episode
stringlengths
45
100
text
stringlengths
1
528
timestamp_link
stringlengths
56
56
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
purpose AI is not going to do anything conceivable in order to attain a goal. All engineering systems
https://karpathy.ai/lexicap/0003-large.html#00:17:50.240
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
are designed to trade off across multiple goals. When we build cars in the first place,
https://karpathy.ai/lexicap/0003-large.html#00:17:57.760
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
we didn't forget to install brakes because the goal of a car is to go fast. It occurred to people,
https://karpathy.ai/lexicap/0003-large.html#00:18:02.800
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
yes, you want it to go fast, but not always. So you would build in brakes too. Likewise,
https://karpathy.ai/lexicap/0003-large.html#00:18:08.880
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
if a car is going to be autonomous and program it to take the shortest route to the airport,
https://karpathy.ai/lexicap/0003-large.html#00:18:13.920
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
it's not going to take the diagonal and mow down people and trees and fences because that's the
https://karpathy.ai/lexicap/0003-large.html#00:18:20.320
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
shortest route. That's not what we mean by the shortest route when we program it. And that's just
https://karpathy.ai/lexicap/0003-large.html#00:18:24.560
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
what an intelligence system is by definition. It takes into account multiple constraints.
https://karpathy.ai/lexicap/0003-large.html#00:18:29.120
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
The same is true, in fact, even more true of so called general intelligence. That is,
https://karpathy.ai/lexicap/0003-large.html#00:18:36.000
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
if it's genuinely intelligent, it's not going to pursue some goal singlemindedly, omitting every
https://karpathy.ai/lexicap/0003-large.html#00:18:41.280
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
other consideration and collateral effect. That's not artificial and general intelligence. That's
https://karpathy.ai/lexicap/0003-large.html#00:18:48.080
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
artificial stupidity. I agree with you, by the way, on the promise of autonomous vehicles for
https://karpathy.ai/lexicap/0003-large.html#00:18:54.400
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
improving human welfare. I think it's spectacular. And I'm surprised at how little press coverage
https://karpathy.ai/lexicap/0003-large.html#00:19:01.120
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
notes that in the United States alone, something like 40,000 people die every year on the highways,
https://karpathy.ai/lexicap/0003-large.html#00:19:06.160
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
vastly more than are killed by terrorists. And we spent a trillion dollars on a war to combat
https://karpathy.ai/lexicap/0003-large.html#00:19:11.680
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
deaths by terrorism, about half a dozen a year. Whereas year in, year out, 40,000 people are
https://karpathy.ai/lexicap/0003-large.html#00:19:18.240
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
massacred on the highways, which could be brought down to very close to zero. So I'm with you on
https://karpathy.ai/lexicap/0003-large.html#00:19:24.160
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
the humanitarian benefit. Let me just mention that as a person who's building these cars,
https://karpathy.ai/lexicap/0003-large.html#00:19:29.920
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
it is a little bit offensive to me to say that engineers would be clueless enough not to engineer
https://karpathy.ai/lexicap/0003-large.html#00:19:34.800
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
safety into systems. I often stay up at night thinking about those 40,000 people that are dying.
https://karpathy.ai/lexicap/0003-large.html#00:19:39.360
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
And everything I tried to engineer is to save those people's lives. So every new invention that
https://karpathy.ai/lexicap/0003-large.html#00:19:45.280
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
I'm super excited about, in all the deep learning literature and CVPR conferences and NIPS, everything
https://karpathy.ai/lexicap/0003-large.html#00:19:50.800
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
I'm super excited about is all grounded in making it safe and help people. So I just don't see how
https://karpathy.ai/lexicap/0003-large.html#00:19:59.280
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
that trajectory can all of a sudden slip into a situation where intelligence will be highly
https://karpathy.ai/lexicap/0003-large.html#00:20:08.240
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
negative. You and I certainly agree on that. And I think that's only the beginning of the
https://karpathy.ai/lexicap/0003-large.html#00:20:13.280
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
potential humanitarian benefits of artificial intelligence. There's been enormous attention to
https://karpathy.ai/lexicap/0003-large.html#00:20:17.920
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
what are we going to do with the people whose jobs are made obsolete by artificial intelligence,
https://karpathy.ai/lexicap/0003-large.html#00:20:24.320
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
but very little attention given to the fact that the jobs that are going to be made obsolete are
https://karpathy.ai/lexicap/0003-large.html#00:20:28.720
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
horrible jobs. The fact that people aren't going to be picking crops and making beds and driving
https://karpathy.ai/lexicap/0003-large.html#00:20:32.560
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
trucks and mining coal, these are soul deadening jobs. And we have a whole literature sympathizing
https://karpathy.ai/lexicap/0003-large.html#00:20:38.960
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
with the people stuck in these menial, mind deadening, dangerous jobs. If we can eliminate
https://karpathy.ai/lexicap/0003-large.html#00:20:45.520
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
them, this is a fantastic boon to humanity. Now granted, you solve one problem and there's another
https://karpathy.ai/lexicap/0003-large.html#00:20:53.040
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
one, namely, how do we get these people a decent income? But if we're smart enough to invent machines
https://karpathy.ai/lexicap/0003-large.html#00:20:58.480
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
that can make beds and put away dishes and handle hospital patients, I think we're smart enough to
https://karpathy.ai/lexicap/0003-large.html#00:21:05.840
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
figure out how to redistribute income to apportion some of the vast economic savings to the human
https://karpathy.ai/lexicap/0003-large.html#00:21:12.400
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
beings who will no longer be needed to make beds. Okay. Sam Harris says that it's obvious that
https://karpathy.ai/lexicap/0003-large.html#00:21:19.200
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
eventually AI will be an existential risk. He's one of the people who says it's obvious.
https://karpathy.ai/lexicap/0003-large.html#00:21:26.000
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
We don't know when the claim goes, but eventually it's obvious. And because we don't know when,
https://karpathy.ai/lexicap/0003-large.html#00:21:31.760
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
we should worry about it now. This is a very interesting argument in my eyes. So how do we
https://karpathy.ai/lexicap/0003-large.html#00:21:38.640
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
think about timescale? How do we think about existential threats when we don't really, we know
https://karpathy.ai/lexicap/0003-large.html#00:21:45.680
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
so little about the threat, unlike nuclear weapons perhaps, about this particular threat, that it
https://karpathy.ai/lexicap/0003-large.html#00:21:51.120
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
could happen tomorrow, right? So, but very likely it won't. Very likely it'd be a hundred years away.
https://karpathy.ai/lexicap/0003-large.html#00:21:58.160
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
So how do we ignore it? How do we talk about it? Do we worry about it? How do we think about those?
https://karpathy.ai/lexicap/0003-large.html#00:22:04.560
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
What is it?
https://karpathy.ai/lexicap/0003-large.html#00:22:12.480
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
A threat that we can imagine. It's within the limits of our imagination,
https://karpathy.ai/lexicap/0003-large.html#00:22:13.840
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
but not within our limits of understanding to accurately predict it.
https://karpathy.ai/lexicap/0003-large.html#00:22:18.560
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
But what is the it that we're afraid of?
https://karpathy.ai/lexicap/0003-large.html#00:22:24.320
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
Sorry. AI being the existential threat.
https://karpathy.ai/lexicap/0003-large.html#00:22:26.880
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
AI. How? Like enslaving us or turning us into paperclips?
https://karpathy.ai/lexicap/0003-large.html#00:22:30.320
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
I think the most compelling from the Sam Harris perspective would be the paperclip situation.
https://karpathy.ai/lexicap/0003-large.html#00:22:35.120
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
Yeah. I mean, I just think it's totally fanciful. I mean, that is don't build a system.
https://karpathy.ai/lexicap/0003-large.html#00:22:39.520
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
Don't give a, don't, first of all, the code of engineering is you don't implement a system with
https://karpathy.ai/lexicap/0003-large.html#00:22:43.440
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
massive control before testing it. Now, perhaps the culture of engineering will radically change.
https://karpathy.ai/lexicap/0003-large.html#00:22:50.080
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
Then I would worry, but I don't see any signs that engineers will suddenly do idiotic things,
https://karpathy.ai/lexicap/0003-large.html#00:22:55.040
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
like put a electric power plant in control of a system that they haven't tested first.
https://karpathy.ai/lexicap/0003-large.html#00:23:00.320
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
Or all of these scenarios, not only imagine almost a magically powered intelligence,
https://karpathy.ai/lexicap/0003-large.html#00:23:07.360
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
including things like cure cancer, which is probably an incoherent goal because there's
https://karpathy.ai/lexicap/0003-large.html#00:23:15.600
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
so many different kinds of cancer or bring about world peace. I mean, how do you even specify that
https://karpathy.ai/lexicap/0003-large.html#00:23:20.240
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
as a goal? But the scenarios also imagine some degree of control of every molecule in the
https://karpathy.ai/lexicap/0003-large.html#00:23:25.600
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
universe, which not only is itself unlikely, but we would not start to connect these systems to
https://karpathy.ai/lexicap/0003-large.html#00:23:31.360
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
infrastructure without testing as we would any kind of engineering system.
https://karpathy.ai/lexicap/0003-large.html#00:23:39.120
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
Now, maybe some engineers will be irresponsible and we need legal and regulatory and legal
https://karpathy.ai/lexicap/0003-large.html#00:23:45.680
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
responsibility implemented so that engineers don't do things that are stupid by their own standards.
https://karpathy.ai/lexicap/0003-large.html#00:23:53.920
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
But the, I've never seen enough of a plausible scenario of existential threat to devote large
https://karpathy.ai/lexicap/0003-large.html#00:24:00.640
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
amounts of brain power to, to forestall it. So you believe in the sort of the power on
https://karpathy.ai/lexicap/0003-large.html#00:24:08.560
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
mass of the engineering of reason, as you argue in your latest book of Reason and Science, to sort of
https://karpathy.ai/lexicap/0003-large.html#00:24:14.240
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
be the very thing that guides the development of new technology so it's safe and also keeps us safe.
https://karpathy.ai/lexicap/0003-large.html#00:24:20.400
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
You know, granted the same culture of safety that currently is part of the engineering mindset for
https://karpathy.ai/lexicap/0003-large.html#00:24:28.000
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
airplanes, for example. So yeah, I don't think that that should be thrown out the window and
https://karpathy.ai/lexicap/0003-large.html#00:24:34.560
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
that untested all powerful systems should be suddenly implemented, but there's no reason to
https://karpathy.ai/lexicap/0003-large.html#00:24:40.480
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
think they are. And in fact, if you look at the progress of artificial intelligence, it's been,
https://karpathy.ai/lexicap/0003-large.html#00:24:45.520
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
you know, it's been impressive, especially in the last 10 years or so, but the idea that suddenly
https://karpathy.ai/lexicap/0003-large.html#00:24:50.400
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
there'll be a step function that all of a sudden before we know it, it will be all powerful,
https://karpathy.ai/lexicap/0003-large.html#00:24:54.160
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
that there'll be some kind of recursive self improvement, some kind of fume is also fanciful.
https://karpathy.ai/lexicap/0003-large.html#00:25:00.080
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
We, certainly by the technology that we, that we're now impresses us, such as deep learning,
https://karpathy.ai/lexicap/0003-large.html#00:25:06.800
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
where you train something on hundreds of thousands or millions of examples,
https://karpathy.ai/lexicap/0003-large.html#00:25:13.200
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
they're not hundreds of thousands of problems of which curing cancer is a typical example.
https://karpathy.ai/lexicap/0003-large.html#00:25:18.320
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
And so the kind of techniques that have allowed AI to increase in the last five years are not the
https://karpathy.ai/lexicap/0003-large.html#00:25:26.000
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
kind that are going to lead to this fantasy of exponential sudden self improvement. I think it's
https://karpathy.ai/lexicap/0003-large.html#00:25:31.520
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
kind of a magical thinking. It's not based on our understanding of how AI actually works.
https://karpathy.ai/lexicap/0003-large.html#00:25:40.320
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
Now give me a chance here. So you said fanciful, magical thinking. In his TED talk,
https://karpathy.ai/lexicap/0003-large.html#00:25:45.440
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
Sam Harris says that thinking about AI killing all human civilization is somehow fun,
https://karpathy.ai/lexicap/0003-large.html#00:25:51.040
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
intellectually. Now I have to say as a scientist engineer, I don't find it fun,
https://karpathy.ai/lexicap/0003-large.html#00:25:55.760
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
but when I'm having beer with my non AI friends, there is indeed something fun and appealing about
https://karpathy.ai/lexicap/0003-large.html#00:26:00.400
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
it. Like talking about an episode of Black Mirror, considering if a large meteor is headed towards
https://karpathy.ai/lexicap/0003-large.html#00:26:08.560
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
Earth, we were just told a large meteor is headed towards Earth, something like this. And can you
https://karpathy.ai/lexicap/0003-large.html#00:26:14.640
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
relate to this sense of fun? And do you understand the psychology of it?
https://karpathy.ai/lexicap/0003-large.html#00:26:20.640
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
Yes. Good question. I personally don't find it fun. I find it kind of actually a waste of time
https://karpathy.ai/lexicap/0003-large.html#00:26:24.560
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
because there are genuine threats that we ought to be thinking about like pandemics, like cyber
https://karpathy.ai/lexicap/0003-large.html#00:26:32.800
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
security vulnerabilities, like the possibility of nuclear war and certainly climate change.
https://karpathy.ai/lexicap/0003-large.html#00:26:39.760
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
You know, this is enough to fill many conversations. And I think Sam did put his
https://karpathy.ai/lexicap/0003-large.html#00:26:46.160
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
finger on something, namely that there is a community, sometimes called the rationality
https://karpathy.ai/lexicap/0003-large.html#00:26:54.320
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
community, that delights in using its brainpower to come up with scenarios that would not occur
https://karpathy.ai/lexicap/0003-large.html#00:27:00.240
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
to mere mortals, to less cerebral people. So there is a kind of intellectual thrill in finding new
https://karpathy.ai/lexicap/0003-large.html#00:27:07.280
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
things to worry about that no one has worried about yet. I actually think, though, that it's
https://karpathy.ai/lexicap/0003-large.html#00:27:14.560
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
not only is it a kind of fun that doesn't give me particular pleasure, but I think there can be a
https://karpathy.ai/lexicap/0003-large.html#00:27:19.840
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
pernicious side to it, namely that you overcome people with such dread, such fatalism, that there
https://karpathy.ai/lexicap/0003-large.html#00:27:25.440
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
are so many ways to die, to annihilate our civilization, that we may as well enjoy life
https://karpathy.ai/lexicap/0003-large.html#00:27:32.400
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
while we can. There's nothing we can do about it. If climate change doesn't do us in, then runaway
https://karpathy.ai/lexicap/0003-large.html#00:27:39.200
Steven Pinker: AI in the Age of Reason | Lex Fridman Podcast #3
robots will. So let's enjoy ourselves now. We've got to prioritize. We have to look at threats that
https://karpathy.ai/lexicap/0003-large.html#00:27:42.880