{"id":23499,"date":"2026-08-20T14:04:40","date_gmt":"2026-08-20T14:04:40","guid":{"rendered":"https:\/\/scannn.com\/are-we-thinking-correctly-about-ai-intelligence\/"},"modified":"2026-08-20T14:04:40","modified_gmt":"2026-08-20T14:04:40","slug":"are-we-thinking-correctly-about-ai-intelligence","status":"publish","type":"post","link":"https:\/\/scannn.com\/lv\/are-we-thinking-correctly-about-ai-intelligence\/","title":{"rendered":"Are We Thinking Correctly About AI Intelligence?"},"content":{"rendered":"\n<div>\n<p>When an LLM answers a question, is it reasoning like humans, or just producing text that looks like reasoning? The distinction isn\u2019t just philosophical, this determines what we can trust AI to do, how closely we need to supervise it, and ultimately what its real-world impact will turn out to be.<\/p>\n<p><a href=\"https:\/\/www.santafe.edu\/people\/profile\/melanie-mitchell\">Melanie Mitchell<\/a> at the Santa Fe Institute argues that we lack adequate methods for measuring machine cognition, and that AI is a form of \u201calien intelligence\u201d that operates through non-human cognitive mechanisms. In this episode of <em>The Joy of Why<\/em>, Mitchell tells Steven Strogatz how methods that psychologists use to study cognition in other kinds of \u201calien intelligence\u201d \u2014 babies and animals \u2014 can be adapted to probe AI, and she lays out six principles for better assessing machine cognition. Their conversation ranges from the challenge of interpreting what\u2019s happening inside these systems, to recent AI-assisted breakthroughs in mathematics, to why a math-performing horse from the early 1900s offers a cautionary tale for how we assess intelligence.<\/p>\n<p>Listen on <a href=\"https:\/\/podcasts.apple.com\/us\/podcast\/the-joy-of-why\/id1608948873\">Apple Podcasts<\/a>, <a href=\"https:\/\/open.spotify.com\/show\/2FoxHraQSKwxV2HgUfwLMp\">Spotify<\/a>, <a href=\"https:\/\/tunein.com\/podcasts\/Science-Podcasts\/The-Joy-of-Why-p1653040\/\">TuneIn<\/a> or your favorite podcasting app, or you can <a href=\"https:\/\/www.quantamagazine.org\/tag\/the-joy-of-why\">stream it from Quanta<\/a>.<\/p>\n<aside class=\"post__aside mb2 relative post__aside--inline hide-on-print\">\n<div class=\"podcast scale0 \">\n            <a href=\"https:\/\/www.quantamagazine.org\/tag\/the-joy-of-why\" class=\"h6 pangram uppercase kern mt0 xbold mbr1 decoration--none no--shadow block align-r\"><br \/>\n            <span class=\"theme__text hover:text-gray-767\">All episodes <svg class=\"icon-s theme__accent\" x=\"0px\" y=\"0px\" viewbox=\"0 0 50 50\" enable-background=\"new 0 0 50 50\" xml:space=\"preserve\"><path fill=\"currentColor\" d=\"M14.7,48.4l-2.2-2.2c-1.1-1.1-1.1-2.8,0-3.9L29.7,25L12.5,7.7c-1.1-1.1-1.1-2.8,0-3.9l2.2-2.2 c1.1-1.1,2.8-1.1,3.9,0L40,23c1.1,1.1,1.1,2.8,0,3.9L18.6,48.4C17.5,49.5,15.7,49.5,14.7,48.4z\"\/><\/svg><\/span><br \/>\n        <\/a><br \/>\n        <audio preload=\"none\" data-role=\"podcast\" data-name=\"podcast-165961\"><source src=\"https:\/\/dts.podtrac.com\/redirect.mp3\/tracking.swap.fm\/track\/sblTq32fyWAjsHzze2LG\/dovetail.prxu.org\/5340\/2f24e110-c7d4-4e5b-8365-21ff422da188\/JOW_AI_w__Melanie_Mitchell_FINAL.mp3\" type=\"audio\/mpeg\">Your browser does not support the audio element.<br \/>\n    <\/source><\/audio><\/p>\n<\/div>\n<\/aside>\n<h2><strong>Transcript<\/strong><\/h2>\n<p>[<em>Music plays<\/em>]<\/p>\n<p><strong>STEVE STROGATZ:<\/strong> I\u2019m Steve Strogatz.<\/p>\n<p><strong>JANNA LEVIN:<\/strong> And I\u2019m Janna Levin.<\/p>\n<p><strong>STROGATZ:<\/strong> And this is <em>The Joy of Why<\/em>.<\/p>\n<p><strong>LEVIN:<\/strong> A podcast from <em>Quanta Magazine<\/em> where we explore some of the biggest unanswered questions in math and science today.<\/p>\n<p><strong>STROGATZ:<\/strong> Well, hello, hello. This is unsurprisingly yet another show about AI.<\/p>\n<p><strong>LEVIN:<\/strong> I\u2019m telling you, it\u2019s a topic people can\u2019t seem to get enough about, and I\u2019m becoming reluctant to pontificate anymore. It\u2019s changing too quickly.<\/p>\n<p><strong>STROGATZ:<\/strong> It\u2019s true. It is moving very fast. Anything we say could be obsolete by next week.<\/p>\n<p><strong>LEVIN: <\/strong>Oh yeah.<\/p>\n<p><strong>STROGATZ:<\/strong> As we speak, it\u2019s July 23rd, 2026.<\/p>\n<p><strong>LEVIN:<\/strong> And it feels different to me than it did in July 23rd, 2025, that\u2019s for sure.<\/p>\n<p><strong>STROGATZ:<\/strong> Mmm. That\u2019s actually relevant, this talking about timelines, because our guest today, Melanie Mitchell, who is a cognitive scientist and computer scientist at Santa Fe Institute, is someone that we had on the show previously. She and I spoke about five years ago, and that is before ChatGPT.<\/p>\n<p><strong>LEVIN:<\/strong> Right. And was she interested in AI then?<\/p>\n<p><strong>STROGATZ:<\/strong> Oh, yes.<\/p>\n<p><strong>LEVIN:<\/strong> Okay, so it wasn\u2019t just cognitive science.<\/p>\n<p><strong>STROGATZ:<\/strong> Absolutely. I, I mean, yes, I should say Melanie has been thinking about AI for a long time, and she\u2019ll tell us about that. But the thing that\u2019s gonna be so interesting, I feel, for us to discuss today is, um, Melanie\u2019s point of view, which is to think about the problem of AI from the standpoint of fields like developmental psychology. Like, how does a baby or a young child get to be as intelligent as they soon become?<\/p>\n<p><strong>LEVIN:<\/strong> Oh, I think that\u2019s so interesting \u2019cause we\u2019re so excited about the artificial mind when we have very little comprehension of the human mind.<\/p>\n<p><strong>STROGATZ:<\/strong> Exactly.<\/p>\n<p><strong>LEVIN: <\/strong>Right, so we\u2019re trying to skip a step.<\/p>\n<p><strong>STROGATZ:<\/strong> Well, that\u2019s right. And not just human mind, but also animal minds, right? So there\u2019s the field of comparative psychology where we look at intelligence in birds or dogs or dolphins, whatever. Um, we have a lot to learn about thinking about intelligences other than our own adult human intelligence.<\/p>\n<p><strong>LEVIN:<\/strong> Yeah, and this idea that we\u2019re going to somehow simply understand a mechanism to generate an artificial intelligence when we, again, don\u2019t understand the mechanism that brings a baby to have its level of intelligence when it\u2019s born or when it\u2019s developing. I mean, I think that\u2019s really interesting to combine those two. So I\u2019m looking forward to this one.<\/p>\n<aside class=\"post__aside mb0 relative post__aside--right\">\n<figure class=\"mb1 mt1 image--shortcode s:mb-0 s:mt-7-5 \">\n<div class=\"relative image mx0\">\n            <\/div><figcaption class=\"image__meta mt1\">\n<\/figcaption><\/figure>\n<\/aside>\n<p><strong>STROGATZ:<\/strong> Well, great. So then let\u2019s dive in with Melanie Mitchell. Here she is.<\/p>\n<p>[<em>Music plays<\/em>]<\/p>\n<p><strong>STROGATZ:<\/strong> Hi there, Melanie.<\/p>\n<p><strong>MELANIE MITCHELL:<\/strong> Hey, Steve.<\/p>\n<p><strong>STROGATZ:<\/strong> Very excited to see you again. This is gonna be fun. We talked a few years ago back when this show was called <em>The Joy of X<\/em>, and I think you may be our first return champion.<\/p>\n<p><strong>MITCHELL:<\/strong> Oh boy, I\u2019m honored.<\/p>\n<p><strong>STROGATZ:<\/strong> Well, you should be. And, I have you back because so much feels like it\u2019s changed in artificial intelligence. We talked, I think it was maybe 2021, and ChatGPT tidal wave hit the world at something like November of 2022. Is that right?<\/p>\n<p><strong>MITCHELL:<\/strong> That\u2019s right.<\/p>\n<p><strong>STROGATZ:<\/strong> So everybody knows that AI is everywhere. We seem to be talking about it. People are worrying about it. Some people are excited about it. It\u2019s certainly very widely used. I suppose I\u2019d like to start by asking, what has surprised you the most about the past few years?<\/p>\n<p><strong>MITCHELL:<\/strong> Oh, wow. So much has surprised me. Just the thought that we could get to where we are now just by training these models on huge amounts of human-generated language and images and so on. I never would\u2019ve dreamed it. So I\u2019ve just been really surprised by what\u2019s happened in AI. Also just the kind of polarized reaction that appeared in the AI community and society at large, I think, has been a little surprising to me, too.<\/p>\n<p><strong>STROGATZ:<\/strong> Polarized in terms of, like, sometimes people will distinguish AI doomers and AI optimists. Is that the kind of thing you\u2019re talking about?<\/p>\n<p><strong>MITCHELL:<\/strong> There\u2019s that dimension, then there\u2019s the dimension of people who believe that AI is smarter than humans and people who think that it\u2019s far, far from being anywhere near human-like intelligence. I guess related to that is sort of the love-it and hate-it. And these are separate dimensions, but maybe they\u2019re correlated.<\/p>\n<p><strong>STROGATZ:<\/strong> Well, and right, and the love-it and hate-it can be also tied to things like the impact on the environment versus, you know, the economic prosperity for certain companies, but then again, what about job loss? There\u2019s so many dimensions to this.<\/p>\n<p><strong>MITCHELL:<\/strong> Oh, there\u2019s so many, yeah.<\/p>\n<p><strong>STROGATZ:<\/strong> But the thing that I really wanna focus on with you today is complex systems, cognitive science, artificial intelligence. You have a lot of different hats but I\u2019m really very curious about the work that you\u2019ve been doing to look at AI through the lens of either developmental psychology, like the way that we try to think about the alien intelligence of human babies, or comparative psychology with the alien intelligence of our pet dogs or smart birds or dolphins or that kind of thing. I mean, it\u2019s a really interesting take on this alien intelligence of AI.<\/p>\n<p><strong>MITCHELL:<\/strong> Yeah. Many people have described AI as an alien kind of intelligence \u2019cause it\u2019s very different from humans, even though it\u2019s been trained on human language and books and everything on the internet and so on. But the way that these systems work, the way that they learn, the way that they reason, the way they do what they do is just really different from the way humans do it.<\/p>\n<p>And this theme was actually picked up by people in developmental psychology, especially, Mike Frank at Stanford, who wrote this paper about how AI people should take some inspiration from the study of babies and young children, developmental psych. And then other people have extended that to, what about animal intelligence? And I guess one of the things that people in cog sci have been urging is that people in AI actually adopt some experimental methodologies that would make AI more like a science.<\/p>\n<p><strong>STROGATZ:<\/strong> Yeah, I really like this point of view, and I think it may not be so familiar to our listeners. I have to admit it wasn\u2019t that familiar to me. You know, I never studied cognitive science, or never took a course in developmental psychology, and people in those fields have been thinking about these issues for\u2026 Well, I don\u2019t know. You tell me.<\/p>\n<p><strong>MITCHELL:<\/strong> Yeah, at least 100 years.<\/p>\n<aside class=\"post__aside mb0 relative mha post__aside--right hide-on-print\">\n<div class=\"post__aside__pullquote relative\">\n<div class=\"pullquote theme__text mb2 align-c\">\n<div class=\"mb1\">\n<p style=\"margin: 0in 0in .25in 0in;\"><span style=\"color: black;\">The thought that we could get to where we are now just by training these models on huge amounts of human-generated language and images\u2026 I never would\u2019ve dreamed it.<\/span><\/p>\n<\/p><\/div>\n<\/p><\/div>\n<\/p><\/div>\n<\/aside>\n<p><strong>STROGATZ:<\/strong> Yeah, 100 years now. Wow. And I was thinking on the way over we constantly talk about AI as a black box. That we can\u2019t read the weights on the neurons very easily, or even if we can, we don\u2019t know what they tell us. But for that matter, couldn\u2019t you say that our own intelligence is in a lot of ways a black box?<\/p>\n<p><strong>MITCHELL:<\/strong> Absolutely. I mean, we have different ways to penetrate the black box. One is neuroscience, where we actually stick probes into neurons, or we use fMRI or other imaging techniques. There\u2019s also psychology, where you actually look at just the behavior of a person or an animal, and try and infer from that underlying mechanisms.<\/p>\n<p>And those two traditions have, for a long time, been quite separate. But the field of cognitive science tried to integrate them, and originally, the field of cognitive science also included AI. Somehow that integration didn\u2019t work.<\/p>\n<p><strong>STROGATZ:<\/strong> You mean it didn\u2019t catch on sociologically, or what do you mean?<\/p>\n<p><strong>MITCHELL:<\/strong> You know, originally it was thought we\u2019re going to program them the way that humans work. And there was a very close connection between human psychology and people trying to build human psychology into AI. And then that actually didn\u2019t yield success in AI the way that we\u2019ve seen neural networks and learning from data rather than trying to program it in.<\/p>\n<p><strong>STROGATZ:<\/strong> I see.<\/p>\n<p><strong>MITCHELL:<\/strong> And neural networks itself was originally inspired by neuroscience, but the way that neural networks work today has diverged considerably from that original inspiration. So I think the field of machine learning has gone much more in the direction of statistics, which is quite separate from how cognitive science works.<\/p>\n<p><strong>STROGATZ:<\/strong> So at this point, I guess I\u2019d like to talk a bit about benchmarks, because they do seem to be a big part of the discussion broadly in society these days. There was something that got a lot of people chattering in the world of math. One of the latest frontier models did something that looked like a kind of creativity, solved an old, longstanding math problem one of the problems that Paul Erd\u0151s, the great, Hungarian mathematician, he left lots of problems for people to think about, and one of them that they call the unit distance problem was recently solved in a very clever way by AI, and it involved putting two parts of math together in a way that hadn\u2019t really been tried before. And so I bring that up because the last time we spoke, we were talking about an old AI that was learning to play some Atari game, or something. And you talked about how it was so good at playing, but then if you move the paddle a couple pixels up or something, it had to relearn all over again. It didn\u2019t know how to play the slightest variation on the original game.<\/p>\n<p>So the thing you said at the time that stuck with me: \u201cThe strange thing is that these machines don\u2019t seem to be able to transfer their brilliance to any other domain than the one they\u2019ve been trained on.\u201d So that was five years ago. Now I guess I wonder, what do you think? Is that still true?<\/p>\n<p><strong>MITCHELL:<\/strong> Yeah, I mean, that particular model was not a large language model. It was a specific model to play the Atari game. Whereas now we have large language models that are trained on everything. So in some sense, they don\u2019t have to transfer anything. They\u2019re already trained. But, people in AI or machine learning talk about things that are in distribution and out of distribution, and that means that is this thing that we\u2019re asking the models to do similar to things that it\u2019s seen in its training data, or is wholly different?<\/p>\n<p>And I think it\u2019s hard to know. We don\u2019t know what it\u2019s been trained on. The model that\u2019s solving these problems has certainly been trained on a lot of math because there\u2019s a lot of math out there on the internet. It\u2019s been trained on textbooks. It\u2019s been trained on all of Steve Stogatz\u2019s videos that are on YouTube. And these models are pretty good at taking things from one area and putting them together with another area.<\/p>\n<p>But, you know, I don\u2019t know how to talk about this notion of transfer when something\u2019s been trained on everything, especially in a field like math.<\/p>\n<p><strong>STROGATZ:<\/strong> Huh.<\/p>\n<p><strong>MITCHELL:<\/strong> Where you know, \u201ctrained on everything\u201d I think has some meaning in a way. If you say it\u2019s been trained on everything that has to do with being human, clearly that\u2019s not the case. But if you say it\u2019s been trained on everything having to do with math or with code, I don\u2019t know. Is all of mathematical knowledge out there in some kind of textual or video format?<\/p>\n<aside class=\"post__aside mb0 relative mha post__aside--right hide-on-print\">\n<div class=\"post__aside__pullquote relative\">\n<div class=\"pullquote theme__text mb2 align-c\">\n<div class=\"mb1\">\n<p style=\"margin: 0in 0in .25in 0in;\"><span style=\"color: black;\">The way that these [AI] systems work, the way that they learn, the way that they reason, the way they do what they do is just really different from the way humans do it.<\/span><\/p>\n<\/p><\/div>\n<\/p><\/div>\n<\/p><\/div>\n<\/aside>\n<p><strong>STROGATZ:<\/strong> Well, you\u2019re asking me. I, so the thing that is roiling our community in math lately as we try to make sense of what just happened is we used to think, \u201cOkay, these machines are very good at searching,\u201d or, \u201cThese programs are good at searching big spaces.\u201d They have a tremendous amount of knowledge because, as you say, they\u2019ve ingested the whole internet and the Library of Congress, and anything you can read, they\u2019ve read.<\/p>\n<p>So anything where knowledge and the ability to search and to compute very fast and to not forget, all that, that plays into their strength. But the, but to spot a connection between different branches that hadn\u2019t been noticed before and to exploit that to solve a longstanding problem, if a human being did that, we would consider that an aesthetic high point.<\/p>\n<p>You know, mathematicians love it when an idea from topology gets used to solve a problem in geometry, or when an idea from algebra helps. But then again, maybe it\u2019s sort of easy. If you know everything that\u2019s been done and you can look for a lot of possible connections, maybe you\u2019ll occasionally get lucky. So that\u2019s what it sort of seems like happened here.<\/p>\n<p><strong>MITCHELL:<\/strong> Yeah. No, I think that\u2019s right. I don\u2019t\u2026 You know, who knows how it happened because we can\u2019t really look at the innards of the- these models very well for many reasons. But it is creative to bring two unexpected things together and have something that\u2019s actually working. I consider that creative. But, it sort of reminds me in a way, there was a math discovery program way back in the \u201870s maybe done by this guy, Douglas Lenat. It was called EURISKO, I think. And basically it was trying to find new ideas in math. And it explicitly tried to bring together things and stick them together, and it would generate hundreds and hundreds and hundreds and hundreds of these things.<\/p>\n<p>Most of them were just junk, but occasionally it would come up with something interesting. A human had to go in and look and say, \u201cIs this interesting?\u201d The machine couldn\u2019t figure it out itself. So how much of that is going on here? I don\u2019t know. I think here the difference is that the machine obviously is at a much bigger scale, and I don\u2019t know how many tokens of reasoning trace that it generated in the course of solving this problem, and how many kind of wrong paths it went down, and how it figured out that it was on the right path. I mean, these are things that I think are part of the science of AI that not enough people are kind of pursuing right now.<\/p>\n<p><strong>STROGATZ:<\/strong> Yeah, let\u2019s get into that now because that\u2019s really where I wanted to go with you. It\u2019s a nice phrase, the science of AI. I\u2019d like to encourage people to look at this article of yours, Melanie, about the six principles to assess cognitive capacity of AI. But just, as a teaser, could you enunciate what are those six and say a little about them?<\/p>\n<p><strong>MITCHELL:<\/strong> Sure. So the first one is to be aware of your own anthropomorphic cognitive biases. So we tend to project human likeness onto things that talk to us in fluent English. So people very much think that these models have human-like qualities when maybe they actually don\u2019t.<\/p>\n<p>The second one\u2019s a very common sense one for scientists. Be skeptical of hypotheses and develop control experiments. That\u2019s just like Science 101, although I\u2019m not sure how often it\u2019s really followed through in science. People tend to like their own hypotheses.<\/p>\n<p>The third is to develop novel variations of your stimuli or your benchmark items in order to test robustness and generalization.<\/p>\n<p>Uh, the fourth one is these systems don\u2019t have to be black boxes. You can probe them in many different ways and we need more people who are very curious about why they\u2019re getting the results that they do get.<\/p>\n<p>Fifth principle is to consider performance versus competence, sort of what you can show that you can do versus what you actually can do, and in the paper I give some examples of that.<\/p>\n<p>The sixth is to analyze failure types and to embrace any negative results. We tend to put papers with negative results in a drawer and forget about them, but actually they can be incredibly enlightening.<\/p>\n<p><strong>STROGATZ:<\/strong> We all have very direct experience with number six, don\u2019t we? When we see the hallucinations, it starts to make you wonder what\u2019s really going on with these systems, and it\u2019s true you learn a lot from the errors.<\/p>\n<p><strong>MITCHELL:<\/strong> Yeah, people celebrate their positive results and they try to explain away their negative results, but it\u2019s important to really understand what\u2019s going on by looking at where it fails.<\/p>\n<p><strong>STROGATZ:<\/strong> So one example that you give in your article, this is not about AI, but this is about the kind of lesson from biology or from psychology that subtle things can be happening that you need to have an alert and skeptical mind to notice what might really be going on. So could you just regale us with the old story of Clever Hans?<\/p>\n<p><strong>MITCHELL:<\/strong> So Clever Hans was a horse who lived in the early 1900s in Germany. And Clever Hans was able to answer arithmetic questions. So you\u2019d say like, \u201cWhat\u2019s 14 plus 12?\u201d And he would tap his hoof that many times. Looked like a genius horse. And people including many scientists living back then, were very convinced that this was an animal who could do mathematics, who could count, who could reason about simple problems in the way that humans do.<\/p>\n<p>And people were very excited. But then a psychologist, named Oskar Pfungst, came along and said, \u201cWell, let\u2019s do some controlled experiments here,\u201d this notion of controlled experiments you know in psychology being kind of a new idea, I think. And let\u2019s see what happens if he can\u2019t see the person who\u2019s asking the question.<\/p>\n<aside class=\"post__aside mb0 relative mha post__aside--right hide-on-print\">\n<div class=\"post__aside__pullquote relative\">\n<div class=\"pullquote theme__text mb2 align-c\">\n<div class=\"mb1\">\n<p>How can we trust the results of these experiments and studies that are done that show that AI can do all these different things?<\/p>\n<\/p><\/div>\n<\/p><\/div>\n<\/p><\/div>\n<\/aside>\n<p><strong>STROGATZ:<\/strong> Okay<\/p>\n<p><strong>MITCHELL:<\/strong> And then he fails. And it turns out what he\u2019s doing is he\u2019s reading subtle cues on the face of the person who\u2019s asking the question. It turns out that if the person who\u2019s asking the question doesn\u2019t know the answer already, he also fails.<\/p>\n<p>\u2019Cause what the person is doing is they\u2019re reacting to his hoof taps, and when he gets to the answer, there\u2019s some unconscious signal they\u2019re sending that he\u2019s reading. So he is a genius horse, just not at the things that people thought he was a genius at. Instead, he\u2019s a genius at reading social signals in human faces.<\/p>\n<p><strong>STROGATZ:<\/strong> And so in this parable then, as far as like when we are impressed by something seemingly genius that AI is doing, what is our lesson? That, that we should be doing controlled experiments, or what?<\/p>\n<p><strong>MITCHELL:<\/strong> Right. So, an AI system was shown to be really good at reasoning about diagrams in scientific papers, let\u2019s say, I think this is, actually a real example, and could answer questions about them. But then the control experiment was give the questions without showing the diagrams. Seems crazy, right? How could you answer questions about a diagram without seeing the diagram? And it turned out that the AI could do this task because somehow there was some kind of spurious association between the words in the questions and the correct answer.<\/p>\n<p><strong>STROGATZ:<\/strong> So that seems like a case of poor experimental design on whoever was doing the benchmark attempt in retrospect.<\/p>\n<p><strong>MITCHELL:<\/strong> In retrospect, and in retrospect this happens all the time in psychology and other fields, I\u2019m sure too, poor experimental design. Experimental design is a very hard thing and there\u2019s all kinds of confounding possibilities. So this is why the notion of replication in science became so important. If one group does an experiment and they get a result, we shouldn\u2019t necessarily believe that result. That result might be due to some other aspect of their experimental design that wasn\u2019t intended. That\u2019s why it\u2019s very important for independent groups to replicate studies. This isn\u2019t something that people in AI do very much.<\/p>\n<p><strong>STROGATZ:<\/strong> No, and why not? Is it that the replication is not very glamorous because you\u2019re coming in second like there\u2019s no incentive. That\u2019s true in all parts of science, right?<\/p>\n<p><strong>MITCHELL:<\/strong> Yeah. I think that\u2019s true in all parts of science. But it\u2019s also because I think most of AI research is done by people whose background is in computer science or a related field that\u2019s not focused on experimental methodology. I\u2019m a computer scientist. I never had to take a course in experimental methodology. No such course was ever offered to me in my department. It wasn\u2019t seen as part of what computer science was all about, and I think that\u2019s one of the things that\u2019s lacking in today\u2019s AI discussion. How can we trust the results of these experiments and studies that are done that show that AI can do all these different things?<\/p>\n<p>[<em>Music plays<\/em>]<\/p>\n<p><strong>LEVIN:<\/strong> Fascinating. So it seems to me that there\u2019s this cognitive science version of the interference of the observer that everyone talks about in quantum mechanics, right? The observer themselves is interfering with the experiment or the outcome of the experiment, and that is such an interesting role. Of course, this Clever Hans is very famous, and I agree that that is a very clever horse for being able to read the social cues.<\/p>\n<p>But how interesting if this is also happening with AI, that it\u2019s, it\u2019s not just the role of the experimenter that\u2019s interfering, it\u2019s actually the role of the psychology of the experimenter that\u2019s interfering.<\/p>\n<p><strong>STROGATZ:<\/strong> Yeah. It\u2019s a whole dimension that many of us in the theoretical sciences and math don\u2019t get trained in, as Melanie freely admits. You know, I never took a course in experimental design. You as a physicist, I assume you had to take some experimental physics, but\u2026<\/p>\n<aside class=\"post__aside mb0 relative mha post__aside--right hide-on-print\">\n<div class=\"post__aside__pullquote relative\">\n<div class=\"pullquote theme__text mb2 align-c\">\n<div class=\"mb1\">\n<p style=\"margin: 0in 0in .25in 0in;\"><span style=\"color: black;\">We had a talk here at Santa Fe Institute from a philosopher who broke down understanding into 25 different types.<\/span><\/p>\n<\/p><\/div>\n<\/p><\/div>\n<\/p><\/div>\n<\/aside>\n<p><strong>LEVIN:<\/strong> Yeah. It doesn\u2019t really weigh in my actual work. It\u2019s really not experimental. Yeah. So I would not be a very good architect of a good experiment.<\/p>\n<p><strong>STROGATZ:<\/strong> Well, and it seems like it is, something that\u2019s a very live issue because these days the AI companies frequently use benchmarks to show how \u2013 well, to assess how \u2013 how far along are their systems on this quest for either artificial general intelligence or superhuman intelligence, that sort of thing. Or even just to out-compete the other AI companies. We would like to know what the capacities are of these new machine learning systems and other AIs.<\/p>\n<p><strong>LEVIN:<\/strong> Well, I think it might be that it\u2019s just, I don\u2019t think we really know how to evaluate human intelligence, or to really know what somebody\u2019s doing when they\u2019re thinking. I don\u2019t think we know about ourselves. I don\u2019t think we can self-report very well. I can\u2019t say to you, \u201cOh, this is how it\u2019s working in here right now as I\u2019m constructing this sentence. I listened to it, and this was the process.\u201d I don\u2019t know, right? It\u2019s just natural. It just comes out. And I\u2019m not that privy to the inner workings, and I feel the AI similarly. A lot of people have said, I\u2019ve had conversations on our show before with other cognitive scientists and computer scientists and they say it\u2019s really hard for the AI to answer questions, \u2019cause a lot of people say, \u201cWhy don\u2019t you just ask it?\u201d And it can\u2019t self-reflect either in an accurate way.<\/p>\n<p><strong>STROGATZ:<\/strong> This whole thought, the mystery of the black box. We use the term black box so often for the AI, but of course, our own intelligence is a black box, not just from mine to you, but even me to myself, as you\u2019re emphasizing. But it makes me wonder if there\u2019s a role for magicians because, you know, magicians or sleight-of-hand people are so good at showing us our own psychophysical limitations. How easily we\u2019re fooled, or the sorts of cognitive errors we tend to make, and there are people who are analogous to the magicians who show the deficits and common sense of the AIs, right? They\u2019re sort of playing games that are almost like magic tricks on the AIs. I wonder how revealing those will be, you know, in a serious scientific way.<\/p>\n<p>Well, Melanie has a lot more to say about the depth of AI cognition and understanding, and also how it might change whole fields of science, including math. We will be hearing more about that after the break.<\/p>\n<p>[<em>Music plays<\/em>]<\/p>\n<p><strong>STROGATZ:<\/strong> Welcome back to <em>The Joy of Why<\/em>. We\u2019re joined today by Santa Fe Institute computer scientist Melanie Mitchell.<\/p>\n<p><strong>STROGATZ:<\/strong> You have been a college professor for much of your life. When you\u2019re working with students they can get the answers right, but as you start to probe what they actually understand, you start to realize that they might be getting the right answers for the wrong reasons. They don\u2019t really know what they\u2019re doing, and that\u2019s important if you wanna be a helpful teacher. This brings up another point: competence versus performance. Can you expand on this idea and, what would it mean in the AI context?<\/p>\n<p><strong>MITCHELL:<\/strong> So competence versus performance is kind of an old distinction from psychology and linguistics. The idea is that you might have the competence for a particular cognitive capacity, but there might be some reasons why you can\u2019t perform the task that I\u2019m giving you. Like they have the competence, they could solve the problems, but they\u2019re just emotionally frozen. There\u2019s some performance block.<\/p>\n<p>But then there\u2019s the other way around, which is performance without competence. So if the student in your office hours, say, had memorized a problem from the textbook and the solution, but they didn\u2019t understand the general principle, so if you gave them a slightly different version of the problem, they couldn\u2019t do it. That\u2019s performance without competence.<\/p>\n<p><strong>STROGATZ:<\/strong> Okay. So if we would say that we\u2019re trying to work out ways of testing whether the AI understands, what would count as evidence? Suppose that, you\u2019re an AI advocate who said that these new systems, because we\u2019ve scaled them up or because we have some nice new architecture with world models or social models or whatever, we\u2019ve now crossed a threshold where they actually understand. It\u2019s not just that they can compute, they understand. What would count as evidence of understanding?<\/p>\n<p><strong>MITCHELL:<\/strong> Oh gosh. I hate to get pedantic about understanding, but there\u2019s so many different meanings of it.<\/p>\n<p><strong>STROGATZ:<\/strong> Ah.<\/p>\n<p><strong>MITCHELL:<\/strong> We had a talk here at Santa Fe Institute from a philosopher who broke down understanding into 25 different types.<\/p>\n<p><strong>STROGATZ:<\/strong> Aha. I didn\u2019t know what I was getting myself into with the question.<\/p>\n<aside class=\"post__aside mb0 relative mha post__aside--right hide-on-print\">\n<div class=\"post__aside__pullquote relative\">\n<div class=\"pullquote theme__text mb2 align-c\">\n<div class=\"mb1\">\n<p style=\"margin: 0in 0in .25in 0in;\"><span style=\"color: black;\">I think it\u2019s almost like a fallacy that if an AI system can do a bunch of tasks, it can do the job of a person that is associated with those tasks.<\/span><\/p>\n<\/p><\/div>\n<\/p><\/div>\n<\/p><\/div>\n<\/aside>\n<p><strong>MITCHELL:<\/strong> So there\u2019s like P understanding and G understanding and there\u2019s this very long typography of understanding. And I\u2019m not sure there is any sort of single notion of real understanding. One of the recent things I and my collaborators have been working on is looking at different dimensions of understanding. One example is you can get one of these language models or chatbots to generate a story. Just generate a short story about something, and they will. They\u2019ll generate a very beautiful little coherent short story. But then if you start asking them questions about the story, they will often will fail in weird ways.<\/p>\n<p><strong>STROGATZ:<\/strong> Hmm.<\/p>\n<p><strong>MITCHELL:<\/strong> even though they generated it. And I think the same thing is true in a lot of different tasks that they understand along one dimension but not along another dimension. And in some sense deep understanding might be just you understand across many different of these dimensions.<\/p>\n<p><strong>STROGATZ:<\/strong> Aha. That sounds like a promising direction. Let\u2019s talk about tasks a little more, because that\u2019s a phrase or a term that I\u2019ve seen in some of your writing, the phrase, the tyranny of tasks. What\u2019s that about?<\/p>\n<p><strong>MITCHELL:<\/strong> I first heard that, from Shannon Vallor, a philosopher. The idea is that in AI, the world is divided in terms of tasks. So when we think about what AI systems can do, people say, \u201cOh, they can make summaries. Let\u2019s test their ability to summarize articles.\u201d Or, \u201cLet\u2019s test their ability to answer questions about diagrams\u201d or I don\u2019t know, some other benchmark.<\/p>\n<p><strong>STROGATZ:<\/strong> Well, I mean, these days, they\u2019ve been benchmarked a lot on International Mathematical Olympiad, very hard high school problems, then there were research level problems. Now there\u2019s open problems that are unsolved in math. These are all like three levels of math benchmarks that are out there.<\/p>\n<p><strong>MITCHELL:<\/strong> Right, their capabilities are defined in terms of these benchmarks. You know, one benchmark might be the bar exam for law students, and they do really well on the bar exam. And so we say, \u201cOh, lawyers, you should be afraid. Your job is threatened because these AI systems are as getting as good as you are.\u201d Uh, But the way that we\u2019re defining that is by looking at how well they do on a specific set of questions or a task. And jobs as a whole are not the same as just one independent task after another. This is, I think it\u2019s almost like a fallacy that if an AI system can do a bunch of tasks, it can do the job of a person that is associated with those tasks.<\/p>\n<p>So just one example of this. So there\u2019s a famous quote from Geoffrey Hinton, where he said something like, \u201cAI systems are incredibly good at diagnosing or interpreting radiology images. Nobody should go to school anymore to be a radiologist. AI is gonna take all the jobs within five years.\u201d<\/p>\n<p>Well, that was 2016. That was 10 years ago. Now we actually have a shortage of radiologists. I don\u2019t know if that\u2019s because he said that, but uh, it turns out that even though AI systems can beat human doctors on these benchmarks, that\u2019s not the same as doing this job out in the real world, which is much more open-ended, which is not just a series of well-defined tasks.<\/p>\n<p><strong>STROGATZ:<\/strong> Still, it does leave you wondering, like in the case of radiology, you could imagine if they are really good at that task, then what\u2019s left for the human radiologist? Should we still be in that part of the game? Like in my own world of math, you know, if they\u2019re very good at proving theorems, but they\u2019re not so great yet at coming up with new concepts, or as we sometimes speak of it, theory building, right? There\u2019s this big distinction between problem-solving and theory building. So is it that we\u2019re sort of gonna find our niche, that we can do the parts that they don\u2019t do? So like in the case of radiology, they have the open-ended part but not the scan reading part? I guess that\u2019s what I\u2019m wondering.<\/p>\n<p><strong>MITCHELL:<\/strong> Yeah.<\/p>\n<p><strong>STROGATZ:<\/strong> Is that how it\u2019s gonna go?<\/p>\n<p><strong>MITCHELL:<\/strong> Maybe. I wouldn\u2019t be at all surprised if jobs like yours change quite a bit because of these new tools. These are going to become incredibly useful tools for mathematicians. So it might change your job. Just like when personal computers came out, but there\u2019s a fantastic book by um, George Lakoff and Rafael N\u00fa\u00f1ez about math and where ideas in math come from, via metaphors. And they feel that human embodiment is a very important part of understanding and mathematics.<\/p>\n<aside class=\"post__aside mb0 relative mha post__aside--right hide-on-print\">\n<div class=\"post__aside__pullquote relative\">\n<div class=\"pullquote theme__text mb2 align-c\">\n<div class=\"mb1\">\n<p style=\"margin: 0in 0in .25in 0in;\"><span style=\"color: black;\">I don\u2019t think that solving all the Erd\u00f6s problems means that the average person has to fear for their job.<\/span><\/p>\n<\/p><\/div>\n<\/p><\/div>\n<\/p><\/div>\n<\/aside>\n<p><strong>STROGATZ:<\/strong> Exactly. I think that\u2019s our only hope \u2019cause right now they the machines don\u2019t have great embodiment. And you\u2019re right, that a lot of great ideas in math are inspired by experience with the world. And that\u2019s what I was gonna say about applied math, that I feel like that\u2019s even more so than pure math, where we get so much inspiration from nature and from engineering and society and all that, that I think we have a lot more chance of being useful as humans in applied math.<\/p>\n<p>But I do think pure math will expire before applied math does, and maybe neither will. Maybe we\u2019ll just keep going forever. What does it look like to you? I mean, math is often thought of as some kind of gold standard like, the AI companies have a lot of use for math, right? They can demonstrate how good their systems are \u2019cause they can verify that they\u2019ve solved a problem or not.<\/p>\n<p><strong>MITCHELL:<\/strong> Well, that\u2019s a big question I have, which is, suppose that your prediction comes right and math, pure math expires in some sense for humans. What does that mean for other fields? Does that mean that these machines are on their way to taking over everything? Or is it more like 1997 or whatever it was that Deep Blue beat Kasparov and that actually beating the best human at chess did not necessarily mean that was gonna go anywhere in other fields.<\/p>\n<p><strong>STROGATZ:<\/strong> I don\u2019t know. What do you think? It feels to me like science is much more open-ended than math in that respect.<\/p>\n<p><strong>MITCHELL:<\/strong> Yeah, I believe that. I don\u2019t think that solving all the Erdos problems means that the average person has to fear for their job.<\/p>\n<p><strong>STROGATZ:<\/strong> Okay, now we have many different things on the table at that point. But even just in the world of pure brainiacs, whether it\u2019s scientists or mathematicians, just the fact that biology there are so many things to be measured, we have so much data that we could collect that we haven\u2019t collected, so many new ways of observing. I mean, that seems very inexhaustible to me compared to math.<\/p>\n<p><strong>MITCHELL:<\/strong> I agree. And even in physics, I think, which is maybe closer to math, there\u2019s so much you know, open-ended questions that aren\u2019t well-formulated, that don\u2019t have something like a proof that can be constructed.<\/p>\n<p><strong>STROGATZ:<\/strong> But so, I do feel like the hope for math is to continue to take inspiration from the real world. And von Neumann had said something like that too, that when math becomes too much art for art\u2019s sake, when it drifts too far from the source, for him the source was nature or reality, if it becomes too far removed it becomes sterile, said von Neumann.<\/p>\n<p>So I think this could be a a really good era for pure math if it starts taking more inspiration from nature. That\u2019s been less so in the 20th and 21st century, but I think if we go back to that, we can probably eke out a few more centuries of human pleasure in math.<\/p>\n<p><strong>MITCHELL:<\/strong> I\u2019ll just say there\u2019s this dictum in AI which is that easy things are hard and hard things are easy.<\/p>\n<p><strong>STROGATZ:<\/strong> Right.<\/p>\n<p><strong>MITCHELL:<\/strong> And pure math is seen by humans as like the most exalted exhibition of intelligence and brilliance. It\u2019s the hard thing, and yet we know that hard things are easier for machines and easier things are harder.<\/p>\n<p><strong>STROGATZ:<\/strong> Yep, and there\u2019s the word soft also, right? In science, we talk about the hard sciences and the soft sciences, and the soft sciences of economics and psychology and anthropology, and those are the really hard ones.<\/p>\n<p><strong>MITCHELL:<\/strong> Right.<\/p>\n<p><strong>STROGATZ:<\/strong> Well, so if we meet again in five years.<\/p>\n<p><strong>MITCHELL:<\/strong> The Joy of Gamma, or something.<\/p>\n<p><strong>STROGATZ:<\/strong> Yes, The Joy of Omega by then, right. What do you hope we would understand about AI systems by then? Or what kinds of tests would we want to be able to do that we can\u2019t do today?<\/p>\n<p><strong>MITCHELL:<\/strong> Yeah, I mean What I really hope will go well in the science of AI is this field called mechanistic interpretability, which is the neuroscience analog, where you\u2019re actually looking at the activations and the weights and the, you know, all the messy innards of the system, and understanding at a higher-level sort of what they are doing.<\/p>\n<aside class=\"post__aside mb0 relative mha post__aside--right hide-on-print\">\n<div class=\"post__aside__pullquote relative\">\n<div class=\"pullquote theme__text mb2 align-c\">\n<div class=\"mb1\">\n<p style=\"margin: 0in 0in .25in 0in;\"><span style=\"color: black;\">There\u2019s this dictum in AI which is that easy things are hard and hard things are easy.<\/span><\/p>\n<\/p><\/div>\n<\/p><\/div>\n<\/p><\/div>\n<\/aside>\n<p>These days, it\u2019s kind of a smallish subfield where people are trying to develop tools that do that, analogous to things like fMRI or whatever. And I don\u2019t think anybody\u2019s really figured out exactly how to do this the right way yet, but I\u2019m hoping that\u2019s something that we can accomplish, and then we would have a genuine way of understanding sort of their limitations, what they can do, what they can\u2019t do, what kinds of mistakes they\u2019re likely to make, and maybe how to fix them.<\/p>\n<p><strong>STROGATZ:<\/strong> Interesting that you put your finger on that because the first time I became aware of you, it was in connection with that in a broad sense. So what I\u2019m thinking of is back when you used to work on something that in the jargon was called GAs for CAs, genetic algorithms for cellular automata, you and Jim Crutchfield were looking at this problem of evolving algorithms that could solve a certain class of problems, hard computer science problems, and you were using this evolutionary algorithm to select better and better algorithms that kept improving through a kind of selection process.<\/p>\n<p>But then the part that you did that I found so creative is once you\u2019ve got a really good system, you looked at it in what felt to me like an analog of mechanistic interpretability. You tried to see what was making that system so smart, analyzing it in terms of particles that were colliding with each other according to certain rules in the diagrams. That\u2019s, I don\u2019t know if I\u2019ve summarized it reasonably well, but it seems like this is a longstanding interest of yours.<\/p>\n<p><strong>MITCHELL:<\/strong> Yeah. that\u2019s true. I hadn\u2019t made that connection exactly, but that\u2019s interesting.<\/p>\n<p><strong>STROGATZ:<\/strong> It is this, though. It\u2019s interpretability.<\/p>\n<p>It is interpretability. And it\u2019s also, I think, in the field of complex systems, people talk about this notion of emergence.<\/p>\n<p><strong>STROGATZ:<\/strong> Yeah.<\/p>\n<p><strong>MITCHELL:<\/strong> And we thought of that as a kind of emergent computation. And I think these AI systems also have emergent computations that are not easy to find, but they\u2019re there, and if we understood them better, we would understand how the system is actually working, doing what it does.<\/p>\n<p><strong>STROGATZ:<\/strong> Yeah, it\u2019s an interesting attitude. It feels honestly to me very sweet and very old school. This hope that\u2026 Okay, you\u2019re chuckling \u2019cause you see where I\u2019m going. It\u2019s a mean thing I\u2019m saying, but this conceit that we with our limited minds can keep doing science, you know, and we\u2019re gonna figure out how these AIs are doing what they\u2019re doing, and that\u2019s what our game will continue to be just like it always has been in science.<\/p>\n<p>And I, the dark side of me, thinks our days are numbered to be able to do that as these gadgets get bigger and bigger. Who says we can keep doing science on them and figuring them out? What\u2019s your reaction to that? We have nothing else to do. We have to try.<\/p>\n<p><strong>MITCHELL:<\/strong> That\u2019s an interesting question. Um, why do we do science in the first place? I mean, you know, we do science \u2019cause we wanna solve problems. That\u2019s one thing. But we also do science \u2019cause we\u2019re driven to understand things.<\/p>\n<p><strong>STROGATZ:<\/strong> Yes.<\/p>\n<p><strong>MITCHELL:<\/strong> You see this in little children. They\u2019re driven to understand. Often one of their first words is why. They ask it constantly. So I think that\u2019s a human drive, and it\u2019s hard to fight against that. And that\u2019s why you and I both went into science, it\u2019s important to us.<\/p>\n<p>Now, I was a little despairing when I went to a panel discussion at a conference on the role of AI in science. And there were a bunch of famous people on the panel talking about how AI was going to revolutionize weather prediction, and genetics, and cosmology, and you name it. And I asked them at the end \u201cWell, like, is this going to contribute to human understanding of the world?\u201d And they\u2019re like, \u201cWhy should we care about that?\u201d<\/p>\n<p><strong>STROGATZ:<\/strong> Yeah. To me, this is the bifurcation that we\u2019re all thinking about now. \u2019Cause science has this double-edged aspect, that it gives us pleasure, we like figuring things out, there is the joy of why, and as you say, it\u2019s deep in our species. So yes, we\u2019re curious, but then there\u2019s the other side that for so long science has been this instrumental thing that helps us in technology and medicine.<\/p>\n<p>And I guess the question I have, and I think a lot of us have, is will we continue to take pleasure in the joy of curiosity when we are no longer the best at solving the important problems? But let me ask you one last thing, for people who haven\u2019t heard our earlier conversation, what was your draw to this field, and if you were starting out today, do you think you\u2019d have the same kind of curiosity?<\/p>\n<p><strong>MITCHELL:<\/strong> Yeah, that\u2019s a great question. When I was a child, I loved logic puzzles, like the knights and the knaves. The knights who always told the truth and the knaves who always lied. There\u2019s a fun several books by Raymond Smullyan, a mathematician who wrote a bunch of puzzles in this genre that I absolutely loved.<\/p>\n<aside class=\"post__aside mb0 relative mha post__aside--right hide-on-print\">\n<div class=\"post__aside__pullquote relative\">\n<div class=\"pullquote theme__text mb2 align-c\">\n<div class=\"mb1\">\n<p>Why do we do science? We do science \u2019cause we wanna solve problems. We also do science \u2019cause we\u2019re driven to understand things.<\/p>\n<\/p><\/div>\n<\/p><\/div>\n<\/p><\/div>\n<\/aside>\n<p>When I got to college, I read Douglas Hofstadter\u2019s book, G\u00f6del, Escher, Bach, which was the real-world version of these in a way. I mean, he was talking about G\u00f6del\u2019s theorem and paradoxes in mathematical logic and how all this related to cognition and thinking and creativity and so on. And I was just completely blown away and that this is what I wanna do in my life. I didn\u2019t exactly know what it was, but it seemed like it might be artificial intelligence. So I pursued Doug as an advisor and got to join his group, and was studying analogy via a new set of puzzles which were analogy puzzles. And, I was very entranced by all of that.<\/p>\n<p>If I were that age today, I would be worried. In fact, I have a son who is getting a PhD in machine learning, and he wants to do research in machine learning, but he\u2019s actually quite nervous that there will be no more roles for humans doing research in machine learning because AI will be doing all the research in machine learning and improving itself and so on and so forth. And I wonder if I\u2019d think the same thing. I don\u2019t know.<\/p>\n<p><strong>STROGATZ:<\/strong> Maybe we do have to revisit this in five years because we may know by then. Given how fast everything is going, who knows? I really appreciate your spending time with us. This has been wide-ranging, a little bit amorphous conversation, but it\u2019s just wide open and I can\u2019t think of a better guide to it. Thank you very much for joining us.<\/p>\n<p><strong>MITCHELL:<\/strong> Thanks, Steve. It\u2019s been great.<\/p>\n<p>[<em>Music plays<\/em>]<\/p>\n<p><strong>LEVIN:<\/strong> Hmm. Hmm. I, I just remember being a student and learning Newton\u2019s laws for the first time, and then Kepler\u2019s laws, which really make Newton\u2019s laws beautiful, this application to the celestial cycles. I didn\u2019t think, \u201cOh, I\u2019m not the best at this, therefore I shouldn\u2019t learn it.\u201d Nor did I think, unless I one day become the best at this, I cannot feel pleasure or joy in my experience of acquiring this information.\u201d<\/p>\n<p>Of course, lots of people study things that other people already know and are better at. So I, I sort of wonder if maybe the AI will know things before us, but we will still need to acquire the understanding ourselves, and in that acquisition is a similar experience. Instead of maybe the AI will be a filter between us and interrogating nature directly, but we\u2019ll still be acquiring, I don\u2019t know, the knowledge and having that experience. I\u2019m not sure. Maybe it\u2019s all gonna pass us by.<\/p>\n<p><strong>STROGATZ:<\/strong> I\u2013 Well, let\u2019s explore this a little more. I like especially your emphasis on not being the best, and how, in a way, unfraught that is. I, I learned as soon as I went to college what it means to not be the best. You know, this, this fixation with being the number one, especially in an age of optimization. There\u2019s so many optimization algorithms. We talk about faster, cheaper. But in our own lives, very often we\u2019re not the best. I\u2019m certainly not the best tennis player. I love to play tennis. I\u2019m not the best chess player, and I\u2019m still happy to play chess. And try to be the best dad, but I may not be. But still, all these things are worth doing for their own sake, right? They give us pleasure.<\/p>\n<p>I do feel very philosophical and almost religious about this. Like, we get a little time on Earth alive and, you know, these questions about AI do tap into questions about the meaning of life. What are we trying to do? If the meaning of life is that you\u2019re gonna be the best in some domain or you\u2019re gonna make a discovery that\u2019s gonna change the world, then most people will have a meaningless life, and I just don\u2019t wanna believe that\u2019s the correct version of the meaning of life.<\/p>\n<p>It was not for my dad. He didn\u2019t even get to go to college. You know, he grew up in the Depression. That was not an option. His life was being a good parent and taking care of the people that bought shoes at the shoe store that he had. And he knew everyone\u2019s shoe size in our little town, and he left a good name when he died. People remembered him well.<\/p>\n<p><strong>LEVIN:<\/strong> Right.<\/p>\n<p><strong>STROGATZ:<\/strong> So okay. What is that doing on our show here about science?<\/p>\n<p><strong>LEVIN:<\/strong> Well, I think that let\u2019s say the meaning for some people of life has to do with acquisition, acquiring wealth. They\u2019re gonna love this stuff, right? \u2019Cause there\u2019s gonna be this new tool that simply leverages all kinds of buttons that they now have faster access to and can exploit and acquire more wealth.<\/p>\n<p>There are people who found meaning in singing songs or writing poetry or being novelists or doing math, and, and I think all of those fields are a little more nervous, right? About reevaluating what the place is going to be for them and, and how to secure that place and how to think about it.<\/p>\n<p>If I\u2019m playing games of what may or may not happen, I mean, there is still a world in which AI is like a supercomputer, and we\u2019ve talked about this before, Steve. Just \u2019cause a supercomputer can crunch all of these numbers, if it presents it to us as a string of symbols, even though it has, in some sense, an answer, it\u2019s not a meaningful answer for us, and none of us value it.<\/p>\n<p>We still, as human beings, have a very important role between us and a supercomputer rendering an image of a galaxy or looking at an image of a biomedical neural map. It hasn\u2019t actually robbed scientists of their work. And so it might be that it really will continue to be a tool and not simply something that overtakes and discards us.<\/p>\n<p><strong>STROGATZ:<\/strong> Well, that\u2019s the question, right? I think there are two plausible scenarios. One is that it continues to be a tool, and we always have some essential role in science and math at the cutting edge. The other option is, and actually in my heart I believe this is the case, that we will not be at the cutting edge, and that will happen very soon. And, so then what is the point?<\/p>\n<p>Then I feel like it\u2019s still meaningful, just like when I was in high school and I discovered things about math. They were discoveries to me. They were not discoveries to the world, you know? I think we may have to all settle for that. We\u2019re not gonna be making genuine discoveries for the world.<\/p>\n<p>The AIs will be doing that. I really do believe that\u2019s gonna happen very soon. I may be wrong. I mean, there may be fundamental reasons why the AIs won\u2019t be able to do that. For instance, they don\u2019t have bodies, they don\u2019t have social life, you know, there\u2019s a lot\u2026 But I just think all that stuff will be solved before long. Anyway, what\u2019s your take?<\/p>\n<p><strong>LEVIN:<\/strong> Well, I think there\u2019s a difference between, making discoveries and understanding, and I guess that\u2019s kind of what I mean in examples. In some sense, maybe the su- supercomputer made the discovery before the person did, but we still say the person did \u2019cause the discovery didn\u2019t count as a discovery until they rendered it in a way that human beings could comprehend.<\/p>\n<p>But, I honestly don\u2019t know. I am not incredibly saddened or pessimistic, so I guess I would have to say that in my heart, intuitively, I am not terrified of this prospect. Maybe I should be, but maybe it\u2019s just sort of a bliss of being naive and I\u2019m just gonna wait for it to sneak up on me.<\/p>\n<p><strong>STROGATZ:<\/strong> There is one thing I think we can be very optimistic about and hopeful about, which is I think we\u2019re gonna have a glorious golden age of science where we will understand, and discoveries by the AIs or by people in conjunction with AIs, that\u2019s all gonna be happening in the next, whatever, five, 10, 15 years, and it\u2019s gonna be a spectacular fireworks time for science. And I think that hopefully with any luck, we\u2019ll be alive to see all that.<\/p>\n<p><strong>LEVIN:<\/strong> Yeah, there\u2019s definitely going to be a transition period where people are moving it fast and furious, and they\u2019re part of the story, and there\u2019s great accomplishment, and it will be exciting to see. I know people, very accomplished, who are very excited about using it. Use it every day. They have multiple things going on, and they just feel like their productivity has doubled or more. And they\u2019re excited, they\u2019re enjoying themselves. I think there\u2019s really nothing we can do but chime in and participate in this, at least, transition phase before we\u2019re obsolete.<\/p>\n<p><strong>STROGATZ:<\/strong> Well, I\u2019m getting choked up just thinking about it. Thanks, Janna. It\u2019s always great to see you, and we\u2019ll see you next time on <em>The Joy of Why<\/em>.<\/p>\n<p><strong>LEVIN:<\/strong> Thanks, Steve.<\/p>\n<p>[<em>Music plays<\/em>]<\/p>\n<p><strong>LEVIN:<\/strong> If you\u2019re enjoying <em>The Joy of Why<\/em> and you\u2019re not already subscribed, hit the subscribe or follow button wherever you\u2019re listening. You can also leave a review for the show. It helps people find this podcast. Find articles, newsletters, videos and more at quantamagazine.org.<\/p>\n<p><strong>STROGATZ:<\/strong> <em>The Joy of Why <\/em>is a podcast from <em>Quanta Magazine<\/em>, an editorially independent publication supported by the Simons Foundation. Funding decisions by the Simons Foundation have no influence on the selection of topics, guests, or other editorial decisions in this podcast or in <em>Quanta Magazine<\/em>. The Joy of Why is produced by PRX Productions.<\/p>\n<p>The production team is Caitlin Faulds, Jade Abdul-Malik, Genevieve Sponsler, and Merritt Jacob. The executive producer of PRX Productions is Jocelyn Gonzales. Edwin Ochoa is our project manager.<\/p>\n<p>From <em>Quanta Magazine<\/em>, Simon Frantz and Samir Patel provided editorial guidance, with support from Samuel Velasco, Kit Sudol, Simone Barr, and Michael Kanyongolo. Samir Patel is <em>Quanta\u2019s<\/em> Editor-in-Chief.<\/p>\n<p>The episode art is by Chanelle Nibbelink and our logo is by Jaki King and Kristina Armitage. Special thanks to Garth Avery at the Cornell Broadcast Studio.<\/p>\n<p>I\u2019m your host, Steve Strogatz. If you have any questions or comments, please email us at <a href=\"https:\/\/www.quantamagazine.org\/cdn-cgi\/l\/email-protection\" class=\"__cf_email__\" data-cfemail=\"59282c38372d38192a303436372a3f362c373d382d30363777362b3e\">[email\u00a0protected]<\/a>. Thanks for listening.<\/p>\n<p>[<em>Music fades<\/em>]<\/p>\n<\/div>\n<p><a href=\"https:\/\/www.quantamagazine.org\/are-we-thinking-correctly-about-ai-intelligence-20260820\/?utm_source=tldrai\">Source link <\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>When an LLM answers a question, is it reasoning like humans, or just producing text that looks like reasoning? The distinction isn\u2019t just philosophical, this determines what we can trust AI to do, how closely we need to supervise it, and ultimately what its real-world impact will turn out to be. Melanie Mitchell at the [&hellip;]<\/p>\n","protected":false},"author":16,"featured_media":23500,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[143],"tags":[],"class_list":["post-23499","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai"],"_links":{"self":[{"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/posts\/23499","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/users\/16"}],"replies":[{"embeddable":true,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/comments?post=23499"}],"version-history":[{"count":0,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/posts\/23499\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/media\/23500"}],"wp:attachment":[{"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/media?parent=23499"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/categories?post=23499"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/tags?post=23499"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}