episode
stringlengths
45
100
text
stringlengths
1
528
timestamp_link
stringlengths
56
56
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
hang out and talk about, you know, document their build progress and
https://karpathy.ai/lexicap/0014-large.html#00:02:12.960
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
everything. And I think I read, I must have read like,
https://karpathy.ai/lexicap/0014-large.html#00:02:16.160
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
you know, tens of thousands of forum posts from basically
https://karpathy.ai/lexicap/0014-large.html#00:02:20.080
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
everything that was out there on what these people were doing. And eventually
https://karpathy.ai/lexicap/0014-large.html#00:02:23.760
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
like sort of triangulated how to put some of these things together.
https://karpathy.ai/lexicap/0014-large.html#00:02:26.960
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
And I ended up doing BattleBots, which was, you know, I was like 13 or 14, which
https://karpathy.ai/lexicap/0014-large.html#00:02:30.400
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
was pretty awesome. I'm not sure if the show is still
https://karpathy.ai/lexicap/0014-large.html#00:02:34.880
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
running, but so BattleBots is, there's not an artificial
https://karpathy.ai/lexicap/0014-large.html#00:02:37.280
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
intelligence component. It's remotely controlled. And it's
https://karpathy.ai/lexicap/0014-large.html#00:02:41.520
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
almost like a mechanical engineering challenge of building things
https://karpathy.ai/lexicap/0014-large.html#00:02:44.800
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
that can be broken. They're radio controlled. So,
https://karpathy.ai/lexicap/0014-large.html#00:02:47.840
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
and I think that they allowed some limited form of autonomy.
https://karpathy.ai/lexicap/0014-large.html#00:02:50.960
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
But, you know, in a two minute match, you're, in the way these things ran,
https://karpathy.ai/lexicap/0014-large.html#00:02:54.720
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
you're really doing yourself a disservice by trying to automate it
https://karpathy.ai/lexicap/0014-large.html#00:02:58.640
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
versus just, you know, do the practical thing, which is drive it yourself.
https://karpathy.ai/lexicap/0014-large.html#00:03:01.120
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
And there's an entertainment aspect, just going on YouTube, there's like an,
https://karpathy.ai/lexicap/0014-large.html#00:03:04.560
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
some of them wield an axe, some of them, I mean, there's that fun.
https://karpathy.ai/lexicap/0014-large.html#00:03:08.800
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
So what drew you to that aspect? Was it the mechanical engineering?
https://karpathy.ai/lexicap/0014-large.html#00:03:12.080
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
Was it the dream to create like Frankenstein and
https://karpathy.ai/lexicap/0014-large.html#00:03:15.280
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
sentient being? Or was it just like the Lego,
https://karpathy.ai/lexicap/0014-large.html#00:03:19.760
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
you like tinkering with stuff? I mean, that was just building something.
https://karpathy.ai/lexicap/0014-large.html#00:03:22.720
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
I think the idea of, you know, this radio controlled machine that
https://karpathy.ai/lexicap/0014-large.html#00:03:25.920
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
can do various things, if it has like a weapon or something was pretty
https://karpathy.ai/lexicap/0014-large.html#00:03:29.760
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
interesting. I agree it doesn't have the same
https://karpathy.ai/lexicap/0014-large.html#00:03:32.320
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
appeal as, you know, autonomous robots, which I,
https://karpathy.ai/lexicap/0014-large.html#00:03:34.880
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
which I, you know, sort of gravitated towards later on. But it was definitely
https://karpathy.ai/lexicap/0014-large.html#00:03:37.600
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
an engineering challenge because everything you did in that
https://karpathy.ai/lexicap/0014-large.html#00:03:40.720
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
competition was pushing components to their limits. So we would
https://karpathy.ai/lexicap/0014-large.html#00:03:44.880
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
buy like these $40 DC motors that came out of a
https://karpathy.ai/lexicap/0014-large.html#00:03:49.200
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
winch, like on the front of a pickup truck or something, and we'd
https://karpathy.ai/lexicap/0014-large.html#00:03:54.400
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
power the car with those and we'd run them at like double or triple their
https://karpathy.ai/lexicap/0014-large.html#00:03:57.760
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
rated voltage. So they immediately start overheating,
https://karpathy.ai/lexicap/0014-large.html#00:04:01.200
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
but for that two minute match you can get,
https://karpathy.ai/lexicap/0014-large.html#00:04:04.000
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
you know, a significant increase in the power output of those motors
https://karpathy.ai/lexicap/0014-large.html#00:04:06.000
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
before they burn out. And so you're doing the same thing for your battery packs,
https://karpathy.ai/lexicap/0014-large.html#00:04:09.520
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
all the materials in the system. And I think there's something, something
https://karpathy.ai/lexicap/0014-large.html#00:04:12.720
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
intrinsically interesting about just seeing like where things break.
https://karpathy.ai/lexicap/0014-large.html#00:04:16.800
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
And did you offline see where they break? Did you
https://karpathy.ai/lexicap/0014-large.html#00:04:20.240
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
take it to the testing point? Like how did you know two minutes? Or was there a
https://karpathy.ai/lexicap/0014-large.html#00:04:23.760
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
reckless let's just go with it and see? We weren't very good at
https://karpathy.ai/lexicap/0014-large.html#00:04:26.640
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
BattleBots. We lost all of our matches the first round.
https://karpathy.ai/lexicap/0014-large.html#00:04:30.720
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
The one I built first, both of them were these wedge shaped robots because
https://karpathy.ai/lexicap/0014-large.html#00:04:34.240
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
wedge, even though it's sort of boring to look
https://karpathy.ai/lexicap/0014-large.html#00:04:38.240
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
at, is extremely effective. You drive towards another robot and
https://karpathy.ai/lexicap/0014-large.html#00:04:40.000
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
the front edge of it gets under them and then they sort of flip over,
https://karpathy.ai/lexicap/0014-large.html#00:04:43.200
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
kind of like a door stopper. And the first one had a
https://karpathy.ai/lexicap/0014-large.html#00:04:46.640
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
pneumatic polished stainless steel spike on the front that would shoot out about
https://karpathy.ai/lexicap/0014-large.html#00:04:49.840
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
eight inches. The purpose of which is what? Pretty,
https://karpathy.ai/lexicap/0014-large.html#00:04:53.360
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
pretty ineffective actually, but it looks cool.
https://karpathy.ai/lexicap/0014-large.html#00:04:56.400
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
And was it to help with the lift? No, it was, it was just to try to poke holes
https://karpathy.ai/lexicap/0014-large.html#00:04:58.720
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
in the other robot. And then the second time I did it, which is the following,
https://karpathy.ai/lexicap/0014-large.html#00:05:02.960
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
I think maybe 18 months later, we had a, well a titanium axe with a, with a
https://karpathy.ai/lexicap/0014-large.html#00:05:07.040
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
hardened steel tip on it that was powered by a hydraulic
https://karpathy.ai/lexicap/0014-large.html#00:05:12.640
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
cylinder which we were activating with liquid CO2, which was,
https://karpathy.ai/lexicap/0014-large.html#00:05:16.640
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
had its own set of problems. So great, so that's kind of on the hardware side.
https://karpathy.ai/lexicap/0014-large.html#00:05:21.920
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
I mean at a certain point there must have been born a fascination
https://karpathy.ai/lexicap/0014-large.html#00:05:26.160
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
on the software side. So what was the first piece of code you've written?
https://karpathy.ai/lexicap/0014-large.html#00:05:31.200
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
Go back there, see what language was it?
https://karpathy.ai/lexicap/0014-large.html#00:05:35.360
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
What, what was that? Was it Emacs? Vim? Was it a more
https://karpathy.ai/lexicap/0014-large.html#00:05:38.480
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
respectable modern IDE? Do you, do you remember any of this?
https://karpathy.ai/lexicap/0014-large.html#00:05:41.760
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
Yeah, well I remember, I think maybe when I was in
https://karpathy.ai/lexicap/0014-large.html#00:05:45.760
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
third or fourth grade, the school I was at, elementary school, had a bunch of
https://karpathy.ai/lexicap/0014-large.html#00:05:49.680
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
Apple II computers and we'd play games on those. And I
https://karpathy.ai/lexicap/0014-large.html#00:05:52.880
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
remember every once in a while something would,
https://karpathy.ai/lexicap/0014-large.html#00:05:56.800
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
would, would crash or wouldn't start up correctly and it would dump you out to
https://karpathy.ai/lexicap/0014-large.html#00:05:58.640
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
what I later learned was like sort of a command prompt.
https://karpathy.ai/lexicap/0014-large.html#00:06:02.560
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
And my teacher would come over and type, I actually remember this to this day for
https://karpathy.ai/lexicap/0014-large.html#00:06:05.680
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
some reason, like PR number six or PR pound six, which is
https://karpathy.ai/lexicap/0014-large.html#00:06:08.640
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
peripheral six, which is the disk drive, which would fire up the disk and load the
https://karpathy.ai/lexicap/0014-large.html#00:06:12.400
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
program. And I just remember thinking wow, she's
https://karpathy.ai/lexicap/0014-large.html#00:06:15.120
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
like a hacker, like teach me these, these codes, these error codes, that is what
https://karpathy.ai/lexicap/0014-large.html#00:06:17.200
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
I called them at the time. But she had no interest in that, so it
https://karpathy.ai/lexicap/0014-large.html#00:06:21.040
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
wasn't until I think about fifth grade that I had a school where you could
https://karpathy.ai/lexicap/0014-large.html#00:06:23.840
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
actually go on these Apple IIs and learn to program. And so it was all in basic,
https://karpathy.ai/lexicap/0014-large.html#00:06:28.400
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
you know, where every line, you know, the line numbers are all number, that every
https://karpathy.ai/lexicap/0014-large.html#00:06:31.520
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
line is numbered and you have to like leave enough space
https://karpathy.ai/lexicap/0014-large.html#00:06:34.720
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
between the numbers so that if you want to tweak your code you go back and
https://karpathy.ai/lexicap/0014-large.html#00:06:37.920
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
the first line was 10 and the second line is 20. Now you have to go back and
https://karpathy.ai/lexicap/0014-large.html#00:06:41.600
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
insert 15 and if you need to add code in front of
https://karpathy.ai/lexicap/0014-large.html#00:06:44.720
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
that, you know, 11 or 12 and you hope you don't run out of line numbers and have
https://karpathy.ai/lexicap/0014-large.html#00:06:47.360
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
to redo the whole thing. And there's go to statements? Yeah, go to
https://karpathy.ai/lexicap/0014-large.html#00:06:50.240
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
and it's very basic, maybe hence the name, but a lot of fun.
https://karpathy.ai/lexicap/0014-large.html#00:06:54.160
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
And that was like, that was, you know, that's when, you know,
https://karpathy.ai/lexicap/0014-large.html#00:06:58.080
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
when you first program you see the magic of it.
https://karpathy.ai/lexicap/0014-large.html#00:07:01.760
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
It's like, it just, just like this world opens up with, you know, endless
https://karpathy.ai/lexicap/0014-large.html#00:07:03.520
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
possibilities for the things you could build or
https://karpathy.ai/lexicap/0014-large.html#00:07:07.040
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
or accomplish with that computer. So you got the bug then, so
https://karpathy.ai/lexicap/0014-large.html#00:07:09.040
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
even starting with basic and then what C++ throughout,
https://karpathy.ai/lexicap/0014-large.html#00:07:12.160
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
what did you, was there computer programming, computer science classes in
https://karpathy.ai/lexicap/0014-large.html#00:07:15.920
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
high school? Not, not where I went, so it was self
https://karpathy.ai/lexicap/0014-large.html#00:07:19.200
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
taught, but I did a lot of programming. The thing that, you know, sort of
https://karpathy.ai/lexicap/0014-large.html#00:07:22.320
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
pushed me in the path of eventually working on self driving cars is actually
https://karpathy.ai/lexicap/0014-large.html#00:07:27.760
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
one of these really long trips driving from my
https://karpathy.ai/lexicap/0014-large.html#00:07:31.280
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
house in Kansas to, to I think Las Vegas where we did the BattleBots competition
https://karpathy.ai/lexicap/0014-large.html#00:07:34.240
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
and I had just gotten my, I think my learner's permit or
https://karpathy.ai/lexicap/0014-large.html#00:07:39.440
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
early driver's permit and so I was driving this,
https://karpathy.ai/lexicap/0014-large.html#00:07:43.680
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
you know, 10 hour stretch across western Kansas where it's just,
https://karpathy.ai/lexicap/0014-large.html#00:07:47.040
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
you're going straight on a highway and it is mind numbingly boring. And I
https://karpathy.ai/lexicap/0014-large.html#00:07:50.480
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
remember thinking even then with my sort of mediocre programming
https://karpathy.ai/lexicap/0014-large.html#00:07:53.760
Kyle Vogt: Cruise Automation | Lex Fridman Podcast #14
background that this is something that a computer can do, right? Let's take a
https://karpathy.ai/lexicap/0014-large.html#00:07:57.440