Teach fundamentals, but emphasize students' own questions

Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

1. 26:25

   what future could my kids possibly be walking into and what kind of knowledge or ways of learning are actually valuable in the future that you imagine so I wonder I'm curious like any takes that you have there what do you expect will be the most useful skills for the future well yeah i'm asking myself that every day right i have i have a 13 year old and a nine year old okay they both you know really like math and science um you know and and uh sort of in in in the old world you know i would have strongly encouraged them to to to go into that right and now i feel like there is so much uncertainty like we could be headed toward this incredibly good world where we can all just pursue whatever we're the most passionate about with no need to worry about you know uh uh supporting ourselves economically right uh uh like the uh luxury space communism as it's called we could

2. 27:23

   be headed for this terrible world where uh um you know where we're sort of ai has disempowered us right? We could be headed for all kinds of things in between. You know, I still encourage my kids to study math and science just because they like it, you know, and, you know, as long as someone enjoys it, I mean, I mean, look, you know, my, my son also spends a lot of time, you know, playing chess, you know, he goes to chess tournaments, okay, humans had no hope against computers at chess already 30 years ago, right? And yet, you know, some people I remember I don't remember Kasparov versus Deep Blue. I remember people predicting that this is going to be the death of chess. And that's not at all what happened. You know, chess has never been more popular than it is now. You know, you've got chess streamers, chess celebrities, right? And, you know, people just decided, you know, this is either a craft or a sport, you know, that they enjoy. And they just continued it as a human practice.

3. 28:29

   so so i think you know certainly math and science can continue at least at that level as a thing that we enjoy doing you know for for i mean for as long as we're we're here at all to to choose to do things you know some people are going to choose to do that and they're going to enjoy that but you know there's also there's a further argument that one can make which is right now we absolutely still do need humans who know how to think okay uh i mean you know and and one proof of that is that we don't yet know how to align ai reliably with human values right we don't you know we're not yet on you know on a course to this you know utopian world with the you know an ai that will you know uh will we are confident will fulfill our wishes and as long as that's not you You know, as long as that's still the case, then I feel like, you know, we need humans at the steering wheel.

4. 29:28

   We need humans, you know, who know how to think. I mean, most obviously, we need humans directly working on the alignment problem or, you know, using whatever skills they have to, you know, help make sure that AI will go well. But more broadly than that, you know, we need people who know math, who know physics, who know programming, who know how to write, who know sort of all the things that humans have, you know, struggled to learn.

5. 29:55

   I mean, you know, look at what has happened, you know, just last week, right? If we didn't, you know, there would be no hope of even vetting these open AI solutions to all these math problems if we didn't have a community of thousands of experts who really know these problems, right? So as long as we are not in this, let's say, post-economic or post-scarcity future world, then we need people who know how to think.

6. 30:32

   And, you know, I think at, you know, universities, you know, like, you know, we have a lot of problems, but we also do know something about teaching people how to think. We, you know, and teaching people how to think, you know, sometimes means, you know, getting them to exercise their brains, even to do problems that, you know, computers have long been able to do better, right? even before ai i mean you know like all the math that someone might do you know up through elementary school maybe even middle school high school you know with stuff that you could put into a calculator you could put into a computer algebra package right and so you could have you know even before ai you could have asked what is the point of any of that right well the point is uh uh you know you you you need to exercise your brain and you know in this way right like you

7. 31:29

   We wouldn't say, oh, you know, you don't need to go running because you can just drive your car around the track, right? That's not going to develop your muscles in the right way, right? So, you know, we could say, yes, you know, people need to learn how to do math, how to program, how to do physics, because that is how we develop their brains to think systematically about anything. Which is a thing that is still needed. So that's kind of, you know, what I tell my students. That's what I tell my kids. You know, my kids are well aware of what's happening, of course, and they, you know, wonder about it.

8. 32:17

   you know, that just a few weeks ago, my son was asking me, like, what is the point of prime factorization? What is it used for? So I, you know, I got excited. I started to answer him. I was going to tell him all about RSA and cryptography. And then my daughter came and she's like, oh, well, you know, prime factorization is not useful for anything anymore because ai will do the math and prime factorization is not going to help us become better plumbers so so you know i mean she was she she was trying to be funny but you know they are aware of this but you know uh they still fortunately want to learn how to think i want them to learn how to think you know maybe you know on the margin like i would i would have them spend less of their time you know maybe just like doing the things you know that that they don't like are fully

9. 33:16

   systematized just you know solving a a a well-defined problem you know they still need to do some of that they still need to put in the reps as i was saying you know get the exercise but you know i would uh maybe like even more than i would before i would encourage them to you know ask their own question right you know think of a weird direction in which you know you would like to take this math problem that maybe you care about more than anyone else, right? Because, you know, that, you know, those kinds of soft skills, right, those are going to be where we have an advantage for the longest, where we will hold out the longest sort of without, you know, having our skills replaced by AI. Right. Yeah. So right now, how are you personally