Leslie Valiant : The Importance of Being Educable

0:00:13 I’m Guy Kawasaki and this is Remarkable People.
0:00:16 We are on a mission to make you remarkable.
0:00:20 Helping me in this episode is Leslie Valiant.
0:00:25 He is a professor of computer science and applied mathematics at Harvard.
0:00:31 Valiant received a BA in mathematics from King’s College, a diploma of Imperial College
0:00:37 from Imperial College, that makes sense, and a PhD in computer science from the University
0:00:38 of Warwick.
0:00:45 Prior to joining Harvard in 1982, he held faculty positions at Carnegie Mellon University,
0:00:49 the University of Leeds, and the University of Edinburgh.
0:00:55 Valiant has made foundational, remarkable contributions to computer science, including
0:01:01 the theory of probably approximately correct learning, the concept of p-completeness in
0:01:06 complexity theory, and the bulk synchronous parallel processing model.
0:01:13 And if you believe I understand all of what I just said, you are in for a big surprise.
0:01:19 Valiant is the recipient of the Turing Award, the highest honor in computer science.
0:01:25 The award committee said this about him, “Rather does one see such a striking combination
0:01:29 of depth and breadth as in Valiant’s work.
0:01:34 He is truly a heroic figure in theoretical computer science and a role model for his
0:01:42 courage and creativity in addressing some of the deepest unsolved problems in science.”
0:01:48 Valiant’s book, The Importance of Being Educable, explores the unique ability of humans to absorb
0:01:50 and apply knowledge.
0:01:56 He argues that understanding our educability is crucial for safeguarding our future, especially
0:02:00 with the rise of artificial intelligence systems.
0:02:09 I’m Guy Kawasaki, this is Remarkable People, and now here is the remarkable Leslie Valiant.
0:02:12 What exactly is wrong with the term “intelligence”?
0:02:18 It is that we don’t know what it means, it’s got no definition.
0:02:24 No one can tell you exactly how to recognize an intelligent person, what’s the behavior
0:02:26 you’re supposed to look for.
0:02:31 One maybe trivial symptom of this, which may not be that important, but just as a symptom.
0:02:37 So there’s much discussion about the SAT test, which many American students do.
0:02:41 And I was surprised to find out recently what the letter A stands for.
0:02:46 So certainly, S stands for standard and T for test, but what does the A stand for?
0:02:52 So apparently, the A used to stand for aptitude, standard aptitude test.
0:02:56 They changed that, then it’s for standard assessment test.
0:03:01 They changed that, then it was SAT colon reasoning, they changed that.
0:03:05 So now officially, the A stands for nothing, it’s just a brand.
0:03:10 So whatever it tests for, the society is buying without any label on what it’s meant to do.
0:03:17 So if you buy a food in a store, it says more than that there’s a correlation between you
0:03:22 being less hungry afterwards and eating this, but there’s a correlation between these SAT
0:03:25 tests and something, and that’s all that’s promised.
0:03:30 So the problem with intelligence is that it’s widely used, but it’s got no definition.
0:03:31 Isn’t that fixable?
0:03:36 Can’t we get MacArthur Fellows and Turing Award winners and all you guys together say,
0:03:38 “Can you just please define intelligence?
0:03:40 How hard could that be?”
0:03:46 So in the book, I quote some reports of American Psychological Association some time ago where
0:03:51 they report on some other groups saying that they asked 24 experts in intelligence and
0:03:54 they all gave somewhat different answers.
0:04:00 So there’s some concepts for which we have a word, but we don’t really know what it
0:04:01 means.
0:04:06 Seeing as how you’re saying that there’s no sort of agreed upon consistent definition
0:04:13 of the word, how are we seemingly freely talking about AI so much right now?
0:04:18 How can we talk about artificial intelligence if we don’t even know what intelligence is?
0:04:19 Exactly.
0:04:25 So in fact, I think using word intelligence didn’t help the AI field to progress because
0:04:26 there was no goal there.
0:04:32 In my view, it progressed because AI adopted a more specialized definition or rather used
0:04:33 one.
0:04:34 And this is learning.
0:04:38 So learning is something which you can define much better.
0:04:41 So learning from examples, after you’ve seen some examples, you can classify future examples
0:04:43 better than you could before.
0:04:47 So learning from examples is something that’s a very specific ability.
0:04:49 It can be defined.
0:04:53 The computer programs are doing quite well and in practice these large language models
0:04:54 do exactly that.
0:04:58 Okay, so there’s some evidence that they do something interesting.
0:05:04 So AI, in my view, has progressed by trying to realize by computer tasks for which there
0:05:05 is a definition.
0:05:10 In a sense, what you’re saying is that intelligence is such a quagmire that you’re just going
0:05:13 to go outside the box and you’re going to create this new standard.
0:05:18 I’m not saying I want to create a new standard or that’s not how I start.
0:05:23 I’m saying how I phrase it in the book is that in some sense, it’s a very old question,
0:05:29 of course, is what’s the intellectual capability which humans have, which other species on
0:05:34 Earth do not, and the capability which enabled us to create our complex civilization, science
0:05:36 and everything else.
0:05:41 And although this may seem overly ambitious question, in some sense, it’s not because
0:05:44 in the evolutionary terms, we evolved pretty fast.
0:05:47 In the short term, this must have evolved.
0:05:52 So I’m trying to write down this capability.
0:05:58 And the first part of this is exactly what I was describing with learning from examples.
0:06:03 So it’s something which we understand well in computers.
0:06:07 Everyone sees how it works with these large language models in the last year or so.
0:06:09 And then I add a few things on.
0:06:12 So I think I’m doing a natural science.
0:06:15 I’m trying to explain human capabilities.
0:06:20 And I think something real, which has a definition, whether it turns out to be useful, that remains
0:06:21 to be seen.
0:06:25 In some sense, it is a replacement for intelligence, but in a special way.
0:06:32 So I’m curious that in the creation of your book and the title and really the writing
0:06:39 of the entire thing, did you ever think about the difference between “educatable” versus
0:06:40 “educable”?
0:06:44 Did you ever think of the nuances between those two words or now in modern use, their
0:06:46 equivalence, or it doesn’t matter?
0:06:51 I’m not quite sure how you mean any of those.
0:06:56 So at one point, these terms were used in the sense of yes/no.
0:07:01 So with children, could be educated in a regular school as opposed to having to be sent out.
0:07:04 So I think that’s the use these terms had maybe 50 years ago.
0:07:07 And maybe that’s the way in which you mean “educatable”, right?
0:07:12 But what I’m trying to capture is to define more ground up, the capability which I think
0:07:17 all of us have to some extent, and I’m not trying to distinguish with us some of us have
0:07:18 more of it than others.
0:07:22 Just generally, we’re trying to distinguish us from other species.
0:07:26 So it’s not to be confused with the other senses in which this word may have been used
0:07:27 in the past.
0:07:29 Just like its new technical term, if you like.
0:07:30 Okay.
0:07:37 Now, from your point of view, is this level of “educableness” if that’s the word?
0:07:45 Is it a person’s or a machine’s innate ability or is it also subject to the methods and resources
0:07:49 and environments around that person or is it just the person?
0:07:54 So the capability itself, some of it there at the beginning, the question is whether
0:07:58 it can be enhanced by some sort of life experience.
0:07:59 I don’t know.
0:08:00 Okay.
0:08:02 So I think that’s a good subject for investigation.
0:08:05 Certainly if you can enhance the rate at which you can learn, for example, that would be
0:08:07 fantastic for society.
0:08:10 But I think that’s a subject for research.
0:08:15 And I should add that most capabilities can be enhanced by some sort of training and the
0:08:16 right environment.
0:08:22 So the default assumption has to be that this also can benefit from suitable interventions.
0:08:23 But I don’t know.
0:08:28 You’re saying theoretically, if we could do this controlled experiment and we took you
0:08:33 or we took Stephen Wolf from or we took Neil deGrasse Tyson or I don’t know, Freeman Dyson
0:08:38 or something and we stuck you in the middle of the Amazon jungle with no internet, no
0:08:40 books, no nothing.
0:08:42 Clearly you’re still the same person.
0:08:44 If we go back 20 years later, are you running the jungle?
0:08:46 Are you the leaders there?
0:08:52 How much did where you are in modern society play into the fact that you’re such overachievers?
0:08:55 Well, well, first, I’m not equating.
0:08:59 So let’s suppose that you can measure adequability.
0:09:01 Different people have different levels of adequability.
0:09:05 Let us assume that I’m not even certain of that, but let’s assume that what it correlates
0:09:08 with in society again, I don’t know.
0:09:10 It’s not obvious.
0:09:16 The most educable people may not be the ones who are most successful in politics or academia
0:09:17 or whatever.
0:09:18 I’m not sure.
0:09:24 I don’t want to speculate, but it would be if you put someone in the jungle, then adequability
0:09:30 would cause one to how fast they can learn what’s going on there relative to what they
0:09:31 know.
0:09:34 Obviously, it may be that they know so little that’s relevant to how to survive there that
0:09:37 the adequability isn’t enough to survive.
0:09:44 But it’s a measure of how fast you can learn and understand what’s going on.
0:09:50 If I had to put my money on you or Stephen Wolfram to figure out Kurare before the random
0:09:53 person in the jungle, I think that’s a pretty good bet.
0:09:57 So I don’t want to speculate, but certainly there are some people, it’s maybe the entrepreneurs
0:10:02 who are very good at picking up lots of information, sitting through lots of information, understanding
0:10:03 what’s going on fast.
0:10:08 So I’m not quite sure which profession exhibit educability the most.
0:10:13 For example, scientists obviously need it, but they may not be the most extreme because
0:10:17 they’re also good at concentrating for a long time to pursue a direction.
0:10:23 Educability may be more where you keep finding out new things all the time and relating everything
0:10:24 to each other and running with it.
0:10:46 Are you familiar with the work of Carol Dweck and the growth mindset?
0:10:52 She’s a professor at Stanford and she in the early 2000s pioneered this dichotomy between
0:10:55 the growth mindset and the fixed mindset.
0:10:59 The fixed mindset basically says that you think you are what you are.
0:11:00 You cannot be anymore.
0:11:03 You cannot learn new skills.
0:11:06 It also means you think you cannot deter your rates.
0:11:10 So if you’re a genius, you believe you’ll always be a genius and Carol Dweck’s theory
0:11:15 is that the growth mindset means you can learn things, you can do things and it seems to
0:11:23 me there’s a great deal of parallel between your theory of educable and the growth mindset.
0:11:27 So I thought that maybe great minds thought alike and you knew each other.
0:11:33 No, but I mean that theory may be more about one’s attitude, but I think this educability
0:11:35 is something which kind of everyone has.
0:11:40 So the fact that most people are gathering information all the time, they’re watching
0:11:44 movies for example, they’re reading novels, they’re looking at their cell phones, they’re
0:11:46 all soaking in information.
0:11:50 Much of the information is of no relevance to them, it’s got no benefit to them, but
0:11:51 that’s what they do.
0:11:52 They soak in information.
0:11:53 Okay.
0:11:57 So I think it’s this distinction whether you regard your ability to soak in information
0:12:00 or whether you’re going to do it usefully to yourself, to benefit yourself, to find
0:12:01 a new profession.
0:12:03 I think that’s not exactly the same thing.
0:12:04 Okay.
0:12:09 I think what I’m describing is something which we freely all have and I quite generously
0:12:10 I think.
0:12:18 In your book you mentioned this example of chimps escaping from their cage in a zoo using
0:12:24 a fallen tree to jump over the wall and then the other chimps watched that and learned
0:12:27 and also jumped over the wall.
0:12:33 And so are you saying then that in this case these animals are educable?
0:12:36 Is that the conclusion to draw that they pass the test for educable?
0:12:43 No, I’m saying that they look pretty smart and they can solve problems and they can obviously
0:12:48 the first one to jump over the wall figured out what to do assuming that hadn’t seen something
0:12:50 like that being done and the rest could copy.
0:12:55 But so by educable, described in great length in the book, so it’s a combination of learning
0:13:01 from experience, being able to train together what you’ve learned and some kind of reasoning.
0:13:06 And the third part is being able to learn from someone else, describing to explicitly
0:13:08 what should be done.
0:13:12 So in some sense, copying behavior is like that, but in humans it’s much more general
0:13:13 that we sit in.
0:13:18 If you go to college, you sit in a lecture room and someone tells you the laws of quantum
0:13:23 mechanics and they sit there for 30 lectures and at the end you can do a lot of stuff.
0:13:28 So I think it’s only humans who have this ability to be able to soak in a lot of information
0:13:29 given explicitly.
0:13:34 So it’s like other people having had the experience, other people having done the experiments,
0:13:38 but they can tell you the conclusions and you can internalize it yourself and use it
0:13:41 as if you had the experience yourself.
0:13:48 Have you made observations yet about the presence of educableness in life?
0:13:50 Is it normally distributed?
0:13:53 Is it distributed differently to genders?
0:13:59 Does your educableness, does it grow or decay chronologically through life?
0:14:04 Have we had some longitudinal studies and cross-cultural studies, cross-gender studies
0:14:06 to understand more about this concept?
0:14:09 No, we haven’t had any studies.
0:14:12 It’s a new concept, we haven’t had any studies.
0:14:16 So to answer any of the questions, you’d have to develop some test which you possibly think
0:14:22 does measure educability and these tests would have to be of the nature that you’re…
0:14:27 It’s like if you do a one-hour test, in principle you could have a test which is like a one semester
0:14:31 course or you could have a one-hour test, but what you’d be measuring is how much you’ve
0:14:34 learned during that one hour.
0:14:39 So you’re given some questions where it wouldn’t help you to have previous knowledge to answer
0:14:40 the questions.
0:14:46 So it’s related to current IQ tests, but it would have a different emphasis.
0:14:52 It’s a new kind of question, but the only related point I’d make is that if one accepts
0:14:57 this notion that what Kersh has assumed is his educability, which means an extreme ability
0:15:03 to learn and soak up information, then this also illustrates why it’s very difficult to
0:15:10 answer the questions you ask, I mean to try to prove differences between groups in abilities
0:15:15 and other abilities is difficult just because if you make some measurement of the performance
0:15:19 of a group today, somewhere else at a different time, the outcome may be different because
0:15:23 the group may have different experiences.
0:15:29 So we are being so subject, so extremely subject to outside influences throughout educability
0:15:35 that we’re the least promising objects of scientific study, if you like.
0:15:39 By doing surveys on humans in different places, you shouldn’t be surprised that these results
0:15:45 are transferable, generalizable from different places, different times, just because we’re
0:15:49 so prone to change the point of educability that we’re so prone to change, so prone to
0:15:55 be influenced by our environment, it’s very hard to answer any of the questions you ask.
0:16:01 If you think about how would you possibly control all the variables in an experiment
0:16:02 like that?
0:16:03 Right?
0:16:07 Although, I gotta tell you, if I had to guess, I would tell you that women are more educable
0:16:08 than men.
0:16:12 There’s no doubt in my mind, but I don’t have any objective proof for that.
0:16:14 Just my life experiences.
0:16:20 So if we’re at such like the starting point of all of this, let’s say somebody’s listening
0:16:24 to this and saying, “Yeah, you know what, really, I want myself, I want my kids, I want
0:16:26 my company to be more educable.
0:16:28 What can I do now?
0:16:33 Give me some tactical and practical stuff, Leslie, help me out here.”
0:16:36 I don’t think I can, I’m sorry.
0:16:44 I think the only rational thing is to do research and develop some tests which possibly
0:16:46 measure this kind of thing.
0:16:49 If we can’t measure it, then there’s not much we can say about it.
0:16:54 So I think it’s a long-term project, the question of whether we can improve our educability.
0:16:55 I think it’s a long-term project.
0:17:01 Of course, many people have ideas about learning to learn and teaching to learn, but to validate
0:17:07 any of these things is difficult without the measure of when you can declare yourself to
0:17:10 be successful in having an enhanced learning.
0:17:17 If you were to just take a social welfare perspective and step back, it’s hard to imagine
0:17:22 more things that could be more important than this.
0:17:28 This could be the key to the survival of mankind to figure out how to make people more educable.
0:17:31 Personally, I agree.
0:17:32 I do.
0:17:37 So I think that, yeah, I think exactly, I think it’s a public issue for discussion.
0:17:40 Listen, let me give you some shallow thinking.
0:17:44 Let me see if this metaphor will work for you.
0:17:52 Is it fair to say that intelligence, even though we can’t really define it, intelligence
0:17:59 is like the chip speed and educableness is programmability.
0:18:00 Is that a fair statement?
0:18:01 No.
0:18:02 I don’t think so.
0:18:10 I think educability is, if supposing, let’s talk about a computer rather than a human.
0:18:15 Eucability is like a discussion of what the capabilities are of this thing.
0:18:17 And intelligence, I don’t know what that is.
0:18:19 I really don’t know what intelligence is.
0:18:25 Some people say intelligence is what a standard intelligence test measures, which is a bit
0:18:26 circular.
0:18:29 But again, I don’t quite know what that is.
0:18:31 Any cognitive capabilities are correlated.
0:18:35 If you’re good at one thing, often you’re good at something else, often a weak correlation,
0:18:37 strong correlation.
0:18:41 So these intelligence tests are just something, maybe arbitrary, which is correlated with
0:18:45 many things, which colleges may want to find this out because it may be correlated with
0:18:47 how our students do.
0:18:49 But many other tests would also.
0:18:51 So I really don’t know what intelligence is.
0:18:59 Personally, I’m really, so I don’t think I go around and say, oh, this person, my neighbor’s
0:19:00 intelligent.
0:19:06 I don’t think I would necessarily say that orthogonal to this is, is there no concept
0:19:11 that we need to take in account something like morality, right?
0:19:16 So that morality, you could be the most educable person in the world.
0:19:20 Not that I believe this, but let’s just suppose for, for a second that we think Donald Trump
0:19:22 is educable.
0:19:24 But I would say he has zero morality, right?
0:19:29 You, you, you could take an educable person who could learn from others and extrapolate
0:19:31 and all do all these great things.
0:19:33 But what if that person is fundamentally evil?
0:19:34 Then what?
0:19:36 Then, then it’s unfortunate.
0:19:44 But certainly, educability only defines exactly, as they say, someone’s basic cognitive capabilities.
0:19:49 And it’s about how facing the book that it’s the not know that I talk about beliefs that
0:19:53 you learn beliefs of the people’s beliefs that tell you their beliefs.
0:19:57 And so in that sense, the theory is about capabilities.
0:19:59 So in one sense, it’s morally totally neutral.
0:20:03 It doesn’t discuss morality because, as you say, the beliefs could be good or could be
0:20:04 bad.
0:20:10 But what one’s reaction to this should be, it shouldn’t be moral neutrality.
0:20:15 As you say, some people have different beliefs and some we think are evil and some we think
0:20:16 are good.
0:20:21 And just because our capabilities are neutral to morality, it doesn’t mean that we should
0:20:22 be.
0:20:25 We should still certainly fight for what you believe is right and against things which
0:20:28 you believe are evil.
0:20:34 But the fact that because our basic capability is kind of neutral on morals, totally bad
0:20:36 beliefs can spread through society.
0:20:39 We don’t seem to have any good defense against that.
0:20:45 If people are listening to this podcast and they’re struggling with this educable concept,
0:20:52 do you have any people that might be well known in the the valiant hall of fame of
0:20:58 educableness that you can say, oh, I understand that now he cites this person as educable.
0:21:02 Is there anybody like that you can say, think of him or think of her when you think about
0:21:03 educability.
0:21:07 You mean someone who shows educability in the extreme?
0:21:08 Yeah.
0:21:09 Who’s your hero?
0:21:11 Who’s in your hall of fame of educableness?
0:21:17 I don’t know, but I sincerely believe that this is something which we all have.
0:21:20 I’m not trying to define something which will separate us.
0:21:24 I’m really trying to define something which we all have and which unifies us.
0:21:27 So I think the educability is very important, otherwise I wouldn’t have spent so much time
0:21:28 on it.
0:21:33 But I don’t know, as we said before, what this correlates with exactly.
0:21:39 What human formative correlates best with, what kind of people exhibit it the most.
0:21:40 Yeah.
0:21:43 So at this point, I just don’t have a good answer to your question.
0:21:47 It’s a natural question, but it’s not the direction in which I’m looking.
0:21:48 Okay.
0:21:53 If we can shift gears slightly towards artificial intelligence, I know I just used the I word,
0:21:55 but that’s what everybody use, right?
0:22:02 At this point, when people say artificial intelligence, do you think it means that it
0:22:09 refers to what a machine can do like a human or what a machine cannot do like a human?
0:22:11 Like what is artificial anymore?
0:22:16 I think the meaning of the term has changed in history, but at the moment in the media,
0:22:21 it clearly means the kind of things which current machine systems can do.
0:22:22 Okay.
0:22:27 So by AI people seem to mean large language models and things for which you can download
0:22:30 some software with which they can do.
0:22:35 So it’s in the area of machine capabilities and manipulating language pictures.
0:22:37 So I think that’s what it means now.
0:22:43 It’s artificial in the sense that it’s a machine doing what humans can do and that’s what makes
0:22:45 it artificial.
0:22:46 Yeah.
0:22:49 The artificial part always means that it’s doing it.
0:22:53 Now, in your opinion, is chat GPT educable?
0:22:59 No, no, basically chat GPT, just the one of the three requirements of educability, which
0:23:01 is learning from examples.
0:23:04 So it’s trained to predict the next syllable of texts.
0:23:06 That’s what it’s trained for doing, basically.
0:23:12 They predict syllable by syllable texts from being given billions of examples.
0:23:19 Now, I understand at a technical standpoint, but from the outside looking in, if you just
0:23:23 give it a series of prompts, I don’t think most people would look at it like, oh, here
0:23:26 comes syllable after syllable very rapidly.
0:23:29 It looks very cogent and salient to me.
0:23:31 What’s going on there?
0:23:32 How does that thing work then?
0:23:39 It’s trained on a very large number of sentences, so the next syllable will be very likely to
0:23:43 come from some sentence or some phrase which has been uttered before many times.
0:23:49 And also, it works on very large windows of texts, so it predicts the next thing on many
0:23:51 hundreds of characters.
0:23:56 So it does give the impression of some sort of stream of consciousness, as if it can remember
0:24:01 what it’s talking about for a while, because it’s got this very large bit of text from
0:24:03 which it’s predicting.
0:24:10 But it uses the fact that it’s got vast numbers of sentences stored, which it can use.
0:24:17 And so some of the mystery is that with machine learning in general, intuitively, what happens
0:24:21 if you’ve got billions of examples is something which is really counterintuitive.
0:24:26 So if you train on these numbers, people never looked at before, then the phenomena are almost
0:24:28 different in kind.
0:24:34 It’s okay being impressed, I’m just impressed, so the smoothness of the sentence is amazing,
0:24:39 but I don’t think one should assume that these things are providing you more than what it’s
0:24:43 doing, predicting the next syllable, so certainly you shouldn’t take it to advice, for example,
0:24:46 for some important decision you have to make.
0:24:53 If you were to take a Turing test orientation, and now, okay, you’re going to have to correct
0:24:57 me if I get this wrong, but the Turing test is if you’re interacting with this thing and
0:25:00 you don’t know if it’s a human or a machine, but you can’t tell the difference between
0:25:03 a human and a machine.
0:25:04 Isn’t it a human?
0:25:05 Yeah.
0:25:06 Okay.
0:25:11 So Turing wrote this paper in 1950, we discussed the word thinking and also intelligence and
0:25:16 basically said that, yes, if you can’t tell the difference between a human and a machine,
0:25:19 then you can’t say that this human is thinking and the machine isn’t thinking.
0:25:22 So this is like almost a definition of thinking.
0:25:26 So in some sense it’s the opposite of what I’m trying to do.
0:25:30 I’m saying that you should define what thinking is and what intelligence is.
0:25:32 So Turing didn’t do this, he said, it is what it is.
0:25:36 If it looks like thinking, it’s thinking.
0:25:42 So Turing test is where the machine can effectively impersonate a person, but again, I don’t know
0:25:44 how far that takes us.
0:25:48 So for example, if you look at these large language models, in some sense it looks like
0:25:55 a human, but so the Turing thing is that you do some 20 questions with it, you ask questions
0:26:01 and see what the answers and for example, you’ll find that large language models don’t
0:26:04 know about yesterday’s news because it was trained a long time ago.
0:26:05 Okay.
0:26:08 So clearly it doesn’t pass the Turing test technically.
0:26:09 Okay.
0:26:15 But the point is that the Turing test philosophically is of course was very important, it’s intrigued
0:26:17 people for decades.
0:26:21 But by itself it doesn’t tell you what to do to make an intelligent machine.
0:26:25 It just says that you shouldn’t quibble philosophically.
0:26:35 So the fact that chat GPT is not educable, does that make you more comfortable with everybody
0:26:38 jumping on large language models or does it scare you more?
0:26:43 Is it a good thing or a bad thing that it isn’t educable at this point?
0:26:49 The only way of saying it is that with a completely at all levels, there are some choices to be
0:26:54 made and it’s not clear what the best choice is, but with large language models, with just
0:27:00 learning from examples, as we all know now, we always did, the training set is all important.
0:27:05 For example, these large language models can have a political orientation, depending on
0:27:09 what text is being trained on, they could have a bias just depends what is trained on.
0:27:13 Even at this level, it’s not clear what you should be doing.
0:27:17 I mean, depending on your political orientation, you may want to train in a different way.
0:27:21 Already we’re disagreeing on what the ideal large language model is.
0:27:27 If you make things educable, then the problems pile up because if you just tell your system
0:27:31 beliefs, then it’s an important question of what beliefs you give it.
0:27:36 So maybe telling it evil beliefs is just such a good idea.
0:27:40 So although we’re wonderful, we think and are educable, there are all kinds of problems
0:27:43 with reducing our educability.
0:27:46 No one knows what the best method of education is.
0:27:49 No one knows what the best beliefs are to instill in people.
0:27:52 No one knows what right beliefs are and wrong beliefs are.
0:27:57 So if we made machines with the same capabilities as humans, it wouldn’t solve anything because
0:28:01 we don’t quite know what the best thing would be is to educate humans.
0:28:02 Okay.
0:28:06 So if you have an educational machine, you have to educate them and make choices to be
0:28:07 made in education.
0:28:35 What do you think poses a greater existential threat to the survival of mankind, mankind
0:28:38 or artificial intelligence?
0:28:39 Mankind.
0:28:47 So the dangers of misusing artificial intelligence or having accidents or things like that are
0:28:52 obviously present as there are with any other dangerous technology, chemistry, nuclear weapons,
0:28:53 whatever.
0:28:58 But I think the extreme viewpoint that somehow machines will take over because they become
0:29:04 so intelligent that they’ll control the world, I think that’s a misplaced fear.
0:29:08 So certainly we have to distinguish control from other capabilities.
0:29:12 So we certainly shouldn’t give control to a machine if you’re not sure what the machine
0:29:13 is going to do to us.
0:29:14 That’s absurd.
0:29:15 Okay.
0:29:17 So we try to not give it control.
0:29:21 But if we don’t give it control, then what it’ll do will be things I think which we kind
0:29:24 of understand what’s going on, bit like with chemistry.
0:29:27 We understand some chemistry, not all of chemistry.
0:29:33 So we have to treat it as any other powerful technology, but I don’t think by its nature
0:29:35 it’s different from other technologies.
0:29:39 So I don’t think some monster is going to emerge from some machine somewhere, which
0:29:40 we don’t understand.
0:29:42 I’m not worried about that.
0:29:43 Okay.
0:29:49 So is it a fair summary to say that we’re at the starting line of educableness and we
0:29:53 are not sure how to foster it, how to grow it and all this.
0:29:58 And so there’s so much research to be done, but it offers great potential to make the
0:30:02 world a better place if we can just figure out how we can get people to learn more from
0:30:05 each other, learn more from what happened before.
0:30:08 And is that kind of the picture you’re trying to paint?
0:30:09 Yes.
0:30:11 That’s a very nice, generous description.
0:30:12 Yes.
0:30:14 I think that’s a very nice way of putting it.
0:30:15 Yes.
0:30:19 And in a sense, I’m asking the same question twice, but if I’m listening to this and I’m
0:30:25 buying into this, just like people listen to Carol Dweck and buy into the growth mindset.
0:30:27 So now what can I do?
0:30:32 I’m looking for some tips that I can use for my teenage son to make it more educable.
0:30:35 I’m not sure whether I can give a tip.
0:30:39 I think possibly being aware of this dimension.
0:30:44 So being aware to the extent to which we’re soaking up information, we easily absorb ideas
0:30:46 which we hear.
0:30:50 So one of the points I’m making in the book is that our educability is very strong.
0:30:55 We soak up information, we can relate information, use information, but evaluating information
0:30:56 is much more difficult.
0:31:03 So if we hear a theory, it’s evaluating whether it’s true or not, isn’t part of our basic
0:31:08 cognitive capabilities, and probably because it’s impossible if you hear some report of
0:31:12 something happening on the other side of the world, that you can’t go and look at it.
0:31:14 You have to see whom you believe.
0:31:20 So I think appreciating that we are so much subject to what we hear and see, and we can
0:31:22 soak it up and agree with it maybe.
0:31:27 So one parameter which is important to be psychologists look at is, that’s what we said
0:31:31 before, is that you hear all these theories, but which ones do we believe?
0:31:33 So what’s our criteria and for whom to believe?
0:31:38 So if we can, in turn, like all these theories, we have to choose which ones too because they
0:31:39 may be contradictory.
0:31:42 So what are our criteria?
0:31:47 And so psychologists, of course, look at this, that we prefer theories which agree with what
0:31:51 we already believe, theories which agree with our friends, et cetera.
0:31:56 So this educability, I think, is also a weakness, it shows us all the time we’re prone to some
0:32:01 new idea which we believe, but we shouldn’t really, we may be tricked into it, it’s our
0:32:08 weakness to, we have some policy for agreeing with things which already have a sympathy,
0:32:09 but maybe wrong.
0:32:14 So I think it’s a viewpoint which is new to me, just showing how vulnerable we are to
0:32:17 the ideas which are swirling around us.
0:32:23 So I think being aware of this educability notion may help people in understanding what’s
0:32:24 going on.
0:32:28 Listen, fundamentally, I am a marketing person.
0:32:33 So I was chief evangelist of Apple, chief evangelist of an online graphics design service
0:32:34 called Canva.
0:32:38 So I’m all about sales and marketing and evangelism.
0:32:45 And I’m telling you from the outside looking in, you are holding in your hand a golden
0:32:47 opportunity to change the world.
0:32:54 If you could just get people off this kind of SAT IQ test, Mensa memberships and all
0:32:59 that and just point out to them that it’s not about this score, it’s about how you can
0:33:02 adapt and learn and coexist with other people.
0:33:09 That’s much more important than your GPA and your SAT and I’m serious, I think you may
0:33:14 be holding the future of humanity in your hand here, Leslie.
0:33:16 You got a big responsibility here.
0:33:17 Thank you for the comments.
0:33:22 I think there are many ways this can go and I certainly need the help of lots of people
0:33:25 to actually go in any of those ways.
0:33:33 But yeah, so I think there’s a new idea here which is very general relevance to us, I think.
0:33:38 There’s no doubt in my mind that Leslie is onto something with this concept of “educable”.
0:33:43 I think it is just the flip side of Carol Dweck’s growth mindset.
0:33:47 And you know how much I love Carol Dweck’s growth mindset.
0:33:53 Carol Dweck and Leslie Valiant, that would be a remarkable combination.
0:33:58 Maybe I’ll send this episode to Carol and see if she’s interested in meeting Leslie.
0:34:00 The world would shake if this happened.
0:34:05 Anyway, I’m Guy Kawasaki, this is Remarkable People.
0:34:10 Once again, I’m going to remind you, please read our new book, Think Remarkable.
0:34:16 It will help you make a difference and change the world and be remarkable.
0:34:22 Now speaking of “educable”, we have a particularly educable group of people on the Think Remarkable
0:34:23 team.
0:34:30 They are, of course, Madison Nizmer, producer, Tessa Nizmer, researcher, Jeff See and Shannon
0:34:39 Hernandez on sound design, and our best buddies, Alexis Nishimura, now at Santa Clara University,
0:34:44 Louise Shortboard, Magana, and finally, Fallen Yates.
0:34:46 This is the Remarkable People team.
0:34:49 We’re on a mission to make you remarkable.
0:35:00 Until next time, mahalo and aloha.