Many People are Einstein but in the Patent Clerk Days - François Chollet | AI Podcast Clips
00:00:00.000 | So yeah, so intelligence explosion, I'm sure your family has the idea, but it's the idea
00:00:11.800 | that if you were to build general AI problem solving algorithms, well, the problem of building
00:00:21.200 | such an AI, that itself is a problem that could be solved by your AI.
00:00:26.840 | And maybe it could be solved better than what humans can do.
00:00:29.920 | So your AI could start tweaking its own algorithm, could start being a better version of itself.
00:00:36.080 | And so on iteratively in a recursive fashion.
00:00:39.540 | And so you would end up with an AI with exponentially increasing intelligence.
00:00:47.360 | And I was basically questioning this idea, first of all, because the notion of intelligence
00:00:54.640 | explosion uses an implicit definition of intelligence that doesn't sound quite right to me.
00:01:01.640 | It considers intelligence as a property of a brain that you can consider in isolation,
00:01:10.000 | like the height of a building, for instance.
00:01:12.960 | But that's not really what intelligence is.
00:01:15.840 | Intelligence emerges from the interaction between a brain, a body, like embodied intelligence,
00:01:23.040 | and an environment.
00:01:24.640 | And if you're missing one of these pieces, then you cannot redefine intelligence anymore.
00:01:30.100 | So just tweaking a brain to make it smarter and smarter doesn't actually make any sense
00:01:34.280 | to me.
00:01:35.280 | So, first of all, you're crushing the dreams of many people, right?
00:01:39.280 | So there's a, let's look at like Sam Harris, actually a lot of physicists, Max Tegmark,
00:01:45.160 | people who think, you know, the universe is a information processing system.
00:01:50.920 | Our brain is kind of an information processing system.
00:01:53.920 | So what's the theoretical limit?
00:01:55.720 | Like, it doesn't make sense that there should be some, it seems naive to think that our
00:02:02.840 | own brain is somehow the limit of the capabilities of this information.
00:02:07.720 | I'm playing devil's advocate here.
00:02:09.880 | This information processing system.
00:02:11.920 | And then if you just scale it, if you're able to build something that's on par with the
00:02:16.240 | brain, you just, the process that builds it just continues and it'll improve exponentially.
00:02:22.640 | So that's the logic that's used actually by almost everybody that is worried about super
00:02:31.480 | human intelligence.
00:02:32.480 | - Yeah.
00:02:33.480 | - So you're trying to make, so most people who are skeptical of that are kind of like,
00:02:38.560 | this doesn't, their thought process, this doesn't feel right.
00:02:42.160 | Like that's for me as well.
00:02:43.880 | So I'm more like, it doesn't, the whole thing is shrouded in mystery where you can't really
00:02:50.520 | say anything concrete, but you could say this doesn't feel right.
00:02:54.200 | This doesn't feel like that's how the brain works.
00:02:56.880 | And you're trying to, with your blog posts and now making it a little more explicit.
00:03:01.880 | So one idea is that the brain doesn't exist alone.
00:03:07.520 | It exists within the environment.
00:03:10.120 | So you can't exponentially, you would have to somehow exponentially improve the environment
00:03:15.480 | and the brain together almost yet in order to create something that's much smarter in
00:03:22.800 | some kind of, of course we don't have a definition of intelligence.
00:03:26.840 | - That's correct.
00:03:27.840 | That's correct.
00:03:28.840 | I don't think, if you look at very smart people today, even humans, not even talking about
00:03:32.760 | AIs, I don't think their brain and the performance of their brain is the bottleneck to the expressed
00:03:40.080 | intelligence, to their achievements.
00:03:43.320 | You cannot just tweak one part of the system, like of this brain body environment system
00:03:49.440 | and expect capabilities like what emerges out of this system to just, you know, explode
00:03:55.760 | exponentially.
00:03:57.240 | Because anytime you improve one part of a system with many interdependencies like this,
00:04:03.520 | there's a new bottleneck that arises, right?
00:04:05.680 | And I don't think even today for very smart people, their brain is not the bottleneck
00:04:10.920 | to the sort of problems they can solve.
00:04:12.960 | Right?
00:04:13.960 | In fact, many very smart people today, you know, they're not actually solving any big
00:04:18.880 | scientific problems.
00:04:19.880 | They're not Einstein.
00:04:20.880 | They're like Einstein, but you know, the patent clerk days, like Einstein became Einstein
00:04:28.000 | because this was a meeting of a genius with a big problem at the right time.
00:04:34.840 | Right?
00:04:35.840 | But maybe this meeting could have never happened.
00:04:38.680 | And then Einstein would have just been a patent clerk.
00:04:40.640 | Right?
00:04:41.640 | And in fact, many people today are probably like genius level smart, but you wouldn't
00:04:47.840 | know because they're not really expressing any of that.
00:04:50.280 | Wow.
00:04:51.280 | That's brilliant.
00:04:52.280 | We can think of the world, earth, but also the universe as just as a space of problems.
00:04:58.880 | So all of these problems and tasks are roaming it of various difficulty.
00:05:03.420 | And there's agents, creatures like ourselves and animals and so on that are also roaming
00:05:08.600 | it.
00:05:09.600 | And then you get coupled with a problem and then you solve it.
00:05:13.920 | But without that coupling, you can't demonstrate your quote unquote intelligence.
00:05:18.760 | Exactly.
00:05:19.760 | So intelligence is the meeting of great problem solving capabilities with a great problem.
00:05:25.000 | And if you don't have the problem, you don't really express an intelligence.
00:05:28.440 | All you're left with is potential intelligence, like the performance of your brain or, you
00:05:32.680 | know, how high your IQ is, which in itself is just a number.
00:05:37.320 | Right.
00:05:38.320 | Yeah.
00:05:38.320 | Thank you.
00:05:43.320 | Thank you.
00:05:48.320 | Thank you.
00:05:53.320 | [BLANK_AUDIO]