Daniel Kahneman: Thinking Fast and Slow, Deep Learning, and AI

[0] The following is a conversation with Daniel Kahneman, winner of the Nobel Prize in Economics for his integration of economic science with the psychology of human behavior, judgment, and decision -making.

[1] He's the author of the popular book Thinking Fast and Slow that summarizes in an accessible way his research of several decades, often in collaboration with Amos Tversky, a cognitive biases, prospect theory, and happiness.

[2] The central thesis of this work is the dichotomy between two modes of thought.

[3] What he calls system one is fast, instinctive, and emotional.

[4] System two is slower, more deliberative and more logical.

[5] The book delineates cognitive biases associated with each of these two types of thinking.

[6] His study of the human mind and his peculiar and fascinating limitations are both instructive and inspiring for those of us seeking to engineer intelligence systems.

[7] This is the artificial intelligence podcast.

[8] If you enjoy it, subscribe on YouTube, give it five stars on Apple Podcast, follow on Spotify, support it on Patreon, or simply connect with me on Twitter, Alex Friedman, spelled F -R -I -D -M -A -N.

[9] I recently started doing ads at the end of the introduction.

[10] I'll do one or two minutes after introducing the episode and never any ads in the middle that can break the flow of the conversation.

[11] I hope that works for you and doesn't hurt the listening experience.

[12] This show is presented by Cash App, the number one finance app in the App Store.

[13] I personally use Cash App to send money to friends, but you can also use it to buy, sell, and deposit Bitcoin in just seconds.

[14] Cash App also has a new investing feature.

[15] You can buy fractions of a stock, say $1 worth, no matter what the stock price is.

[16] Brokers services are provided by Cash App investing, a subsidiary of Square and member SIPC.

[17] I'm excited to be working with Cash App to support one of my favorite organizations called First, best known for their first robotics and Lego competitions.

[18] They educate and aspire hundreds of thousands of students in over 110 countries and have a perfect rating, a charity navigator, which means that donated money is used to maximum effectiveness.

[19] When you get Cash App from the App Store or Google Play and use code Lex Podcast, you'll get $10, and Cash App will also donate $10 to First, which again is an organization that I've personally seen inspire girls and boys the dream of engineering a better world.

[20] And now, here's my conversation with Daniel Kahneman.

[21] You tell a story of an SS soldier early in the war, World War II, in Nazi -occupied France and Paris, where you grew up.

[22] He picked you up and hugged you and showed you a picture of a boy, maybe not realizing that you were Jewish.

[23] Not maybe, certainly not.

[24] So I told you I'm from the Soviet Union that was significantly impacted by the war as well, and I'm Jewish as well.

[25] What do you think World War II taught us about human psychology broadly?

[26] Well, I think the only big surprise is the extermination policy, genocide, by the German people.

[27] That's when you look back on it.

[28] And, you know, I think that's a major surprise.

[29] It's a surprise because...

[30] It's a surprise that they could do it.

[31] It's a surprise that enough people willingly participated in that.

[32] This is a surprise.

[33] Now it's no longer a surprise, but it's changed many people's views, I think, about human beings.

[34] Certainly for me, the Achman trial in...

[35] that teaches you something because it's very clear that if it could happen in Germany, it could happen anywhere.

[36] It's not that the Germans were special.

[37] This could happen anywhere.

[38] So what do you think that is?

[39] Do you think we're all capable of evil?

[40] We're all capable of cruelty?

[41] I don't think in those terms.

[42] I think that what is certainly possible is you can dehumanize people.

[43] so that you treat them not as people anymore, but as animals.

[44] And the same way that you can slaughter animals without feeling much of anything.

[45] It can the same.

[46] And when you feel that I think the combination of dehumanizing the other side and having uncontrolled power over other people, I think that doesn't bring out the most generous aspect of human nature.

[47] So that Nazi soldier, you know, he was a good man. I mean, you know, and he was perfectly capable of killing a lot of people, and I'm sure he did.

[48] But what did the Jewish people mean to Nazis?

[49] So what, the dismissal of Jewish as worthy of...

[50] Again, this is surprising that it was so extreme, but it's not one thing in human nature.

[51] I don't want to call it evil, but the distinction between the in -group and the out -group, that is very basic.

[52] So that's built in.

[53] The loyalty and affection towards in -group and the willingness to dehumanize the out group, that is inhuman nature.

[54] And that's what I think probably didn't need the Holocaust to teach us that.

[55] But the Holocaust is a very sharp lesson of, you know, what can happen to people and what people can do.

[56] So the effect of the in -group and the out -group?

[57] You know, it's clear.

[58] Those were people, you know, you could shoot.

[59] you could, you know, they were not human, there were not, there was no empathy or very, very little empathy left.

[60] So occasionally, you know, there might have been, and very quickly, by the way, the empathy disappeared if there was initially, and the fact that everybody around you was doing it, that completely the group doing it, and everybody's shooting, and everybody Jews, I think that makes it permissible.

[61] Now, how much, you know, whether it could happen in every culture or whether the Germans were just particularly efficient and disciplined, so they could get away with it.

[62] That's a question.

[63] It's an interesting question.

[64] Are these artifacts of history or is it human nature?

[65] I think that's really human nature.

[66] You know, you put some people in a position of power relative to other people, and then they become less human, they become different.

[67] But in general, in war, outside of concentration camps in World War II, it seems that war brings out darker size of human nature, but also the beautiful things about human nature.

[68] Well, you know, I mean, would it be...

[69] What it brings out is the loyalty among soldiers.

[70] I mean, it brings out the bonding, male bonding, I think is a very real thing that happens.

[71] And so, and there is a certain thrill to friendship, and there is certainly a certain thrill to friendship under risk and to shared risk.

[72] And so people have very profound emotions up to the point where it gets so traumatic, that little is left.

[73] So let's talk about psychology a little bit.

[74] In your book, Thinking Fast and Slow, you described two modes of thoughts as the one, the fast instinctive, an emotional one, the system two, the slower, deliberate, logical one, at the risk of asking Darwin to discuss theory of evolution, Can you describe distinguishing characteristics for people who have not read your book of the two systems?

[75] Well, I mean, the word system is a bit misleading, but at the same time it's misleading.

[76] It's also very useful.

[77] But what I call system one, it's easier to think of it as a family of activities.

[78] And primarily the way I describe it is there are different ways.

[79] for ideas to come to mind.

[80] And some ideas come to mind automatically.

[81] And the example, a standard example, is two plus two, and then something happens to you.

[82] And in other cases, you've got to do something.

[83] You've got to work in order to produce the idea.

[84] And my example, I always give the same pair of numbers, is 27 times 14, I think.

[85] You have to perform some algorithm in your head, some steps.

[86] Yes.

[87] And it takes time.

[88] It's a very difference.

[89] Nothing comes to mind except something comes to mind, which is the algorithm, I mean, that you've got to perform.

[90] And then it's work, and it engages short -term memory, and it engages executive function, and it makes you incapable of doing other things at the same time.

[91] So the main characteristic of system, too, that there is mental effort involved, and there is a limited capacity.

[92] for mental effort, whereas system one is effortless, essentially.

[93] That's the major distinction.

[94] So you talk about their, you know, it's really convenient to talk about two systems, but you also mention just now and in general that there is no distinct two systems in the brain from a neurobiological, even from psychology perspective.

[95] But why does it seem to, from the experiments you've conducted, there does seem to be, kind of emergent two modes of thinking.

[96] So at some point these kinds of systems came into a brain architecture.

[97] Maybe mammals share it or do you not think of it at all in those terms that it's all a mush and these two things just emerge.

[98] You know, evolutionary theorizing about this is cheap and easy.

[99] Fair enough.

[100] So it's the way I think about it is that it's very clear that animals have a perceptual system and that includes an ability to understand the world, at least to the extent that they can predict, they can't explain anything, but they can anticipate what's going to happen.

[101] And that's a key form of understanding the world.

[102] And my crude idea is that we, what I call system two, well, system two grew out of this.

[103] And, you know, there is language and there is the capacity of manipulating ideas and the capacity of imagining futures and of imagining counterfactuals thing that haven't happened and to do conditional thinking.

[104] and there are really a lot of abilities that without language and without the very large brain that we have compared to others would be impossible.

[105] Now, system one is more like what the animals are, but system one also can talk.

[106] I mean, it has language, it understands language.

[107] Indeed, it speaks for us.

[108] I mean, you know, I'm not choosing every word as a deliberate process.

[109] the words, I have some idea, and then the words come out.

[110] And that's automatic and effortless.

[111] And many of the experiments you've done is to show that, listen, system one exists and it does speak for us, and we should be careful about the voice it provides.

[112] Well, I mean, you know, we have to trust it because it's the speed at which it acts.

[113] System 2.

[114] If we depend on System 2 for survival, we wouldn't survive very long because it's very slow.

[115] Yeah, crossing the street.

[116] Crossing the street.

[117] I mean, many things depend on there being automatic.

[118] One very important aspect of System 1 is that it's not instinctive.

[119] You use the word instinctive.

[120] It contains skills that clearly have been learned.

[121] So that skilled behavior like driving a car or speaking, in fact, skilled behavior has to be learned and so it doesn't you know you don't come equipped with with driving you have to learn how to drive and and you have to go through a period where driving is not automatic before it becomes automatic so yeah you construct I mean this is where you talk about heuristic and biases is you to make it automatic you create a pattern and then system one essentially matches a new experience against a previously seen pattern and when that match is not a good one that's when the cognitive all the mess happens but it's most of the time it works and so it's pretty most of the time the anticipation of what's going to happen next is correct and and most of the time the plan about what you have to do is correct and so most of the time everything works just fine but What's interesting, actually, is that in some sense, system one is much better at what it does than system two is at what it does.

[122] That is, there is that quality of effortlessly solving enormously complicated problems, which clearly exists, so that a chess player, a very good chess player, all the moves that come to their mind are strong moves.

[123] So all the selection of strong moves, happens unconsciously and automatically and very, very fast.

[124] And all of that is in system one.

[125] So a system two verifies.

[126] So along this line of thinking, really what we are are machines that construct pretty effective system one.

[127] You could think of it that way.

[128] So we're now talking about humans, but if we think about building artificial intelligence systems, robots.

[129] Do you think all the features and bugs that you have highlighted in human beings are useful for constructing AI systems?

[130] So both systems are useful for perhaps instilling in robots?

[131] What is happening these days is that actually what is happening in deep learning is more like a system one product than like a system two product.

[132] I mean, deep learning matches patterns and anticipate what's going to happen, so it's highly predictive.

[133] What deep learning doesn't have, and many people think that this is a critical, it doesn't have the ability to reason, so there is no system two there.

[134] But I think very importantly, it doesn't have any causality or any to represent meaning and to represent real interaction.

[135] So until that is solved, what can be accomplished is marvelous and very exciting but limited.

[136] That's actually really nice to think of current advances in machine learning is essentially system one advances.

[137] So how far can we get with just system one?

[138] If we think of deep learning and artificial technology systems and some more.

[139] I mean, you know, it's very clear that deep mind has already gone way beyond what people thought was possible.

[140] I think the thing that has impressed me most about the developments in AI is the speed.

[141] It's that things, at least in the context of deep learning, and maybe this is about to slow down, but things moved a lot faster than anticipated.

[142] The transition from solving chess to solving Go was, I mean, that's bewildering how quickly it went.

[143] The move from AlphaGo to Alpha Zero is sort of bewildering the speed at which they accomplished that.

[144] Now, clearly, so there are many problems that you can solve that way, but there are some problems for which you need something else.

[145] Something like reasoning.

[146] Well, reasoning and also, you know, one of the real mysteries, psychologist Gary Marcus, who is also a critic of AI, I mean, he, what he points out, and I think he has a point, is that humans learn quickly, children don't need a million examples, they need to two or three examples.

[147] So clearly there is a fundamental difference.

[148] And what enables a machine to learn quickly, what you have to build into the machine, because it's clear that you have to build some expectations or something in the machine to make it ready to learn quickly, that at the moment seems to be unsolved.

[149] I'm pretty sure that deep mind is working on it, but if they have solved it, I haven't heard yet.

[150] They're trying to actually, them and Open AI are trying to start to get to use neural networks to reason.

[151] So assemble knowledge, of course, causality is, temporal causality is out of reach to most everybody.

[152] You mentioned the benefits of System 1 is essentially that it's fast, allows us to function in the world.

[153] Fast and skilled, yeah.

[154] It's skill.

[155] And it has a model of the world.

[156] You know, in a sense, I mean, there was the earlier phase of AI attempted to model reasoning.

[157] And they were moderately successful, but, you know, reasoning by itself doesn't get you much.

[158] Deep learning has been much more successful in terms of, you know, what they can do.

[159] But now, it's an interesting question.

[160] it's approaching its limits.

[161] What do you think?

[162] I think absolutely.

[163] So I just talked to Gian Lecun.

[164] He mentioned, you know.

[165] I know him.

[166] So he thinks that the limits were not going to hit the limits with neural networks, that ultimately this kind of system one pattern matching will start to start to look like system two without significant transformation of the architecture.

[167] So I'm more with the majority of the people who think that, yes, all networks will hit a limit in their capability.

[168] On the one hand, I have heard him tell Demisassabis essentially that what they have accomplished is not a big deal, that they have just touched, that basically they can't do unsupervised learning in an effective way.

[169] But you're telling me that he thinks that the current, within the current architecture, you can do causality and reasoning?

[170] So he's very much a pragmatist in a sense that's saying that we're very far away, that there's still, I think there's this idea that he says is we can only see one or two mountain peaks ahead and there might be either a few more after or thousands more after, yeah.

[171] So that kind of idea.

[172] I heard that metaphor.

[173] Right.

[174] But nevertheless, it doesn't see the final answer not fundamentally looking like one that we currently have.

[175] So neural networks being a huge part of that.

[176] Yeah, I mean, that's very likely because pattern matching is so much of what's going on.

[177] And you can think of neural networks as processing information sequentially.

[178] Yeah, I mean, you know, there is an important aspect to, for example, you get systems that translate and they do a very good job.

[179] but they really don't know what they're talking about.

[180] And for that, I'm really quite surprised.

[181] For that, you would need an AI that has sensation, an AI that is in touch with the world.

[182] Yes, as self -awareness, and maybe even something resembles consciousness kind of ideas.

[183] Certainly awareness of what's going on so that the words have meaning or can get, are in touch with some perception or some action.

[184] Yeah, so that's a big thing for Jan and what he refers to as grounding to the physical space.

[185] So that's what we're talking about the same thing.

[186] Yeah, so how do you ground?

[187] I mean, the grounding, without grounding, then you get a machine that doesn't know what it's talking about, because it is talking about the world, ultimately.

[188] The question, the open question, is what it means to ground.

[189] I mean, we're very human -centric in our thinking, but what does it mean for a machine to understand what it means to be in this world?

[190] Does it need to have a body?

[191] Does it need to have a finiteness like we humans have?

[192] All of these elements, it's a very...

[193] It's an operation.

[194] You know, I'm not sure about having a body, but having a perceptual system, having a body would be very helpful, too.

[195] I mean, if you think about human, mimicking human, but having a perception that seems to be essential so that you can build, you can accumulate knowledge about the world.

[196] So if you can imagine a human completely paralyzed and there is a lot that the human brain could learn, you know, with a paralyzed body.

[197] So if we got a machine that could do that, that would be a big deal.

[198] And then the flip side of that, that something you see in children and something in machine learning world is called active learning, maybe it is also, is being able to play with the world.

[199] How important for developing system one or system two do you think it is to play with the world?

[200] To be able to interact with the world.

[201] Certainly a lot of what you learn as you learn to anticipate the outcomes of your actions.

[202] I mean, you can see that how babies learn it.

[203] you know, with their hands, how they learn, you know, to connect, you know, the movements of their hands with something that clearly is something that happens in the brain, and the ability of the brain to learn new patterns.

[204] So, you know, it's the kind of thing that you get with artificial limbs, that you connect it, and then people learn to operate the artificial limb, you know, really impressively quickly, at least, from what I hear.

[205] So we have a system that is ready to learn the world through action.

[206] At the risk of going into way too mysterious of land, what do you think it takes to build a system like that?

[207] Obviously, we're very far from understanding how the brain works, but how difficult is it to build this?

[208] mind of ours.

[209] You know, I mean, I think that Yan Le Coon's answer, that we don't know how many mountains there are, I think that's a very good answer.

[210] I think that, you know, if you look at what Ray Kurzweil is saying, that strikes me as off the wall.

[211] But I think people are much more realistic than that where actually Demisasabis is and Jan is.

[212] and so the people are actually doing the work fairly realistic, I think.

[213] To maybe phrase it another way, from a perspective not of building it, but from understanding it, how complicated are human beings in the following sense?

[214] You know, I work with autonomous vehicles and pedestrians, so we tried to model pedestrians.

[215] How difficult is it to model a human being, their perception, of the world, the two systems they operate under, sufficiently to be able to predict whether the pedestrian is going to cross the road or not?

[216] I'm fairly optimistic about that, actually, because what we're talking about is a huge amount of information that every vehicle has and that feeds into one system, into one gigantic system and so anything that any vehicle learns becomes part of what the whole system knows and with a system multiplier like that there is a lot that you can do so human beings are very complicated but and you know system is going to make mistakes but human makes mistakes I think that they'll be able to I think they are able to anticipate pedestrians otherwise a lot would happen, they're able to get into a roundabout and into traffic, so they must know both to expect or to anticipate how people will react when they're sneaking in.

[217] And there's a lot of learning that's involved in that.

[218] Currently, the pedestrians are treated as things that cannot be hit and they're not treated as agents with whom you interact in a game theoretic way so I mean it's not it's a totally open problem and every time somebody tries to solve it seems to be harder than we think and nobody's really tried to seriously solve the problem of that dance because I'm not sure if you've thought about the problem of pedestrians but you're really putting your life in the hands of the driver.

[219] You know, there is a dance, there's part of the dance, that would be quite complicated, but, for example, when I cross the street and there is a vehicle approaching, I look the driver in the eye, and I think many people do that.

[220] And, you know, that's a signal that I'm sending, and I would be sending that machine to an autonomous vehicle, and it had better understand it, because it means I'm crossing.

[221] so and there's another thing you do that actually so i'll tell you what you do because we watch i've watched hundreds of hours of video on this is when you step in the street you do that before you step in the street and when you step in the street you actually look away yeah yeah uh now what what is that what that's saying is mean you're trusting that the car who hasn't slowed down yet will slow down.

[222] Yeah.

[223] And you're telling him, I'm committed.

[224] I mean, this is like in a game of chicken, so I'm committed.

[225] And if I'm committed, I'm looking away.

[226] So there is, you just have to stop.

[227] So the question is whether a machine that observes that needs to understand mortality.

[228] Here, I'm not sure that it's got to understand so much as it's got to anticipate.

[229] And here, But, you know, you're surprising me because here I would think that maybe you can anticipate without understanding because I think this is clearly what's happening in playing goal or in playing chess.

[230] There's a lot of anticipation and there is zero understanding.

[231] So I thought that you didn't need a model of the human and a model of the human mind to avoid hitting.

[232] pedestrians.

[233] But you are suggesting that actually you do, yeah, you do.

[234] Then it's, then it's a lot harder.

[235] So this is, and I have a follow question to see where your intuition lies is it seems that almost every robot human collaboration system is a lot harder than people realize.

[236] So do you think it's possible for robots and humans to collaborate successfully?

[237] We talked a little bit about semi -autonomous vehicles, like in the Tesla, autopilot, but just in tasks in general.

[238] If you think we talked about current neural networks being kind of System 1, do you think those same systems can borrow humans for System 2 type tasks and collaborate successfully?

[239] Well, I think that in any system where humans and the machine interact, that the human, the will be superfluous within a fairly short time, that is, if the machine is advanced enough so that it can really help the human, then it may not need the human for a long time.

[240] Now, it would be very interesting if there are problems that for some reason the machine doesn't, cannot solve, but that people could solve, then you would have to build into the machine and ability to recognize that it is in that kind of problematic situation and to call the human.

[241] That cannot be easy without understanding.

[242] It must be very difficult to program a recognition that you are in a problematic situation without understanding the problem.

[243] That's very true.

[244] In order to understand the full scope of situations that are problematic, you almost need to be smart enough to solve all those problems.

[245] It's not clear to me how much the machine will need the human.

[246] I think the example of chess is very instructive.

[247] I mean, there was a time at which Kasparov was saying that human machine combinations will beat everybody.

[248] Even stockfish doesn't need people.

[249] and alpha zero certainly doesn't need people.

[250] The question is, just like you said, how many problems are like chess and how many problems are the ones where are not like chess?

[251] Well, every problem probably in the end is like chess.

[252] The question is how long is that transition period?

[253] I mean, you know, that's a question I would ask you in terms of, I mean, autonomous vehicle just driving is probably a lot more complicated than go to solve that.

[254] Yes.

[255] And that's surprising because it's open.

[256] No, I mean, you know, that's not surprising to me because the, because there is a hierarchical aspect to this, which is recognizing a situation and then within the situation bringing up the relevant knowledge.

[257] And for that hierarchical type of system to work, you need a more complicated system than we currently have.

[258] A lot of people think because as human beings, this is probably the cognitive biases.

[259] They think of driving as pretty simple because they think of their own experience.

[260] This is actually a big problem for AI researchers or people thinking about AI because they evaluate how hard a particular problem is based on very limited knowledge basically on how hard it is for them to do the task.

[261] And then they take for granted, maybe you can speak to that because most people tell me driving is trivial, and humans, in fact, are terrible at driving is what people tell me. And I see humans, and humans are actually incredible at driving, and driving is really terribly difficult.

[262] Yeah.

[263] So is that just another element of the effects that you've described in your work on the psychology side?

[264] No, I mean, I haven't really, you know, I would say that my research has contributed nothing to understanding the ecology and to understanding the structure of situations and the complexity of problems.

[265] So all we know is very clear that that go, it's endlessly complicated, but it's very constrained.

[266] And in the real world, there are far fewer constraints and many more potential surprises.

[267] So that's obviously, because it's not always obvious to people, right?

[268] So when you think about...

[269] Well, I mean, you know, people thought that reasoning was hard and perceiving was easy, but they quickly learned that actually modeling vision was tremendously complicated and modeling, even proving theorems, was relatively straightforward.

[270] To push back on that a little bit on the quickly part, it took several decades to learn that, and most people still haven't learned that.

[271] I mean, our intuition, of course, AI researchers have, but you drift a little bit outside the specific AI field, the intuition is still perceptual.

[272] Oh, yeah.

[273] No, I mean, that's true.

[274] I mean, the intuitions of the public haven't changed radically.

[275] And they are, as you said, they're evaluating the complexity of problems by how difficult it is for them to solve the problems.

[276] And that's got very little to do with the complexities of solving them in AI.

[277] How do you think from the perspective of AI researcher, do we deal with the intuitions of the public?

[278] So in trying to think, arguably, the combination of hype investment and the public intuition is what led to the AI winters.

[279] I'm sure that same can be applied to tech or the intuition of the public leads to media hype, leads to companies investing in the tech, and then the tech doesn't make the company's money, and then there's a crash.

[280] is there a way to educate people sort of to fight the let's call it system one thinking in general no I think that's the simple answer and it's going to take a long time before the understanding of what those systems can do becomes you know part and becomes public knowledge And then, and the expectations, you know, there are several aspects that are going to be very complicated.

[281] And the fact that you have a device that cannot explain itself is a major, major difficulty.

[282] And we're already seeing that.

[283] I mean, this is really something that is happening.

[284] So it's happening in the judicial system.

[285] So you have a system that are clearly better at predicting parole violations than judges, but they can't explain their reasoning.

[286] And so people don't want to trust them.

[287] We seem to, in system one, even, use cues to make judgments about our environment.

[288] So this explainability point, do you think humans can explain stuff?

[289] No, but I mean, there is a very interesting aspect of that.

[290] Humans think they can explain themselves.

[291] So when you say something, and I ask you, why do you believe that, then reasons will occur to you and you will.

[292] But actually, my own belief is that in most of the, cases, the reasons have very little to do with why you believe what you believe, so that the reasons are a story that comes to your mind when you need to explain yourself.

[293] But people traffic in those explanations.

[294] I mean, the human interaction depends on those shared fictions and the stories that people tell themselves.

[295] You just made me actually realized, and we'll talk about stories in a second, not to be cynical about it, but perhaps there's a whole movement of people trying to do explainable AI.

[296] And really, we don't necessarily need to explain, AI does need to explain itself, it just needs to tell a convincing story.

[297] Yeah, absolutely.

[298] It doesn't necessarily need to reflect the truth as it just needs to be convincing.

[299] There's something to that.

[300] You can say exactly the same thing in a way that sounds cynical or doesn't sound cynical.

[301] But the objective of having an explanation is to tell a story that will be acceptable to people.

[302] And for it to be acceptable and to be robustly acceptable, it has to have some elements of truth, but the objective is for people to accept it.

[303] It's quite brilliant, actually.

[304] But so on the stories that we tell, sorry to ask you the question that most people know the answer to, but you talk about two selves in terms of how life has lived.

[305] The experience of and remembering self.

[306] Can you describe the distinction between the two?

[307] Well, sure.

[308] I mean, there is an aspect of life that occasionally, you know, most of the time we just live and we have experiences and they're better and they're worse and it goes on over time.

[309] And mostly we forget everything that happens or we forget most of what happens.

[310] Then occasionally, you, when something ends or a different point, points, you evaluate the past and you form a memory, and the memory is schematic.

[311] It's not that you can roll a film of an interaction.

[312] You construct, in effect, the elements of a story about an episode.

[313] So there is the experience, and there is the story that is created about the experience.

[314] And that's what I call the remembering.

[315] So I had the image of two selves.

[316] So there is a self that lives, and there is a self that evaluates life.

[317] Now, the paradox and the deep paradox in that is that we have one system or one self that does the living, but the other system, the remembering self, is all we get to keep.

[318] And basically, decision -making and everything that we do, is governed by our memories, not by what actually happened.

[319] It's governed by the story that we told ourselves or by the story that we're keeping.

[320] So that's the distinction.

[321] I mean, there's a lot of brilliant ideas about the pursuit of happiness that come out of that.

[322] What are the properties of happiness which emerge from a remember -the -self?

[323] There are properties of how we construct stories that are really important.

[324] So, I studied a few, but a couple are really very striking.

[325] And one is that in stories, time doesn't matter.

[326] There's a sequence of events, or there are highlights, and how long it took.

[327] You know, they lived happily ever after, and three years later, something.

[328] Time really doesn't matter.

[329] stories, events matter, but time doesn't.

[330] That leads to a very interesting set of problems because time is all we got to live.

[331] I mean, you know, time is the currency of life.

[332] And yet time is not represented basically in evaluative memories.

[333] So that creates a lot of paradoxes that I've thought about.

[334] Yeah, that are fascinating.

[335] But if If you were to give advice on how one lives a happy life based on such properties, what's the optimal?

[336] You know, I gave up, I abandoned happiness research because I couldn't solve that problem.

[337] I couldn't, I couldn't see.

[338] And in the first place, it's very clear that if you do talk in terms of those two cells, then that what makes the remembering self -happy and what makes experiencing self -happy are different things.

[339] And I asked the question of, suppose you're planning a vacation and you're just told that at the end of the vacation, you'll get an amnesic drug so you remember nothing, and they'll also destroy all your photos, so there'll be nothing.

[340] Would you still go to the same vacation?

[341] And it's, it turns out we go to vacations in large part to construct memories, not to have experiences, but to construct memories.

[342] And it turns out that the vacation that you would want for yourself, if you knew you will not remember, is probably not the same vacation that you will want for yourself if you will remember.

[343] So I have no solution to these problems, but clearly those are big issues.

[344] And you've talked about how many minutes or hours you spend about the vacation.

[345] It's an interesting way to think about it because that's how you really experience the vacation outside the being in it.

[346] But there's also a modern, I don't know if you think about this or interact with it.

[347] there's a modern way to magnify the remembering self, which is by posting on Instagram, on Twitter, on social networks.

[348] A lot of people live life for the picture that you take, that you post somewhere.

[349] And now thousands of people share in it potentially, potentially millions, and then you can relive it even much more than just those minutes.

[350] Do you think about that magnification much?

[351] You know, I'm too old for social networks.

[352] I've never seen Instagram, so I cannot really speak intelligently about those things.

[353] I'm just too old.

[354] But it's interesting to watch the exact effects you described.

[355] I think it will make a very big difference.

[356] I mean, and it will also make a difference, and that I don't know, whether it's clear that in some ways the devices that serve.

[357] of us supplant function.

[358] So you don't have to remember phone numbers.

[359] You really don't have to know facts.

[360] I mean, the number of conversations I'm involved with where somebody says, well, let's look it up.

[361] So in a way, it's made conversations, well, it means that it's much less important to know things.

[362] No, it used to be very important to know things.

[363] this is changing so the requirements of that that we have for ourselves and for other people are changing because of all those supports and because and I have no idea what Instagram does but it's well I'll tell you I knew I mean I wish I could just have the my remembering self could enjoy this conversation, but I'll get to enjoy it even more by watching it and then talking to others it'll be about 100 ,000 people as scary as to say, well, listen or watch this, right?

[364] It changes things.

[365] It changes the experience of the world.

[366] That you seek out experiences which could be shared in that way.

[367] And I haven't seen, it's the same effects that you described.

[368] And I don't think the psychology of that magnification has been described yet?

[369] Because it's a new world.

[370] You know, the sharing, there was a time when people read books.

[371] And you could assume that your friends had read the same books that you read.

[372] So there was.

[373] Kind of invisible sharing.

[374] There was a lot of sharing going on.

[375] And there was a lot of assumed common knowledge and, you know, that was built in.

[376] I mean, it was obvious that you had read the New York Times.

[377] It was obvious that you had read the reviews.

[378] I mean, so a lot was taken for granted that was shared.

[379] And, you know, when there were, when there were three television channels, it was obvious that you'd seen one of them probably the same.

[380] So sharing has always been there.

[381] It was always there.

[382] It was just different.

[383] At the risk of inviting mockery from you, let me say that I'm also a fan of Sartre and Camus and existentialist philosophers.

[384] And I'm joking, of course, about mockery, but from the perspective of the two selves, what do you think of the existentialist philosophy of life?

[385] So trying to really emphasize the experiencing self as the proper way to, or the best way to live life.

[386] I don't know enough philosophy to answer that, but it's not, you know, the emphasis on experience is also the emphasis in Buddhism.

[387] Right, that's right.

[388] So that's, you just have got to experience things and not to evaluate and not to pass judgment and not to score, not to keep score.

[389] So if when you look at the grand picture of experience, you think there's something to that that one of the ways to achieve contentment and maybe even happiness is letting go of any of the things, any of the procedures of the remembering self?

[390] Well, yeah, I mean, I think, you know, if one could imagine a life in which people don't score themselves, it feels as if that would be a better life, as if the self -scoring and, you know, how am I doing kind of question is not a very happy thing to have.

[391] but I got out of that field because I couldn't solve that problem.

[392] And that was because my intuition was that the experiencing self, that's reality.

[393] But then it turns out that what people want for themselves is not experiences.

[394] They want memories and they want a good story about their life.

[395] And so you cannot have a theory of happiness that doesn't correspond to what people want for themselves.

[396] And when I realized that this was where things were going, I really sort of left the field of research.

[397] Do you think there's something instructive about this emphasis of reliving memories in building AI systems?

[398] So currently, artificial intelligence systems are more like experiencing self.

[399] In that they react to the environment, there's some pattern formation like, learning and so on, but you really don't construct memories except in reinforcement learning every once in a while that you replay over and over.

[400] Yeah, but you know, that would, in principle, would not be...

[401] Do you think that's useful?

[402] Do you think it's a feature or a bug of human beings that we look back?

[403] Oh, I think that's definitely a feature.

[404] That's not a bug.

[405] I mean, you have to look back in order to look forward.

[406] So without looking back, you couldn't really intelligently look forward.

[407] You're looking for the echoes of the same kind of experience in order to predict what the future holds?

[408] Yeah.

[409] Though Victor Frankel, in his book, Man's Search for Meaning, I'm not sure if you've read, describes his experience at the concentration camps during World War II as a way to describe that finding, identifying a purpose in life, a positive purpose in life, can save one from suffering.

[410] First of all, do you connect with the philosophy that he describes there?

[411] Not really.

[412] So I can really see that somebody who has that feeling of purpose and meaning and so on, that that could sustain you.

[413] I, in general, don't have that feeling.

[414] And I'm pretty sure that if I were in a concentration camp, I'd give up and die.

[415] You know, so he talks, he is a survivor.

[416] Yeah.

[417] And, you know, he survived with that.

[418] And I'm not sure how essential to survival the sense is.

[419] But I do know when I think about myself that I would have given up.

[420] at oh this isn't going anywhere and there is a sort of character that manages to survive in conditions like that and then because they survive they tell stories and it sounds as if they survived because of what they were doing we have no idea they survived because of the kind of people that they are and they are the kind of people who survives and will tell themselves stories of a particular kind So I'm not...

[421] So you don't think seeking purpose is a significant driver in our behavior?

[422] I mean, it's a very interesting question because when you ask people whether it's very important to have meaning in their life, they say, oh yes, that's the most important thing.

[423] But when you ask people, what kind of a day did you have?

[424] And, you know, what were the experiences that you remember?

[425] You don't get much meaning.

[426] you get social experiences, then, and some people say that, for example, in child, you know, in taking care of children, the fact that they are your children and you're taking care of them makes a very big difference.

[427] I think that's entirely true, but it's more because of a story that we're telling ourselves, which is a very different story when we're taking care of our children or when we're taking care of other things.

[428] Jumping around a little bit, in doing a lot of experiments, let me ask a question.

[429] Most of the work I do, for example, is in the real world, but most of the clean good science that you can do is in the lab.

[430] So that distinction, do you think we can understand the fundamentals of human behavior through controlled experiments in the lab.

[431] If we talk about pupil diameter, for example, it's much easier to do when you can control lighting conditions.

[432] Yeah, of course.

[433] So when we look at driving, lighting variation destroys almost completely your ability to use pupil diameter.

[434] But in the lab, as I mentioned, semi -autonomous or autonomous vehicles, in driving simulators, we don't capture true, honest human behavior in that particular domain.

[435] So what's your intuition?

[436] How much of human behavior can we study in this controlled environment of the lab?

[437] A lot, but you'd have to verify it, you know, that your conclusions are basically limited to the situation, to the experimental situation.

[438] then you have to jump a big inductive leap to the real world and that's the flare that's where the difference I think between the good psychologist and others that are mediocre is in the sense that your experiment capture something that's important and something that's real and others are just running experiments so what is that like the birth of an idea to his development in your mind to something that leads to an experiment.

[439] Is that similar to maybe like what Einstein or a good physicist do is your intuition?

[440] You basically use your intuition to build up.

[441] Yeah, but I mean, you know, it's very skilled intuition.

[442] Right.

[443] I mean, I just had that experience, actually.

[444] I had an idea that turns out to be a very good idea a couple of days ago.

[445] And you have a sense of that building up.

[446] So I'm working with a collaborator.

[447] And he essentially was saying, you know, what are you doing?

[448] You know, what's going on?

[449] And I was really, I couldn't exactly explain it, but I knew this is going somewhere.

[450] But, you know, I've been around that game for a very long time.

[451] And so I can, you develop that anticipation that, yes, this is worth following.

[452] going up.

[453] There's something here.

[454] That's part of the skill.

[455] Is that something you can reduce to words in describing a process in the form of advice to others?

[456] No. Follow your heart essentially.

[457] I mean, you know, it's like trying to explain what it's like to drive.

[458] It's not, you've got to break it apart and it's not.

[459] And then you lose.

[460] And then you lose the experience.

[461] You mentioned collaboration.

[462] You've written about your collaboration.

[463] with Amos Tversky, that this is you writing, the 12 or 13 years in which most of our work was joint were years of interpersonal and intellectual bliss.

[464] Everything was interesting, almost everything was funny, and there was a current joy of seeing an idea take shape.

[465] So many times in those years we shared the magical experience of one of us saying something which the other one would understand more deeply than a speaker had done.

[466] contrary to the old laws of information theory it was common for us to find that more information was received than had been sent i have almost never had the experience with anyone else if you have not had it you don't know how marvelous collaboration can be so let me ask a perhaps a silly question how does one find and creates such a collaboration that may be asking like, how does one find love?

[467] Yeah, you have to be lucky.

[468] And I think you have to have the character for that because I've had many collaborations.

[469] I mean, none were as exciting as with Amos people.

[470] But I've had, and I'm having, just very, so it's a skill.

[471] I think I'm good at it.

[472] Not everybody is good at it.

[473] and then it's the luck of finding people who are also good at it.

[474] Is there advice in a form for a young scientist who also seeks to violate this law of information theory?

[475] I really think it's so much luck is involved.

[476] And in those really serious collaborations, at least in my experience, are a very personal experience.

[477] and I have to like the person I'm working with.

[478] Otherwise, you know, I mean, there is that kind of collaboration, which is like an exchange, a commercial exchange, I'm giving this, you give me that.

[479] But the real ones are interpersonal.

[480] They're between people who like each other and who like making each other think and who like the way that the other person responds to your thoughts you have to be lucky.

[481] Yeah, I mean, but I already noticed the past, even just me showing up here, you've quickly started to digging in a particular problem I'm working on and already new information started to emerge.

[482] Is that a process, just the process of curiosity of talking to people about problems and seeing?

[483] I'm curious about anything to do with AI and robotics, and so, and I knew, you were dealing with that.

[484] So I was curious.

[485] Just follow your curiosity.

[486] Jumping around on the psychology front, a dramatic sounding terminology of replication crisis, but really just the, at times, this effect at times studies do not, are not fully generalizable.

[487] They don't.

[488] You are being polite.

[489] It's worse than that.

[490] Is it?

[491] So I'm actually not fully familiar to the degree how bad it is, right?

[492] So what do you think is the source?

[493] Where do you think?

[494] I think I know what's going on, actually.

[495] I mean, I have a theory about what's going on.

[496] And what's going on is that there is, first of all, a very important distinction between two types of experiments.

[497] And one type is within subject.

[498] So it's the same person has two.

[499] experimental conditions.

[500] And the other type is between subjects, where some people are this condition, other people are that condition.

[501] They are different worlds.

[502] And between subject experiments are much harder to predict and much harder to anticipate.

[503] And the reason, and they're also more expensive because you need more people and it's just...

[504] So between subject experiments is, where the problem is.

[505] It's not so much in within -subject experiments, it's really between.

[506] And there is a very good reason why the intuitions of researchers about between -subject experiments are wrong.

[507] And that's because when you are a researcher, you're in a within -subject situation.

[508] That is, you are imagining the two conditions and you see the causality and you feel it, But in the between -subject condition, they live in one condition, and the other one is just nowhere.

[509] So our intuitions are very weak about between -subject experiments, and that, I think, is something that people haven't realized.

[510] And in addition, because of that, we have no idea about the power of manipulations, of experimental manipulations, because the same manipulation is much more powerful when you are in the two conditions than when you live in only one condition.

[511] And so the experimenters have very poor intuitions about between subject experiments.

[512] and there is something else, which is very important, I think, which is that almost all psychological hypotheses are true.

[513] That is, in the sense that, you know, directionally, if you have a hypothesis that A really causes B, that it's not true that A causes the opposite of B. Maybe A just has very little effect, But hypotheses are true mostly, except mostly they're very weak, they're much weaker than you think when you are having images of.

[514] So the reason I'm excited about that is that I recently heard about some friends of mine who they essentially funded 53 studies of behavioral treatment.

[515] change by 20 different teams of people with a very precise objective of changing the number of time that people go to the gym, but you know so.

[516] And the success rate was zero, not one of the 53 studies worked.

[517] Now, what's interesting about that is those are the best people in the field, and they have no idea what's going on.

[518] So they're not calibrated.

[519] They think that it's going to be powerful because they can imagine it.

[520] But actually, it's just weak because you're focusing on your manipulation and it feels powerful to you.

[521] There's a thing that I've written about that's called the focusing illusion.

[522] That is that when you think about something, it looks very important.

[523] More important than it really is.

[524] More important than it really is, but if you don't see that effect, the 53 studies, doesn't that mean you just report that?

[525] So what's, I guess, the solution to that?

[526] Well, I mean, the solution is for people to trust their intuitions less or to try out their intuitions before.

[527] I mean, experiments have to be pre -registered, and by the time you run an experiment, you have to be committed to.

[528] it and you have to run the experiment seriously enough and in a public and so this is happening and the interesting thing is what happens before and how do people prepare themselves and how they run pilot experiments it's going to train the way psychology is done and it's already happening do you have a hope for this might connect to the study sample sample size.

[529] Do you have a hope for the internet?

[530] Well, I mean, you know, this is really happening.

[531] M -Turk.

[532] Everybody is running experiments on M -Turk, and it's very cheap and very effective.

[533] Do you think that changes psychology, essentially?

[534] Because you can not run 10 ,000 subjects.

[535] Eventually, it will.

[536] I mean, I can't put my finger on how exactly, but it's, That's been true in psychology.

[537] Whenever an important new method came in, it changes the field.

[538] So an M -Turk is really a method because it makes it very much easier to do something, to do some things.

[539] Is there undergrad students who'll ask me how big an neural network should be for a particular problem?

[540] So let me ask you an equivalent question.

[541] How big, how many subjects that a study have for it to have a conclusive result in general?

[542] Well, it depends on the strength of the effect.

[543] So if you're studying visual perception or the perception of color, many of the classic results in visual, in color perception, we're done on three or four people, and I think one of them was colorblind, but partly colorblind.

[544] But on vision, you know, it's highly reliable.

[545] Many people don't need a lot of replications for some type of neurological experiment.

[546] When you're studying weaker phenomena, and especially when you're studying them between subjects, then you need a lot more subjects than people have been running.

[547] and that is that's one of the thing that are happening in psychology now is that the power is a statistical power of the experiments is increasing rapidly does the between subjects as the number of subjects goes to infinity approach well I mean you know it goes to infinity is exaggerated but people the standard number of subjects who are an experiment psychology with 30 or 40 and for a weak effect, that's simply not enough.

[548] And you may need a couple of hundred.

[549] I mean, it's that sort of order of magnitude.

[550] What are the major disagreements in theories and effects that you've observed throughout your career?

[551] That still stand today.

[552] We've worked on several fields.

[553] But what still is out there as a major disagreement that pops into your mind?

[554] I've had one extreme experience of controversy with somebody who really doesn't like the work that Amoski and I did and he's been after us for 30 years or more, at least.

[555] Do you want to talk about it?

[556] Well, I mean, his name is good, Gigranzer.

[557] He's a well -known German psychologist.

[558] And that's the one controversy I have, which I, it's been unpleasant and no, I don't particularly want to talk about it.

[559] But is there open questions, even in your own mind, every once in a while, you know, we talked about semi -autonomous vehicles.

[560] In my own mind, I see what the data says, but I also am constantly torn.

[561] Do you have things where your studies have found something, but you're also in.

[562] intellectually torn about what it means and there's maybe it has been, maybe disagreements within your own mind about particular things.

[563] I mean, it's, you know, one of the things that are interesting is how difficult it is for people to change their mind.

[564] Essentially, you know, once they are committed, people just don't change their mind about anything that matters.

[565] And that is surprisingly, but it's true about scientists.

[566] So the controversy that I described, you know, that's been going on like 30 years and it's never going to be resolved.

[567] And you build a system and you live within that system and other systems of ideas look foreign to you and there is very little contact and very little mutual influence.

[568] That happens a fair amount.

[569] Do you have a hopeful advice or message on that?

[570] Thinking about science, thinking about politics, thinking about things that have impact on this world, how can we change our mind?

[571] I think that, I mean, on things that matter, you know, which are political or religious, and people just don't change their mind.

[572] and by and large, and there's very little that you can do about it.

[573] What does happen is that if leaders change their minds.

[574] So, for example, the American public doesn't really believe in climate change, doesn't take it very seriously.

[575] But if some religious leaders decided this is a major threat to humanity, that would have a big effect.

[576] So that we have the opinions that we have, not because we know why we have them, but because we trust some people and we don't trust other people.

[577] And so it's much less about evidence than it is about stories.

[578] So the way, one way to change your mind isn't at the individual level is that the leaders in the communities you look up with the stories change and therefore your mind changes with them.

[579] So there's a guy named Alan Turing, came up with a Turing test.

[580] Yeah.

[581] What do you think is a good test of intelligence?

[582] Perhaps we're drifting in a topic that we're maybe philosophizing about, but what do you think is a good test for intelligence for an artificial intelligence system?

[583] Well, the standard definition of artificial general intelligence, that it can do anything that people can do and it can do them better.

[584] Yes.

[585] What we are seeing is that in many domains, you have domain -specific and, you know, devices or programs or software, and they beat people easily in specified way.

[586] What we are very far from is that generally ability, a general purpose intelligence.

[587] So we, in machine learning, people are approaching something more general.

[588] I mean, for Alpha Zero was much more general than AlphaGo.

[589] But it's still extraordinarily narrow and specific in what it can do.

[590] So we're quite far from something that can, in every domain, think like a human, makes a better.

[591] What aspect, so the touring task has been criticized, it's natural language conversation that is too simplistic.

[592] It's easy to quote unquote pass under constraints specified what aspect of conversation would impress you if you heard it?

[593] Is it humor?

[594] Is it what would impress the heck out of you if you saw it in conversation?

[595] Yeah, I mean certainly wit would be impressive and humor would be more impressive than just factual conversation, which I think is easy, and allusions would be interesting, and metaphors would be interesting, I mean, but new metaphors, not practiced metaphors.

[596] So there is a lot that would be sort of impressive that it's completely natural in conversation, but that you really wouldn't expect.

[597] Does the possibility of creating a human -level intelligence or superhuman -level intelligence system excite you, scare you?

[598] How does it make you feel?

[599] I find the whole thing fascinating, absolutely fascinating.

[600] So exciting.

[601] I think, and exciting.

[602] It's also terrifying, you know, But I'm not going to be around to see it.

[603] And so I'm curious about what is happening now.

[604] But I also know that predictions about it are silly.

[605] We really have no idea, but it will look like 30 years from now.

[606] No idea.

[607] Speaking of silly, bordering on the profound, they may ask the question of, in your view, What is the meaning of it all, the meaning of life?

[608] These descendant of great apes that we are, what drives us as a civilization, as a human being, as a force behind, everything that you've observed and studied?

[609] Is there any answer or is it all just a beautiful mess?

[610] There is no answer that I can understand.

[611] And I'm not, and I'm not actively.

[612] looking for one.

[613] Do you think an answer exists?

[614] No. There is no answer that we can understand.

[615] I'm not qualified to speak about what we cannot understand, but there is, I know, that we cannot understand reality, you know, and, I mean, there are a lot of things that we can do.

[616] I mean, you know, gravity waves, I mean, that's a big moment for humanity.

[617] And when you imagine that ape, you know, being able to go back to the Big Bang, that's, that's, but.

[618] The why is bigger than us.

[619] The why is hopeless, really.

[620] Danny, thank you so much.

[621] It was an honor.

[622] Thank you for speaking today.

[623] Thank you.

[624] Thanks for listening to this conversation.

[625] And thank you to our presenting sponsor, Cash App.

[626] Download it, use code Lex Podcast.

[627] You'll get $10.

[628] and $10 will go to first, a STEM education nonprofit that inspires hundreds of thousands of young minds to become future leaders and innovators.

[629] If you enjoy this podcast, subscribe on YouTube, give it five stars on Apple Podcasts, follow on Spotify, support it on Patreon, or simply connect with me on Twitter.

[630] And now, let me leave you with some words of wisdom from Daniel Kahneman.

[631] Intelligence is not only the ability to reason.

[632] It is also the ability to find relevant material and memory, and to deploy attention when needed.

[633] Thank you for listening, and hope to see you next time.