception that my aggressive
link |
driving style is getting me there faster.
link |
Well, so that's, you know, some of the things I've studied, some of the things
link |
I'm fascinated by the psychology of that. I don't think it matters
link |
that it doesn't get you there faster. It's the emotional release.
link |
Driving is a place, being inside of a car,
link |
somebody said it's like the real world version of being a troll.
link |
So you have this protection, this mental protection, you're able to sort of yell
link |
at the world, like release your anger, whatever.
link |
So there's an element of that that I think autonomous vehicles would also
link |
have to, you know, giving an outlet to people, but it doesn't have to be
link |
through, through, through driving or honking or so on.
link |
There might be other outlets, but I think to just sort of even just put that aside,
link |
the baseline is really, you know, that's the focus.
link |
That's the thing you need to solve.
link |
And then the fun human things can be solved after.
link |
But so from the baseline of just solving autonomous driving, you're working in
link |
San Francisco, one of the more difficult cities to operate in.
link |
What is, what is the, in your view, currently the hardest
link |
aspect of autonomous driving?
link |
Is it negotiating with pedestrians?
link |
Is it edge cases of perception?
link |
Is there a mechanical engineering?
link |
Is it data, fleet stuff?
link |
What are your thoughts on the challenge, the more challenging aspects there?
link |
That's a, that's a good question.
link |
I think before, before we go to that, though, I just want to, I like what you
link |
said about the psychology aspect of this,
link |
because I think one observation I've made is I think I read somewhere that I
link |
think it's maybe Americans on average spend, you know, over an hour a day on
link |
social media, like staring at Facebook.
link |
And so that's just, you know, 60 minutes of your life, you're not getting back.
link |
It's probably not super productive.
link |
And so that's 3,600 seconds, right?
link |
And that's, that's time, you know, it's a lot of time you're giving up.
link |
And if you compare that to people being on the road, if another vehicle, whether
link |
it's a human driver or autonomous vehicle, delays them by even three
link |
seconds, they're laying in on the horn, you know, even though that's, that's, you
link |
know, one, one thousandth of the time they waste looking at Facebook every day.
link |
So there's, there's definitely some.
link |
You know, psychology aspects of this, I think that are pretty interesting road
link |
And then the question of course is if everyone is in self driving cars,
link |
do they even notice these three second delays anymore?
link |
Cause they're doing other things or reading or working or just talking to
link |
So it'll be interesting to see where that goes.
link |
In a certain aspect, people, people need to be distracted by something
link |
entertaining, something useful inside the car.
link |
So they don't pay attention to the external world.
link |
And then, and then they can take whatever psychology and bring it back to
link |
Twitter and then focus on that as opposed to sort of interacting, sort of putting
link |
the emotion out there into the world.
link |
So it's a, it's an interesting problem, but baseline autonomy.
link |
I guess you could say self driving cars, you know, at scale will lower the
link |
collective blood pressure of society probably by a couple of points without
link |
all that road rage and stress.
link |
So that's a good, good external.
link |
So back to your question about the technology and the, I guess the biggest
link |
And I have a hard time answering that question because, you know, we've been
link |
at this like specifically focusing on driverless cars and all the technology
link |
needed to enable that for a little over four and a half years now.
link |
And even a year or two in, I felt like we had completed the functionality needed
link |
to get someone from point A to point B.
link |
As in, if we need to do a left turn maneuver, or if we need to drive around
link |
at, you know, a double parked vehicle into oncoming traffic or navigate
link |
through construction zones, the scaffolding and the building blocks was
link |
there pretty early on.
link |
And so the challenge is not any one scenario or situation for which, you
link |
know, we fail at 100% of those.
link |
It's more, you know, we're benchmarking against a pretty good or pretty high
link |
standard, which is human driving.
link |
All things considered, humans are excellent at handling edge cases and
link |
unexpected scenarios where computers are the opposite.
link |
And so beating that baseline set by humans is the challenge.
link |
And so what we've been doing for quite some time now is basically, it's this
link |
continuous improvement process where we find sort of the most, you know,
link |
uncomfortable or the things that could lead to a safety issue or other
link |
things, all these events.
link |
And then we sort of categorize them and rework parts of our system to make
link |
incremental improvements and do that over and over and over again.
link |
And we just see sort of the overall performance of the system, you know,
link |
actually increasing in a pretty steady clip.
link |
But there's no one thing.
link |
There's actually like thousands of little things and just like polishing functionality
link |
and making sure that it handles, you know, every version and possible
link |
permutation of a situation by either applying more deep learning systems or
link |
just by, you know, adding more test coverage or new scenarios that we
link |
develop against and just grinding on that.
link |
We're sort of in the unsexy phase of development right now, which is doing
link |
the real engineering work that it takes to go from prototype to production.
link |
You're basically scaling the grinding, sort of taking seriously that the
link |
process of all those edge cases, both with human experts and machine
link |
learning methods to cover all those situations.
link |
And the exciting thing for me is I don't think that grinding ever stops because
link |
there's a moment in time where you've crossed that threshold of human
link |
performance and become superhuman.
link |
But there's no reason, there's no first principles reason that AV capability
link |
will tap out anywhere near humans.
link |
Like there's no reason it couldn't be 20 times better, whether that's, you
link |
know, just better driving or safer driving or more comfortable driving or
link |
even a thousand times better given enough time.
link |
And we intend to basically chase that, you know, forever to build the best
link |
Better and better and better.
link |
And always new edge cases come up and new experiences.
link |
So, and you want to automate that process as much as possible.
link |
So what do you think in general in society?
link |
When do you think we may have hundreds of thousands of fully autonomous
link |
vehicles driving around?
link |
So first of all, predictions, nobody knows the future.
link |
You're a part of the leading people trying to define that future, but even
link |
then you still don't know.
link |
But if you think about hundreds of thousands of vehicles, so a significant
link |
fraction of vehicles in major cities are autonomous.
link |
Do you think, are you with Rodney Brooks, who is 2050 and beyond, or are you
link |
more with Elon Musk, who is, we should have had that two years ago?
link |
Well, I mean, I'd love to have it two years ago, but we're not there yet.
link |
So I guess the way I would think about that is let's flip that question
link |
So what would prevent you to reach hundreds of thousands of vehicles?
link |
And that's a good, that's a good rephrasing.
link |
So the, I'd say the, it seems the consensus among the people developing
link |
self driving cars today is to sort of start with some form of an easier
link |
environment, whether it means, you know, lacking inclement weather or, you
link |
know, mostly sunny or whatever it is.
link |
And then add, add capability for more complex situations over time.
link |
And so if you're only able to deploy in areas that meet sort of your
link |
criteria or the current domain, you know, operating domain of the
link |
software you developed, that may put a cap on how many cities you could
link |
But then as those restrictions start to fall away, like maybe you add
link |
capability to drive really well and safely in heavy rain or snow, you
link |
know, that, that probably opens up the market by two, two or three fold
link |
in terms of the cities you can expand into and so on.
link |
And so the real question is, you know, I know today if we wanted to, we
link |
could produce that, that many autonomous vehicles, but we wouldn't be
link |
able to make use of all of them yet.
link |
Cause we would sort of saturate the demand in the cities in which we
link |
would want to operate initially.
link |
So if I were to guess like what the timeline is for those things falling
link |
away and reaching hundreds of thousands of vehicles, I would say that
link |
thousands of vehicles, maybe a range is better, I would say less than
link |
five years, less than five years.
link |
And of course you're working hard to make that happen.
link |
So you started two companies that were eventually acquired for each
link |
four billion dollars.
link |
So you're a pretty good person to ask, what does it take to build a
link |
successful startup?
link |
I think there's, there's sort of survivor bias here a little bit, but
link |
I can try to find some common threads for the things that worked for
link |
me, which is, you know, in, in both of these companies, I was really
link |
passionate about the core technology.
link |
I actually like, you know, lay awake at night thinking about these
link |
problems and how to solve them.
link |
And I think that's helpful because when you start a business, there
link |
are like to this day, there are these crazy ups and downs.
link |
Like one day you think the business is just on, you're just on top of
link |
the world and unstoppable.
link |
And the next day you think, okay, this is all going to end, you know,
link |
it's just, it's just going south and it's going to be over tomorrow.
link |
And and so I think like having a true passion that you can fall back
link |
on and knowing that you would be doing it, even if you weren't getting
link |
paid for it, helps you weather those, those tough times.
link |
So that's one thing.
link |
I think the other one is really good people.
link |
So I've always been surrounded by really good cofounders that are
link |
logical thinkers are always pushing their limits and have very high
link |
levels of integrity.
link |
So that's Dan Kahn and my current company and actually his brother and
link |
a couple other guys for Justin TV and Twitch.
link |
And then I think the last thing is just I guess persistence or
link |
perseverance, like, and, and that, that can apply to sticking to sort
link |
of, or having conviction around the original premise of your idea and
link |
sticking around to do all the, you know, the unsexy work to actually
link |
make it come to fruition, including dealing with, you know, whatever
link |
it is that you, that you're not passionate about, whether that's
link |
finance or, or HR or, or operations or those things, as long as you
link |
are grinding away and working towards, you know, that North star
link |
for your business, whatever it is, and you don't give up and you're
link |
making progress every day, it seems like eventually you'll end up in a
link |
And the only things that can slow you down are, you know, running out
link |
of money or I suppose your competitors destroying you.
link |
But I think most of the time it's, it's people giving up or, or somehow
link |
destroying things themselves rather than being beaten by their competition
link |
or running out of money.
link |
If you never quit, eventually you'll arrive.
link |
So, uh, it's a much more concise version of what I was trying to say.
link |
Yeah, that was good.
link |
So you went the Y Combinator route twice.
link |
What do you think in a quick question, do you think is the best way to
link |
raise funds in the early days or not just funds, but just community
link |
develop your idea and so on.
link |
Can you do it solo or maybe with a co founder with like self funded?
link |
Do you think Y Combinator is good?
link |
Is it good to do VC route?
link |
Is there no right answer or is there from the Y Combinator experience
link |
something that you could take away that that was the right path to take?
link |
There's no one size fits all answer, but if your ambition I think is to, you
link |
know, see how big you can make something or, or, or rapidly expand and capture
link |
a market or solve a problem or whatever it is, then, then, you know, going to
link |
venture back route is probably a good approach so that, so that capital doesn't
link |
become your primary constraint.
link |
Y Combinator I love because it puts you in this, uh, sort of competitive
link |
environment where you're, where you're surrounded by, you know, the top, maybe
link |
1% of other really highly motivated, you know, peers who are in the same, same
link |
place and that, uh, that environment I think just breeds breed success, right?
link |
If you're surrounded by really brilliant, hardworking people, you're going to
link |
feel, you know, sort of compelled or inspired to, to try to emulate them and
link |
And, uh, so even though I had done it once before and I felt like, yeah, I'm
link |
pretty self motivated.
link |
I thought like, look, this is going to be a hard problem.
link |
I can use all the help I can get.
link |
So surrounding myself with other entrepreneurs is going to make me work a
link |
little bit harder or push a little harder than it's worth it.
link |
And so that's why I, why I did it, you know, for example, the second time.
link |
Let's, uh, let's go philosophical existential.
link |
If you go back and do something differently in your life, starting in the
link |
high school and MIT leaving MIT, you could have gone the PhD route doing the
link |
startup, going to see about a startup in California and you, or maybe some
link |
aspects of fundraising.
link |
Is there something you regret, something you not necessarily regret, but if
link |
you go back, you could do differently.
link |
I think I've made a lot of mistakes, like, you know, pretty much everything
link |
I think I've screwed up at least once, but I, you know, I don't regret those
link |
I think it's, it's hard to, it's hard to look back on things, even if it didn't
link |
go well and call it a regret, because hopefully it took away some new knowledge
link |
or learning from that.
link |
So I would say there was a period.
link |
The closest I can, I can come to is there's a period, um, in, in Justin
link |
TV, I think after seven years where, you know, the company was going one
link |
direction, which is towards Twitch, uh, in video gaming.
link |
I'm not a video gamer.
link |
I don't really even use Twitch at all.
link |
And I was still, uh, working on the core technology there, but my, my heart
link |
was no longer in it because the business that we were creating was not something
link |
that I was personally passionate about.
link |
It didn't meet your bar of existential impact.
link |
And I'd say I probably spent an extra year or two working on that.
link |
And, uh, and I'd say like, I would have just tried to do something different
link |
sooner because those, those were two years where I felt like, um, you know,
link |
from this philosophical or existential thing, I just, I just felt that
link |
something was missing.
link |
And so I would have, I would have, if I could look back now and tell myself,
link |
it's like, I would have said exactly that.
link |
Like, you're not getting any meaning out of your work personally right now.
link |
You should, you should find a way to change that.
link |
And that's, that's part of the pitch I use to basically everyone who joins
link |
Cruise today, it's like, Hey, you've got that now by coming here.
link |
Well, maybe you needed the two years of that existential dread to develop
link |
the feeling that ultimately it was the fire that created Cruise.
link |
So, you never know.
link |
You can't, good theory.
link |
So last question, what does 2019 hold for Cruise?
link |
After this, I guess we're going to go and I'll talk to your class.
link |
But one of the big things is going from prototype to production, uh, for
link |
autonomous cars and what does that mean?
link |
What does that look like?
link |
And 2019 for us is the year that we try to cross over that threshold and reach,
link |
you know, superhuman level of performance to some degree with the software and,
link |
uh, have all the other of the thousands of little building blocks in place to,
link |
to launch, um, you know, our, our first, uh, commercial product.
link |
So that's, that's, what's in store for us or in store for us.
link |
And we've got a lot of work to do.
link |
We've got a lot of brilliant people working on it.
link |
So it's, it's all up to us now.
link |
From Charlie Miller and Chris Vells, like the people I've crossed paths with.
link |
If you, it sounds like you have an amazing team.
link |
So, um, like I said, it's one of the most, I think one of the most important
link |
problems in artificial intelligence of the century.
link |
It'll be one of the most defining, the super exciting that you work on it.
link |
And, uh, the best of luck in 2018, I'm really excited to see what
link |
Cruz comes up with.
link |
Thanks for having me today.