Charles Isbell: Computing, Interactive AI, and Race in America

00:00:00.000 | The following is a conversation with Charles Isbell,

00:00:03.120 | Dean of the College of Computing at Georgia Tech,

00:00:06.360 | a researcher and educator

00:00:08.400 | in the field of artificial intelligence,

00:00:10.720 | and someone who deeply thinks about what exactly is

00:00:14.980 | the field of computing and how do we teach it.

00:00:17.880 | He also has a fascinatingly varied set of interests,

00:00:21.440 | including music, books, movies, sports, and history

00:00:25.560 | that make him especially fun to talk with.

00:00:28.240 | When I first saw him speak, his charisma

00:00:30.960 | immediately took over the room,

00:00:32.960 | and I had a stupid excited smile on my face,

00:00:35.640 | and I knew I had to eventually talk to him on this podcast.

00:00:39.400 | Quick mention of each sponsor,

00:00:41.120 | followed by some thoughts related to the episode.

00:00:44.280 | First is Neuro, the maker of functional sugar-free gum

00:00:48.080 | and mints that I use to give my brain a quick caffeine boost.

00:00:52.280 | Second is Decoding Digital,

00:00:54.420 | a podcast on tech and entrepreneurship

00:00:57.080 | that I listen to and enjoy.

00:00:59.240 | Third is Masterclass, online courses that I watch

00:01:02.120 | from some of the most amazing humans in history.

00:01:05.020 | And finally, Cash App, the app I use to send money

00:01:07.840 | to friends for food and drinks.

00:01:10.760 | Please check out these sponsors in the description

00:01:13.080 | to get a discount and to support this podcast.

00:01:16.360 | As a side note, let me say that I'm trying to make it

00:01:18.600 | so that the conversations with Charles, Eric Weinstein,

00:01:22.360 | and Dan Carlin will be published

00:01:24.180 | before Americans vote for president on November 3rd.

00:01:28.280 | There's nothing explicitly political in these conversations,

00:01:31.480 | but they do touch on something in human nature

00:01:34.520 | that I hope can bring context to our difficult time,

00:01:37.880 | and maybe, for a moment, allow us to empathize

00:01:41.240 | with people we disagree with.

00:01:43.260 | With Eric, we talk about the nature of evil.

00:01:45.880 | With Charles, besides AI and music,

00:01:49.240 | we talk a bit about race in America

00:01:51.480 | and how we can bring more love and empathy

00:01:54.380 | to our online communication.

00:01:57.060 | And with Dan Carlin, well, we talk about

00:02:00.340 | Alexander the Great, Genghis Khan, Hitler, Stalin,

00:02:04.820 | and all the complicated parts of human history in between,

00:02:07.980 | with a hopeful eye toward a brighter future

00:02:10.420 | for our humble little civilization here on Earth.

00:02:14.020 | The conversation with Dan will hopefully be posted tomorrow

00:02:17.740 | on Monday, November 2nd.

00:02:20.260 | If you enjoy this thing, subscribe on YouTube,

00:02:22.760 | review it with Five Stars on Apple Podcasts,

00:02:25.040 | follow on Spotify, support on Patreon,

00:02:27.680 | or connect with me on Twitter @LexFriedman.

00:02:31.240 | And now, here's my conversation with Charles Isbell.

00:02:35.120 | You've mentioned that you love movies and TV shows.

00:02:40.040 | Let's ask an easy question,

00:02:43.160 | but you have to be definitively, objectively conclusive.

00:02:46.720 | What's your top three movies of all time?

00:02:49.880 | - So you're asking me to be definitive and to be conclusive.

00:02:52.460 | That's a little hard, I'm gonna tell you why.

00:02:54.460 | It's very simple.

00:02:55.340 | It's because movies is too broad of a category.

00:02:58.340 | I gotta pick sub-genres.

00:02:59.660 | But I will tell you that of those genres,

00:03:02.140 | I'll pick one or two from each of the genres.

00:03:04.180 | I'll get us to three, so I'm not gonna cheat.

00:03:06.060 | So my favorite comedy of all times,

00:03:09.860 | which is probably my favorite movie of all time,

00:03:12.060 | is "His Girl Friday," which is probably a movie

00:03:15.220 | that you've not ever heard of,

00:03:16.540 | but it's based on a play called "The Front Page"

00:03:19.060 | from, I don't know, early 1900s.

00:03:21.660 | And the movie is a fantastic film.

00:03:26.020 | - What's the story?

00:03:26.860 | What's the independent film?

00:03:28.300 | - No, no, no. - What are we talking about?

00:03:29.700 | - This is one of the movies

00:03:31.180 | that would have been very popular.

00:03:32.260 | It's a screwball comedy.

00:03:33.340 | You ever see "Moonlighting," the TV show?

00:03:35.100 | You know what I'm talking about?

00:03:35.940 | So you've seen these shows where there's a man and a woman,

00:03:38.860 | and they clearly are in love with one another,

00:03:40.380 | and they're constantly fighting

00:03:41.540 | and always talking over each other.

00:03:43.300 | Banter, banter, banter, banter, banter.

00:03:45.300 | This was the movie that started all that

00:03:47.540 | as far as I'm concerned.

00:03:48.980 | It's very much of its time, so it's, I don't know,

00:03:52.340 | must have come out sometime between 1934 and 1939.

00:03:55.260 | I'm not sure exactly when the movie itself came out.

00:03:58.300 | It's black and white.

00:03:59.740 | It's just a fantastic film.

00:04:01.500 | It is hilarious.

00:04:02.420 | - So it's mostly conversation?

00:04:04.380 | - Not entirely, but mostly, mostly.

00:04:06.260 | Just a lot of back and forth.

00:04:07.620 | There's a story there.

00:04:08.940 | Someone's on death row, and they're newspapermen,

00:04:13.940 | including her.

00:04:14.900 | They're all newspapermen.

00:04:16.380 | They were divorced, the editor, the publisher, I guess,

00:04:19.260 | and the reporter, they were divorced.

00:04:22.540 | But they clearly, he's thinking,

00:04:24.260 | trying to get back together,

00:04:25.580 | and there's this whole other thing that's going on.

00:04:27.300 | But none of that matters.

00:04:28.140 | The plot doesn't matter.

00:04:28.980 | - Yeah, it's just a little play in conversation.

00:04:31.660 | - It's fantastic, and I just love everything

00:04:33.980 | about the conversation, because at the end of the day,

00:04:35.860 | sort of narrative and conversation

00:04:37.180 | are the sort of things that drive me,

00:04:38.340 | and so I really like that movie for that reason.

00:04:41.380 | Similarly, I'm now gonna cheat,

00:04:43.140 | and I'm gonna give you two movies as one,

00:04:45.180 | and they're Crouching Tiger, Hidden Dragon, and John Wick.

00:04:49.040 | Both relatively modern.

00:04:50.900 | John Wick, of course, is--

00:04:51.740 | - One, two, or three.

00:04:52.700 | - One.

00:04:53.540 | It gets increasingly, I love them all for different reasons,

00:04:56.220 | and increasingly more ridiculous.

00:04:57.660 | Kind of like loving Alien and Aliens,

00:04:59.660 | despite the fact they're two completely different movies.

00:05:01.780 | But the reason I put Crouching Tiger, Hidden Dragon,

00:05:04.420 | and John Wick together is 'cause I actually think

00:05:06.580 | they're the same movie, or what I like about them,

00:05:08.820 | the same movie, which is both of them create a world

00:05:13.660 | that you're coming in the middle of,

00:05:15.620 | and they don't explain it to you.

00:05:17.740 | But the story is done so well that you pick it up.

00:05:20.100 | So anyone who's seen John Wick, you know,

00:05:22.140 | you have these little coins, and they're headed out,

00:05:24.660 | and there are these rules, and apparently,

00:05:26.340 | every single person in New York City is an assassin.

00:05:29.140 | There's like two people who come through who aren't,

00:05:30.740 | but otherwise they are.

00:05:31.580 | But there's this complicated world,

00:05:33.180 | and everyone knows each other.

00:05:34.220 | They don't sit down and explain it to you,

00:05:35.300 | but you figure it out.

00:05:36.140 | Crouching Tiger, Hidden Dragon's a lot like that.

00:05:38.180 | You get the feeling that this is chapter nine

00:05:40.020 | of a 10-part story, and you've missed

00:05:42.100 | the first eight chapters, and they're not gonna

00:05:43.940 | explain it to you, but there's this sort of

00:05:45.300 | rich world behind you.

00:05:46.140 | - But you get pulled in anyway, like immediately.

00:05:47.460 | - You get pulled in anyway.

00:05:48.300 | So it's just excellent storytelling in both cases,

00:05:50.940 | and very, very different.

00:05:51.780 | - And also you like the outfit, I assume,

00:05:53.460 | the John Wick outfit.

00:05:54.660 | - Oh yeah, of course, of course, yes.

00:05:56.660 | I think John Wick outfit's perfect.

00:05:58.140 | And so that's number two, and then--

00:05:59.860 | - But sorry to pause on that, martial arts,

00:06:02.100 | you have a long list of hobbies.

00:06:04.020 | Like it scrolls off the page, but I didn't see

00:06:06.460 | martial arts as one of them.

00:06:07.820 | - I do not do martial arts, but I certainly--

00:06:09.540 | - Watch it. - Watch martial arts.

00:06:10.380 | Oh, I appreciate it very much.

00:06:11.420 | - Oh, we could talk about every Jackie Chan movie

00:06:13.300 | ever made, and I would be on board with that.

00:06:15.820 | - The Couch Hour 2, like that kind of,

00:06:17.660 | the comedy of it, cop.

00:06:18.940 | - Yes, yes.

00:06:20.100 | By the way, my favorite Jackie Chan movie

00:06:21.980 | would be Drunken Master 2, known in the States

00:06:26.140 | usually as Legend of the Drunken Master.

00:06:28.140 | Actually, Drunken Master, the first one,

00:06:30.780 | is the first kung fu movie I ever saw,

00:06:33.500 | but I did not know that.

00:06:35.020 | - First Jackie Chan movie?

00:06:36.140 | - No, first one ever that I saw and remember,

00:06:38.140 | but I had no idea--

00:06:39.660 | - What is this?

00:06:40.500 | - I didn't know what it was, and I didn't know

00:06:41.340 | that was Jackie Chan.

00:06:42.180 | That was like his first major movie.

00:06:44.180 | I was a kid, it was done in the '70s.

00:06:46.460 | I only later rediscovered that that was actually--

00:06:49.820 | - And he creates his own martial art by,

00:06:52.500 | was he actually drinking, or was he play drinking?

00:06:57.500 | - You mean as an actor, or--

00:06:59.660 | - No. (laughing)

00:07:00.980 | I'm sure as an actor he was--

00:07:02.700 | - No, he was-- - It was the '70s,

00:07:03.820 | or whatever.

00:07:04.660 | - He was definitely drinking, and in the end,

00:07:07.620 | he drinks industrial-grade alcohol.

00:07:09.620 | - Ah, yeah.

00:07:10.660 | - Yeah, and has one of the most fantastic fights ever

00:07:13.980 | in that sub-genre.

00:07:15.220 | Anyway, that's my favorite one of his movies,

00:07:17.180 | but I'll tell you the last movie.

00:07:19.540 | It's actually a movie called Nothing But a Man,

00:07:21.660 | which is a 1960s, starred Ivan Dixon,

00:07:26.380 | who you'll know from Hogan's Heroes, and Abbie Lincoln.

00:07:31.380 | It's just a really small little drama.

00:07:34.140 | It's a beautiful story, but my favorite scenes,

00:07:36.500 | I'm cheating, my favorite,

00:07:38.700 | one of my favorite movies just for the ending

00:07:41.380 | is The Godfather.

00:07:42.780 | I think the last scene of that is just fantastic.

00:07:45.660 | It's the whole movie all summarized in just eight,

00:07:48.100 | nine seconds.

00:07:48.940 | - Godfather Part One?

00:07:49.760 | - Part One.

00:07:50.600 | - How does it end?

00:07:51.500 | I don't think you can, you need to worry about spoilers

00:07:54.020 | if you haven't seen The Godfather.

00:07:55.100 | - Spoiler alert.

00:07:55.940 | It ends with the wife coming to Michael,

00:08:00.700 | and he says, "Just this once,

00:08:02.740 | "I'll let you ask me my business."

00:08:04.180 | And she asks him if he did this terrible thing,

00:08:06.620 | and he looks her in the eye, and he lies,

00:08:07.940 | and he says, "No," and she says, "Thank you,"

00:08:10.180 | and she walks out the door,

00:08:12.620 | and you see him, you see her going out of the door,

00:08:17.620 | and all these people are coming in,

00:08:20.100 | and they're kissing Michael's hands, and Godfather.

00:08:23.420 | And then the camera switches perspectives,

00:08:25.580 | so instead of looking at him, you're looking at her,

00:08:28.360 | and the door closes in her face,

00:08:31.540 | and that's the end of the movie,

00:08:32.460 | and that's the whole movie right there.

00:08:34.620 | - Do you see parallels between that

00:08:36.060 | and your position as dean at Georgia Tech, Carl?

00:08:38.380 | - Ha!

00:08:39.220 | Just kidding, trick question.

00:08:40.380 | (laughing)

00:08:42.140 | Sometimes, certainly if the door gets closed

00:08:43.860 | on me every once in a while.

00:08:45.540 | - Okay, that was a rhetorical question.

00:08:48.120 | You've also mentioned that you, I think,

00:08:50.960 | enjoy all kinds of experiments, including on yourself,

00:08:54.160 | but I saw a video where you said you did an experiment

00:08:56.980 | where you tracked all kinds of information about yourself,

00:09:00.020 | and a few others, sort of wiring up your home,

00:09:05.540 | and this little idea that you mentioned in that video,

00:09:09.060 | which is kind of interesting,

00:09:10.180 | that you thought that two days' worth of data

00:09:14.660 | is enough to capture majority of the behavior

00:09:17.420 | of the human being.

00:09:18.420 | First, can you describe what the heck you did

00:09:23.140 | to collect all that data, 'cause it's fascinating,

00:09:25.220 | just like little details of how you collect that data,

00:09:27.820 | and also what your intuition behind the two days is.

00:09:31.080 | - So, first off, it has to be the right two days,

00:09:32.540 | but I was thinking of a very specific experiment.

00:09:34.940 | There's actually a suite of them that I've been a part of,

00:09:36.980 | and other people have done this, of course.

00:09:38.140 | I just sort of dabbled in that part of the world.

00:09:40.460 | But to be very clear, the specific thing

00:09:41.980 | that I was talking about had to do with recording

00:09:45.100 | all the IR going on in my, infrared going on in my house.

00:09:48.820 | So this is a long time ago, so everything's being controlled

00:09:52.020 | by pressing buttons on remote controls,

00:09:54.340 | as opposed to speaking to Alexa, or Siri,

00:09:56.900 | or someone like that.

00:09:58.020 | And I was just trying to figure out

00:09:59.480 | if you could get enough data on people

00:10:00.900 | to figure out what they were gonna do

00:10:02.660 | with their TVs or their lights.

00:10:04.200 | My house was completely wired up at the time.

00:10:06.460 | But you know, what, I'm about to look at a movie,

00:10:09.500 | or I'm about to turn on the TV, or whatever,

00:10:10.900 | and just see what I could predict from it.

00:10:12.940 | It was kind of surprising, it shouldn't have been.

00:10:16.500 | But that's all very easy to do, by the way,

00:10:18.340 | just capturing all the little stuff.

00:10:19.580 | I mean, it's a bunch of computer systems.

00:10:20.700 | It's really easy to capture,

00:10:21.540 | if you know what you're looking for.

00:10:23.180 | At Georgia Tech, long before I got there,

00:10:24.620 | we had this thing called the Aware Home,

00:10:26.540 | where everything was wired up,

00:10:27.780 | and you saw, you captured everything that was going on.

00:10:29.780 | Nothing even difficult, not with video,

00:10:31.500 | or anything like that, just the way that the system

00:10:33.580 | was just capturing everything.

00:10:35.700 | So it turns out that, and I did this with myself,

00:10:39.780 | and then I had students, and they worked with

00:10:41.540 | many other people, and it turns out at the end of the day,

00:10:44.500 | people do the same things over and over and over again.

00:10:48.120 | So it has to be the right two days, like a weekend,

00:10:50.100 | but it turns out, not only can you predict

00:10:52.720 | what someone's going to do next,

00:10:53.820 | at the level of what button they're gonna press next

00:10:55.660 | on a remote control, but you can do it

00:10:58.980 | with something really, really simple,

00:11:00.740 | like a, you don't even need a hidden Markov model,

00:11:02.740 | it's like a Mark, just simply, I press this,

00:11:05.220 | this is my prediction of the next thing.

00:11:06.420 | And it turns out, you get 93% accuracy,

00:11:09.900 | just by doing something very simple and stupid,

00:11:11.900 | and just counting statistics.

00:11:13.780 | But what was actually more interesting,

00:11:15.220 | is that you could use that information.

00:11:16.580 | This comes up again and again in my work.

00:11:18.520 | If you try to represent people or objects

00:11:21.780 | by the things they do, the things you can measure about them

00:11:25.060 | that have to do with action in the world,

00:11:27.500 | so distribution over actions, and you try to represent them

00:11:30.420 | by the distribution of actions that are done on them,

00:11:34.740 | then you do a pretty good job of sort of

00:11:37.020 | understanding how people are,

00:11:38.380 | and they cluster remarkably well, in fact, irritatingly so.

00:11:43.220 | And so, by clustering people this way, you can,

00:11:47.260 | maybe, you know, I got the 93% accuracy

00:11:49.620 | of what's the next button you're gonna press,

00:11:51.340 | but I can get 99% accuracy, or somewhere there's about,

00:11:54.620 | on the collections of things you might press.

00:11:56.460 | And it turns out, the things that you might press

00:11:58.340 | are all related to each other in exactly

00:12:00.700 | the way that you would expect.

00:12:01.540 | So, for example, all the numbers on a keypad,

00:12:05.900 | it turns out, all have the same behavior,

00:12:07.420 | with respect to you as a human being.

00:12:09.140 | And so, you would naturally cluster them together,

00:12:11.420 | and you discover that numbers are all related

00:12:15.140 | to one another in some way, and all these other things.

00:12:17.100 | And then, and here's the part that I think's important.

00:12:19.700 | I mean, you can see this in all kinds of things.

00:12:22.060 | Every individual is different,

00:12:25.060 | but any given individual is remarkably predictable,

00:12:28.460 | because you keep doing the same things over and over again.

00:12:30.740 | And the two things that I've learned,

00:12:32.460 | in the long time that I've been thinking about this,

00:12:34.100 | is people are easily predictable,

00:12:36.720 | and people hate when you tell them

00:12:37.900 | that they're easily predictable.

00:12:39.580 | But they are, and there you go.

00:12:42.100 | - What about, let me play devil's advocate,

00:12:45.140 | and philosophically speaking, is it possible to say

00:12:48.540 | that what defines humans is the outlier?

00:12:51.580 | So, even though some large percentage of our behaviors,

00:12:55.560 | whatever the signal we measure is the same,

00:12:57.660 | and it would cluster nicely,

00:12:59.380 | but maybe it's the special moments

00:13:01.420 | of when we break out of the routine,

00:13:03.960 | is the definitive things,

00:13:05.680 | and the way we break out of that routine

00:13:07.540 | for each one of us might be different?

00:13:09.260 | - It's possible.

00:13:10.100 | I would say it a little differently, I think.

00:13:12.660 | I would make two things.

00:13:13.660 | One is, I'm gonna disagree with the premise, I think,

00:13:16.500 | but that's fine.

00:13:17.340 | I think the way I would put it is,

00:13:21.480 | there are people who are very different

00:13:23.140 | from lots of other people, but they're not 0%.

00:13:25.920 | They're closer to 10%, right?

00:13:27.560 | So, in fact, even if you do this kind of clustering

00:13:29.180 | of people, that'll turn out to be

00:13:30.580 | this small number of people.

00:13:31.700 | They all behave like each other,

00:13:33.420 | even if they individually behave very differently

00:13:35.500 | from everyone else.

00:13:37.180 | So, I think that's kind of important.

00:13:38.780 | But what you're really asking, I think,

00:13:40.300 | and I think this is really a question,

00:13:41.620 | is what do you do when you're faced

00:13:44.140 | with a situation you've never seen before?

00:13:46.580 | What do you do when you're faced

00:13:47.580 | with an extraordinary situation

00:13:48.860 | maybe you've seen others do,

00:13:49.900 | and you're actually forced to do something?

00:13:51.520 | You react to that very differently,

00:13:52.720 | and that is the thing that makes you human.

00:13:54.000 | I would agree with that, at least at a philosophical level,

00:13:56.520 | that it's the times when you are faced

00:13:59.800 | with something difficult, a decision that you have to make,

00:14:04.180 | where the answer isn't easy,

00:14:05.760 | even if you know what the right answer is,

00:14:07.680 | that's sort of what defines you as the individual,

00:14:09.560 | and I think what defines people broadly.

00:14:11.840 | It's the hard problem, it's not the easy problem.

00:14:13.440 | It's the thing that's gonna hurt you.

00:14:14.500 | It's not the thing.

00:14:15.440 | It's not even that it's difficult.

00:14:17.520 | It's just that you know that the outcome

00:14:19.700 | is going to be highly suboptimal for you,

00:14:22.300 | and I do think that that's a reasonable place to start

00:14:25.380 | for the question of what makes us human.

00:14:28.140 | - So before we talk about, sort of explore

00:14:30.100 | the different ideas underlying

00:14:31.780 | interactive artificial intelligence,

00:14:33.300 | which we are working on, let me just go along this thread

00:14:36.940 | to skip to kind of our world of social media,

00:14:39.660 | which is something that,

00:14:41.660 | at least on the artificial intelligence side,

00:14:43.500 | you think about there's a popular narrative.

00:14:47.060 | I don't know if it's true,

00:14:48.680 | but that we have these silos in social media,

00:14:53.480 | and we have these clusterings, as you're kind of mentioning,

00:14:56.600 | and the idea is that, along that narrative,

00:15:00.840 | is that we wanna break each other out of those silos

00:15:05.840 | so we can be empathetic to other people.

00:15:11.000 | If you're a Democrat, you'd be empathetic to the Republican.

00:15:14.160 | If you're a Republican, you're empathetic to a Democrat.

00:15:16.420 | Those are just two silly bins

00:15:18.480 | that we seem to be very excited about,

00:15:21.000 | but there's other binnings that we can think about.

00:15:24.120 | Is there, from an artificial intelligence perspective,

00:15:28.120 | 'cause you're just saying we cluster along the data,

00:15:30.920 | but then interactive artificial intelligence

00:15:33.680 | is referring to throwing agents into that mix,

00:15:36.680 | AI systems in that mix,

00:15:38.400 | helping us, interacting with us humans,

00:15:41.100 | and maybe getting us out of those silos.

00:15:43.900 | Is that something that you think is possible?

00:15:46.800 | Do you see a hopeful possibility

00:15:50.240 | for artificial intelligence systems

00:15:52.160 | in these large networks of people

00:15:55.320 | to get us outside of our habits,

00:15:58.320 | in at least the idea space,

00:16:00.200 | to where we can sort of be empathetic

00:16:04.240 | to other people's lived experiences,

00:16:07.900 | other people's points of view, all that kind of stuff?

00:16:11.440 | - Yes, and I actually don't think it's that hard.

00:16:13.320 | Well, it's not hard in this sense.

00:16:15.300 | So imagine that you can,

00:16:17.180 | now let's make life simple for a minute.

00:16:20.600 | Let's assume that you can do a kind of partial ordering

00:16:23.660 | over ideas or clusterings of behavior.

00:16:27.080 | It doesn't even matter what I mean here,

00:16:28.920 | so long as there's some way that this is a cluster,

00:16:30.840 | this is a cluster, there's some edge between them, right?

00:16:33.600 | They don't quite touch even, or maybe they come very close.

00:16:36.020 | If you can imagine that conceptually,

00:16:38.620 | then the way you get from here to here

00:16:40.040 | is not by going from here to here.

00:16:41.100 | The way you get from here to here is you find the edge

00:16:43.020 | and you move slowly together, right?

00:16:44.800 | And I think that machines are actually very good

00:16:46.560 | at that sort of thing, once we can kind of define the problem

00:16:48.720 | either in terms of behavior or ideas or words or whatever.

00:16:51.420 | So it's easy in the sense that

00:16:52.980 | if you already have the network

00:16:54.600 | and you know the relationships,

00:16:56.000 | the edges and sort of the strings on them,

00:16:57.400 | and you kind of have some semantic meaning for them,

00:17:00.720 | the machine doesn't have to, you do as the designer,

00:17:03.800 | then yeah, I think you can kind of move people along

00:17:05.680 | and sort of expand them.

00:17:06.920 | But it's harder than that.

00:17:07.760 | And the reason it's harder than that,

00:17:09.560 | or sort of coming up with the network structure itself

00:17:12.960 | is hard is because I'm gonna tell you a story

00:17:14.500 | that someone else told me, and I don't,

00:17:17.460 | I may get some of the details a little bit wrong,

00:17:19.140 | but it's roughly, it roughly goes like this.

00:17:22.280 | You take two sets of people from the same backgrounds

00:17:25.740 | and you want them to solve a problem.

00:17:27.820 | So you separate them up, which we do all the time,

00:17:29.740 | right, oh, you know, we're gonna break out in the,

00:17:31.380 | we're gonna break out groups,

00:17:32.220 | you're gonna go over there and you're gonna talk about this,

00:17:33.540 | you're gonna go over there and you're gonna talk about this.

00:17:35.120 | And then you have them sort of in this big room,

00:17:37.700 | but far apart from one another,

00:17:38.740 | and you have them sort of interact with one another.

00:17:41.060 | When they come back to talk about what they learned,

00:17:43.440 | you wanna merge what they've done together,

00:17:45.080 | it can be extremely hard because they don't,

00:17:48.160 | they basically don't speak the same language anymore.

00:17:50.080 | Like when you create these problems and you dive into them,

00:17:52.200 | you create your own language.

00:17:53.680 | So the example this one person gave me,

00:17:55.280 | which I found kind of interesting

00:17:57.080 | 'cause we were in the middle of that at the time,

00:17:58.560 | was they're sitting over there

00:18:00.960 | and they're talking about these rooms that you can see,

00:18:03.560 | but you're seeing them from different vantage points,

00:18:05.480 | depending upon what side of the room you're on.

00:18:07.520 | They can see a clock very easily.

00:18:09.840 | And so they start referring to the room

00:18:11.220 | as the one with the clock.

00:18:12.580 | This group over here, looking at the same room,

00:18:15.540 | they can see the clock,

00:18:16.420 | but it's not in their line of sight or whatever,

00:18:18.420 | so they end up referring to it by some other way.

00:18:22.700 | When they get back together

00:18:23.620 | and they're talking about things,

00:18:25.240 | they're referring to the same room

00:18:26.720 | and they don't even realize they're referring

00:18:27.980 | to the same room.

00:18:28.820 | And in fact, this group doesn't even see

00:18:30.060 | that there's a clock there

00:18:30.900 | and this group doesn't see whatever it is.

00:18:32.180 | The clock on the wall is the thing that stuck with me.

00:18:33.980 | So if you create these different silos,

00:18:35.620 | the problem isn't that the ideologies disagree,

00:18:38.340 | it's that you're using the same words

00:18:40.940 | and they mean radically different things.

00:18:42.800 | The hard part is just getting them to agree on the,

00:18:45.720 | well, maybe we'd say the axioms in our world, right?

00:18:48.880 | But just get them to agree on some basic definitions.

00:18:52.560 | Because right now they're talking past each other,

00:18:55.020 | just completely talking past each other.

00:18:56.680 | That's the hard part, getting them to meet,

00:18:58.700 | getting them to interact, that may not be that difficult.

00:19:01.360 | Getting them to see where their language

00:19:04.180 | is leading them to lead past one another,

00:19:05.760 | that's the hard part.

00:19:07.480 | - It's a really interesting question to me.

00:19:09.160 | It could be on the layer of language,

00:19:10.380 | but it feels like there's multiple layers to this.

00:19:12.520 | Like it could be worldview, it could be,

00:19:14.520 | I mean, it all boils down to empathy,

00:19:16.000 | being able to put yourself in the shoes of the other person,

00:19:19.560 | to learn the language, to learn visually

00:19:22.960 | how they see the world, to learn the,

00:19:27.680 | I mean, I experience this now with trolls,

00:19:30.960 | the degree of humor in that world.

00:19:33.500 | For example, I talk about love a lot.

00:19:35.040 | I'm very lucky to have this amazing community

00:19:39.240 | of loving people, but whenever I encounter trolls,

00:19:42.200 | they always roll their eyes at the idea of love

00:19:45.040 | because it's so quote unquote cringe.

00:19:47.360 | - Yeah.

00:19:48.200 | - So they show love by derision, I would say.

00:19:53.200 | And I think about, on the human level,

00:19:58.200 | that's a whole 'nother discussion,

00:19:59.280 | that's psychology, that's sociology, so on.

00:20:01.800 | But I wonder if AI systems can help somehow

00:20:06.080 | and bridge the gap of what is this person's life like?

00:20:11.080 | Encourage me to just ask that question,

00:20:15.280 | to put myself in their shoes,

00:20:17.280 | to experience the agitations, the fears,

00:20:20.280 | the hopes they have, to experience,

00:20:23.420 | even just to think about what was their upbringing like,

00:20:28.440 | like having a single parent home

00:20:31.440 | or a shitty education or all those kinds of things,

00:20:35.400 | just to put myself in that mind space.

00:20:38.440 | It feels like that's really important

00:20:40.480 | for us to bring those clusters together,

00:20:44.200 | to find that similar language,

00:20:45.760 | but it's unclear how AI can help that

00:20:48.240 | because it seems like AI systems need

00:20:49.880 | to understand both parties first.

00:20:52.360 | - So the word understander is doing a lot of work, right?

00:20:55.040 | - Yeah, yes.

00:20:56.160 | - Do you have to understand it

00:20:57.120 | or do you just simply have to note

00:20:58.960 | that there is something similar as a point to touch, right?

00:21:03.320 | So you use the word empathy,

00:21:06.320 | and I like that word for a lot of reasons.

00:21:08.560 | I think you're right in the way that you're using

00:21:10.200 | and the way that you're describing it,

00:21:11.080 | but let's separate it from sympathy, right?

00:21:13.080 | So sympathy is feeling sort of for someone.

00:21:17.060 | Empathy is kind of understanding where they're coming from

00:21:19.240 | and how they feel, right?

00:21:20.260 | And for most people, those things go hand in hand.

00:21:23.960 | For some people, some are very good at empathy

00:21:25.940 | and very bad at sympathy.

00:21:28.040 | Some people cannot experience,

00:21:29.840 | well, my observation would be, I'm not a psychologist,

00:21:32.320 | my observation would be that some people

00:21:34.280 | seem incapable of feeling sympathy

00:21:36.080 | unless they feel empathy first.

00:21:38.080 | You can understand someone,

00:21:39.320 | understand where they're coming from and still think,

00:21:40.920 | "No, I can't support that," right?

00:21:43.520 | It doesn't mean that the only way,

00:21:45.840 | because if that isn't the case,

00:21:47.640 | then what it requires is that you must,

00:21:51.600 | the only way that you can,

00:21:52.840 | to understand someone means you must agree

00:21:54.820 | with everything that they do.

00:21:56.840 | Which, is it right, right?

00:21:58.880 | And if the only way I can feel for someone

00:22:02.000 | is to completely understand them

00:22:03.520 | and make them like me in some way,

00:22:06.240 | well, then we're lost, right?

00:22:07.800 | Because we're not all exactly like each other.

00:22:09.880 | I don't have to understand everything

00:22:10.880 | that you've gone through.

00:22:11.700 | It helps, clearly.

00:22:12.920 | But they're separable ideas, right?

00:22:14.360 | Even though they get clearly tangled up in one another.

00:22:16.920 | So what I think AI could help you do, actually,

00:22:19.440 | is if, and I'm being quite fanciful, as it were,

00:22:23.180 | but if you think of these as kind of,

00:22:25.280 | I understand how you interact,

00:22:26.660 | the words that you use, the actions you take,

00:22:28.880 | I have some way of doing this,

00:22:29.720 | let's not worry about what that is,

00:22:31.960 | but I can see you as a kind of distribution

00:22:34.320 | of experiences and actions taken upon you,

00:22:36.880 | things you've done, and so on.

00:22:38.040 | And I can do this with someone else,

00:22:39.660 | and I can find the places where there's

00:22:41.680 | some kind of commonality, a mapping, as it were,

00:22:44.700 | even if it's not total.

00:22:46.120 | If I think of it as distribution, right,

00:22:48.520 | then I can take the cosine of the angle between you,

00:22:51.100 | and if it's zero, you've got nothing in common.

00:22:53.840 | If it's one, you're completely the same person.

00:22:56.000 | Well, you're probably not one,

00:22:59.240 | you're almost certainly not zero.

00:23:00.520 | If I can find the place where there's the overlap,

00:23:02.580 | then I might be able to introduce you on that basis,

00:23:04.560 | or connect you in that way,

00:23:07.740 | and make it easier for you to take that step of empathy.

00:23:10.880 | It's not impossible to do,

00:23:14.520 | although I wonder if it requires that everyone involved

00:23:18.840 | is at least interested in asking the question.

00:23:21.320 | So maybe the hard part is just getting them interested

00:23:23.040 | in asking the question.

00:23:23.880 | In fact, maybe if you can get them to ask the question,

00:23:25.560 | how are we more alike than we are different,

00:23:27.400 | they'll solve it themselves.

00:23:28.400 | Maybe that's the problem that AI should be working on,

00:23:30.280 | not telling you how you're similar or different,

00:23:32.820 | but just getting you to decide

00:23:34.200 | that it's worthwhile asking the question.

00:23:35.800 | - So, no.

00:23:36.640 | - It feels like an economist's answer, actually.

00:23:39.120 | - Well, people, okay, first of all,

00:23:40.360 | people like it when I would disagree.

00:23:42.020 | So let me disagree slightly,

00:23:43.760 | which is I think everything you said is brilliant,

00:23:46.040 | but I tend to believe, philosophically speaking,

00:23:49.840 | that people are interested underneath it all,

00:23:52.640 | and I would say that AI,

00:23:54.400 | the possibility that an AI system

00:23:58.640 | would show the commonality is incredible.

00:24:01.400 | That's a really good starting point.

00:24:02.720 | I would say if on social media,

00:24:05.940 | I could discover the common things,

00:24:09.160 | deep or shallow, between me and a person

00:24:12.000 | who there's tension with,

00:24:14.240 | I think that my basic human nature

00:24:17.080 | would take over from there,

00:24:18.400 | and I think enjoy that commonality

00:24:22.520 | and there's something sticky about that

00:24:25.320 | that my mind will linger on,

00:24:27.040 | and that person in my mind will become warmer and warmer,

00:24:30.880 | and I'll start to feel more and more compassion towards them.

00:24:34.240 | I think for majority of the population, that's true,

00:24:37.560 | but that's a hypothesis.

00:24:39.600 | - Yeah, I mean, it's an empirical question, right?

00:24:41.320 | You'd have to figure it out.

00:24:42.840 | I mean, I want to believe you're right,

00:24:44.200 | and so I'm gonna say that I think you're right.

00:24:46.600 | Of course, some people come to those things

00:24:49.440 | for the purpose of trolling, right?

00:24:52.300 | And it doesn't matter.

00:24:53.140 | They're playing a different game.

00:24:54.880 | - Yeah.

00:24:55.720 | - But I don't know.

00:24:56.540 | My experience is it requires two things.

00:24:59.000 | It requires, in fact, maybe this is really,

00:25:00.960 | at the end, what you're saying,

00:25:01.800 | and I do agree with this for sure.

00:25:03.840 | So it's hard to hold onto that kind of anger

00:25:08.840 | or to hold onto just the desire

00:25:13.400 | to humiliate someone for that long.

00:25:14.880 | It's just difficult to do.

00:25:15.920 | It takes a toll on you,

00:25:17.800 | but more importantly, we know this,

00:25:20.080 | both from people having done studies on it,

00:25:21.960 | but also from our own experiences,

00:25:23.700 | that it is much easier to be dismissive of a person

00:25:27.100 | if they're not in front of you, if they're not real, right?

00:25:29.840 | So much of the history of the world

00:25:33.020 | is about making people other, right?

00:25:35.180 | So if you're on social media, if you're on the web,

00:25:37.760 | if you're doing whatever on the internet,

00:25:39.680 | being forced to deal with someone as a person,

00:25:42.760 | some equivalent to being in the same room,

00:25:46.040 | makes a huge difference, 'cause then you're,

00:25:47.520 | one, you're forced to deal with their humanity

00:25:49.280 | 'cause it's in front of you.

00:25:50.280 | The other is, of course, that they might punch you

00:25:52.460 | in the face if you go too far,

00:25:53.540 | so both of those things kind of work together,

00:25:55.800 | I think, to the right end.

00:25:57.360 | So I think bringing people together

00:25:59.880 | is really a kind of substitute for forcing them

00:26:04.020 | to see the humanity in another person

00:26:06.120 | and to not be able to treat them as bits.

00:26:08.020 | It's hard to troll someone

00:26:09.080 | when you're looking them in the eye.

00:26:10.540 | This is very difficult to do.

00:26:12.760 | - Agreed.

00:26:13.800 | Your broad set of research interests

00:26:15.340 | fall under interactive AI, as I mentioned,

00:26:19.140 | which is a fascinating set of ideas,

00:26:21.460 | and you have some concrete things

00:26:23.240 | that you're particularly interested in,

00:26:25.520 | but maybe could you talk about how you think

00:26:28.760 | about the field of interactive artificial intelligence?

00:26:31.920 | - Sure.

00:26:32.760 | So let me say up front that if you look at,

00:26:34.960 | certainly my early work,

00:26:35.800 | but even if you look at most of it,

00:26:37.600 | I'm a machine learning guy.

00:26:38.920 | Right, I do machine learning.

00:26:40.680 | First paper I ever published was in NIPS.

00:26:43.360 | Back then it was NIPS, now it's NIRPS.

00:26:45.640 | It's a long story there.

00:26:46.800 | Anyway, that's another thing.

00:26:48.380 | So I'm a machine learning guy.

00:26:49.220 | I believe in data, I believe in statistics

00:26:50.940 | and all those kind of things.

00:26:51.780 | - Yes.

00:26:52.600 | - And the reason I'm bringing that up

00:26:53.440 | is even though I'm a newfangled

00:26:54.900 | statistical machine learning guy

00:26:56.580 | and have been for a very long time,

00:26:58.100 | the problem I really care about is AI.

00:27:00.500 | I care about artificial intelligence.

00:27:01.980 | I care about building some kind of intelligent artifact,

00:27:05.820 | however that gets expressed,

00:27:07.900 | that would be at least as intelligent as humans

00:27:12.060 | and as interesting as humans,

00:27:13.580 | perhaps in their own way.

00:27:16.660 | - So that's the deep underlying love and dream

00:27:18.840 | is the bigger AI.

00:27:20.800 | - Yes.

00:27:21.640 | - It's the bigger, whatever the heck that is.

00:27:22.840 | - Yeah, the machine learning in some ways

00:27:24.080 | is a means to the end.

00:27:25.840 | It is not the end.

00:27:26.880 | And I don't understand how one could be intelligent

00:27:29.480 | without learning, so therefore I gotta figure out

00:27:31.240 | how to do that, right?

00:27:32.320 | So that's important.

00:27:33.160 | But machine learning, by the way, is also a tool.

00:27:35.440 | I said statistical because that's what most people

00:27:37.520 | think of themselves, machine learning people.

00:27:39.200 | That's how they think.

00:27:40.040 | I think it's what Pat Langley might disagree,

00:27:41.440 | or at least 1980s Pat Langley might disagree

00:27:44.520 | with what it takes to do machine learning.

00:27:46.820 | But I care about the AI problem,

00:27:48.460 | which is why it's interactive AI,

00:27:49.620 | not just interactive ML.

00:27:50.980 | I think it's important to understand

00:27:52.620 | that there's a long-term goal here,

00:27:54.500 | which I will probably never live to see,

00:27:56.160 | but I would love to have been a part of,

00:27:58.140 | which is building something truly intelligent

00:28:00.740 | outside of ourselves.

00:28:02.380 | - Can we take a tiny tangent?

00:28:03.900 | - Sure.

00:28:04.740 | - Or am I interrupting?

00:28:05.580 | Which is, is there something you can say concrete

00:28:10.100 | about the mysterious gap between the subset ML

00:28:14.240 | and the bigger AI?

00:28:15.980 | What's missing?

00:28:16.900 | What do you think?

00:28:18.220 | I mean, obviously it's totally unknown,

00:28:21.040 | not totally, but in part unknown at this time,

00:28:23.220 | but is it something like what Pat Langley's,

00:28:25.700 | is it knowledge, like expert system reasoning

00:28:28.460 | type of kind of thing?

00:28:29.820 | - So AI is bigger than ML, but ML is bigger than AI.

00:28:33.060 | This is kind of the real problem here,

00:28:35.620 | is that they're really overlapping things

00:28:36.980 | that are really interested in slightly different problems.

00:28:39.540 | I tend to think of ML, and there are many people out there

00:28:41.500 | are gonna be very upset at me about this,

00:28:42.700 | but I tend to think of ML being much more concerned

00:28:44.600 | with the engineering of solving a problem.

00:28:46.320 | I'm an AI about the sort of more philosophical goal

00:28:48.880 | of true intelligence, and that's the thing

00:28:50.960 | that motivates me, even if I end up finding myself

00:28:53.480 | living in this kind of engineering-ish space.

00:28:56.120 | I've now made Michael Jordan upset,

00:28:57.840 | but you know, it's, to me, they just feel very different.

00:29:01.640 | You're just measuring them differently.

00:29:03.400 | Your sort of goals of where you're trying to be

00:29:06.360 | are somewhat different, but to me, AI is about

00:29:08.600 | trying to build that intelligent thing.

00:29:10.520 | And typically, but not always, for the purpose

00:29:14.500 | of understanding ourselves a little bit better.

00:29:16.820 | Machine learning is, I think, trying to solve the problem,

00:29:19.860 | whatever that problem is.

00:29:21.020 | Now, that's my take.

00:29:22.140 | Others, of course, would disagree.

00:29:23.520 | - So on that note, so with the interactive AI,

00:29:26.860 | do you tend to, in your mind, visualize AI

00:29:29.060 | as a singular system, or is it as a collective,

00:29:32.260 | huge amount of systems interacting with each other?

00:29:34.340 | Like, is the social interaction of us humans

00:29:38.380 | and of AI systems fundamental to intelligence?

00:29:41.920 | - I think, well, it's certainly fundamental

00:29:43.240 | to our kind of intelligence, right?

00:29:44.960 | And I actually think it matters quite a bit.

00:29:46.800 | So the reason the interactive AI part matters to me

00:29:51.040 | is because I don't, this is gonna sound simple,

00:29:55.660 | but I don't care whether a tree makes a sound

00:30:00.180 | when it falls and there's no one around,

00:30:01.720 | because I don't think it matters, right?

00:30:03.720 | If there's no observer, in some sense.

00:30:05.320 | And I think what's interesting about the way

00:30:06.800 | that we're intelligent is we're intelligent

00:30:09.780 | with other people, right, or other things, anyway.

00:30:12.960 | And we go out of our way to make other things intelligent.

00:30:15.600 | We're hardwired to find intention,

00:30:17.600 | even whether there is no intention.

00:30:19.120 | Why, I mean, anthropomorphize everything.

00:30:21.000 | I think, anyway.

00:30:21.840 | I think the interactive AI part is being intelligent

00:30:25.280 | in and of myself in isolation is a meaningless act,

00:30:28.840 | in some sense.

00:30:30.520 | The correct answer is you have to be intelligent

00:30:32.520 | in the way that you interact with others.

00:30:33.520 | That's also efficient, because it allows you

00:30:35.440 | to learn faster, because you can import from past history.

00:30:39.760 | It also allows you to be efficient

00:30:41.160 | in the transmission of that.

00:30:42.440 | So we ask ourselves about me.

00:30:44.920 | Am I intelligent?

00:30:45.800 | Clearly, I think so.

00:30:47.600 | But I'm also intelligent as a part of a larger species

00:30:50.300 | and group of people, and we're trying to move

00:30:52.040 | the species forward as well.

00:30:53.520 | And so I think that notion of being intelligent

00:30:56.320 | with others is kind of the key thing,

00:30:57.920 | because otherwise you come and you go,

00:30:59.840 | and then it doesn't matter.

00:31:01.520 | And so that's why I care about that aspect of it.

00:31:04.600 | And it has lots of other implications.

00:31:07.320 | One is not just building something intelligent with others,

00:31:10.320 | but understanding that you can't always communicate

00:31:12.360 | with those others.

00:31:13.180 | They have been in a room where there's a clock on the wall

00:31:15.480 | that you haven't seen, which means you have to spend

00:31:17.240 | an enormous amount of time communicating

00:31:18.720 | with one another constantly in order to figure out

00:31:21.920 | what each other wants.

00:31:24.000 | So, I mean, this is why people project, right?

00:31:25.620 | You project your own intentions and your own reasons

00:31:28.740 | for doing things onto others as a way of understanding them

00:31:31.040 | so that you know how to behave.

00:31:32.960 | But by the way, you, completely predictable person,

00:31:35.920 | I don't know how you're predictable,

00:31:37.040 | I don't know you well enough, but you probably eat

00:31:38.880 | the same five things over and over again,

00:31:40.400 | or whatever it is that you do, right?

00:31:41.840 | I know I do.

00:31:43.200 | If I'm going to a new Chinese restaurant,

00:31:44.720 | I will get General Gao's chicken,

00:31:45.840 | because that's the thing that's easy to get.

00:31:47.740 | I will get hot and sour soup.

00:31:49.840 | People do the things that they do,

00:31:51.120 | but other people get the chicken and broccoli.

00:31:53.800 | I think I can push this analogy way too far.

00:31:55.840 | The chicken and broccoli--

00:31:56.680 | - I don't know what's wrong with those people.

00:31:57.840 | - I don't know what's wrong with them either.

00:32:00.200 | - That's not good.

00:32:01.040 | - We have all had our trauma.

00:32:02.520 | So they get their chicken and broccoli

00:32:03.720 | and their egg drop soup or whatever.

00:32:06.000 | We got to communicate, and it's going to change, right?

00:32:08.440 | So it's not, interactive AI is not just about

00:32:11.960 | learning to solve a problem or a task.

00:32:14.040 | It's about having to adapt that over time,

00:32:16.440 | over a very long period of time,

00:32:17.720 | and interacting with other people,

00:32:19.620 | who will themselves change?

00:32:20.880 | This is what we mean about things like adaptable models,

00:32:22.840 | right, that you have to have a model,

00:32:24.200 | that model's going to change.

00:32:25.160 | And by the way, it's not just the case

00:32:26.500 | that you're different from that person,

00:32:28.040 | but you're different from the person you were 15 minutes ago,

00:32:30.320 | or certainly 15 years ago,

00:32:31.920 | and I have to assume that you're at least going to drift.

00:32:34.760 | Hopefully not too many discontinuities,

00:32:36.320 | but you're going to drift over time.

00:32:38.040 | And I have to have some mechanism for adapting to that

00:32:42.360 | as you and an individual over time,

00:32:43.680 | and across individuals over time.

00:32:46.040 | - On the topic of adaptive modeling,

00:32:48.020 | and you talk about lifelong learning,

00:32:51.000 | which is, I think, a topic that's understudied,

00:32:56.000 | or maybe because nobody knows what to do with it.

00:32:59.520 | But if you look at Alexa,

00:33:01.520 | or most of our artificial intelligence systems

00:33:04.360 | that are primarily machine learning based systems,

00:33:07.000 | or dialogue systems, all those kinds of things,

00:33:08.960 | they know very little about you,

00:33:12.360 | in the sense of the lifelong learning sense,

00:33:15.640 | that we learn, as humans,

00:33:19.760 | we learn a lot about each other,

00:33:22.080 | not in the quantity effects,

00:33:24.880 | but the temporally rich set of information

00:33:29.880 | that seems to pick up the crumbs along the way

00:33:34.360 | that somehow seems to capture a person pretty well.

00:33:37.080 | Do you have any ideas how to do lifelong learning?

00:33:42.080 | Because it seems like most of the machine learning community

00:33:44.560 | does not.

00:33:45.480 | - No, well, by the way,

00:33:46.320 | not only does the machine learning community

00:33:47.320 | not spend a lot of time on lifelong learning,

00:33:49.840 | I don't think they spend a lot of time on learning, period,

00:33:52.760 | in the sense that they tend to be very task-focused.

00:33:55.120 | Everybody is over-fitting to whatever problem

00:33:57.520 | is they happen to have.

00:33:58.360 | They're over-engineering their solutions to the task.

00:34:01.480 | Even the people, and I think these people do,

00:34:03.720 | are trying to solve a hard problem of transfer learning,

00:34:05.920 | right, I'm gonna learn on one task,

00:34:07.080 | then learn the other task.

00:34:08.040 | You still end up creating the task.

00:34:09.680 | You know, it's like looking for your keys

00:34:11.060 | where the light is, 'cause that's where the light is, right?

00:34:12.700 | It's not because the keys have to be there.

00:34:14.920 | I mean, one could argue that we tend to do this in general.

00:34:18.400 | We tend to kind of do it as a group.

00:34:20.600 | We tend to hill climb and get stuck in local optima.

00:34:23.280 | And I think we do this in the small as well.

00:34:26.360 | I think it's very hard to do.

00:34:28.360 | Because, so, look, here's the hard thing about AI, right?

00:34:32.240 | The hard thing about AI is it keeps changing on us, right?

00:34:34.600 | You know, what is AI?

00:34:35.440 | AI is the art and science of making computers act

00:34:38.520 | the way they do in the movies, right?

00:34:39.840 | That's what it is, right?

00:34:40.680 | (laughing)

00:34:41.500 | - That's a good definition.

00:34:42.340 | - But beyond that, it's--

00:34:43.920 | - And they keep coming out with new movies.

00:34:45.280 | - Yes, and they just, right, exactly.

00:34:47.300 | We are driven by this kind of need

00:34:49.700 | to the sort of ineffable quality of who we are.

00:34:52.980 | Which means that the moment you understand something

00:34:55.660 | is no longer AI, right?

00:34:57.140 | Well, we understand this.

00:34:58.060 | That's just, you take the derivative and you divide by two

00:35:00.300 | and then you average it out over time in the window.

00:35:02.500 | So therefore, that's no longer AI.

00:35:03.940 | So the problem is unsolvable

00:35:05.140 | because it keeps kind of going away.

00:35:07.020 | This creates a kind of illusion,

00:35:08.020 | which I don't think is an entire illusion,

00:35:09.440 | of either there's very simple task-based things

00:35:11.340 | you can do very well and over-engineer.

00:35:13.700 | There's all of AI, and there's like nothing in the middle.

00:35:16.560 | Like it's very hard to get from here to here,

00:35:18.740 | and it's very hard to see how to get from here to here.

00:35:21.740 | And I don't think that we've done a very good job of it

00:35:24.220 | because we get stuck trying to solve the small problem

00:35:27.020 | that's in front of us, myself included.

00:35:28.140 | I'm not gonna pretend that I'm better at this

00:35:30.340 | than anyone else.

00:35:31.300 | And of course, all the incentives in academia

00:35:34.340 | and in industry are set to make that very hard

00:35:37.140 | 'cause you have to get the next paper out,

00:35:38.960 | you have to get the next product out,

00:35:40.460 | you have to solve this problem,

00:35:41.820 | and it's very sort of naturally incremental.

00:35:43.740 | And none of the incentives are set up

00:35:47.000 | to allow you to take a huge risk

00:35:48.860 | unless you're already so well-established

00:35:51.120 | you can take that big risk.

00:35:52.460 | And if you're that well-established

00:35:54.880 | that you can take that big risk,

00:35:56.280 | then you've probably spent much of your career

00:35:57.840 | taking these little risks, relatively speaking.

00:35:59.920 | And so you have got a lifetime of experience

00:36:02.720 | telling you not to take that particular big risk, right?

00:36:04.760 | So the whole system's set up to make progress very slow.

00:36:07.240 | That's fine.

00:36:08.180 | It's just the way it is.

00:36:09.320 | But it does make this gap seem really big,

00:36:11.160 | which is my long way of saying

00:36:12.620 | I don't have a great answer to it

00:36:14.020 | except that stop doing n equals one.

00:36:17.820 | At least try to get n equal two and maybe n equal seven

00:36:20.820 | so that you can say I'm gonna,

00:36:22.100 | or maybe T is a better variable here.

00:36:24.060 | I'm gonna not just solve this problem,

00:36:25.500 | I'm gonna solve this problem and another problem.

00:36:26.900 | I'm not gonna learn just on you.

00:36:28.300 | I'm gonna keep living out there in the world

00:36:30.180 | and just seeing what happens

00:36:31.580 | and that we'll learn something as designers

00:36:33.180 | and our machine learning algorithms

00:36:34.900 | and our AI algorithms can learn as well.

00:36:37.060 | But unless you're willing to build a system

00:36:39.700 | which you're gonna have live for months at a time

00:36:42.160 | in an environment that is messy and chaotic

00:36:44.100 | you cannot control,

00:36:45.740 | then you're never going to make progress in that direction.

00:36:48.420 | So I guess my answer to you is yes.

00:36:50.300 | My idea is that you should, it's not no.

00:36:52.380 | It's yes, you should be deploying these things

00:36:55.740 | and making them live for months at a time

00:36:58.740 | and be okay with the fact

00:36:59.940 | that it's gonna take you five years to do this.

00:37:01.780 | Not rerunning the same experiment over and over again

00:37:04.060 | and refining the machine

00:37:05.600 | so it's slightly better at whatever,

00:37:07.340 | but actually having it out there

00:37:08.900 | and living in the chaos of the world

00:37:11.580 | and seeing what it's learning algorithm, say, can learn,

00:37:14.800 | what data structure it can build

00:37:16.060 | and how it can go from there.

00:37:17.360 | Without that, you're gonna be stuck ultimately.

00:37:19.800 | - What do you think about the possibility of N equals one

00:37:23.840 | growing, it's probably a crude approximation,

00:37:27.240 | but growing like if we look at language models like GPT-3,

00:37:31.040 | if you just make it big enough, it'll swallow the world.

00:37:34.360 | Meaning like it'll solve all your T to infinity

00:37:37.680 | by just growing in size of this.

00:37:41.120 | Taking the small over-engineered solution

00:37:43.820 | and just pumping it full of steroids in terms of compute,

00:37:47.300 | in terms of size of training data

00:37:49.820 | and the Yann LeCun style self-supervised

00:37:52.340 | or open AI self-supervised.

00:37:54.240 | Just throw all of YouTube at it

00:37:57.860 | and it will learn how to reason, how to paint,

00:38:01.880 | how to create music, how to love,

00:38:04.300 | all of that by watching YouTube videos.

00:38:06.180 | - I mean, I can't think of a more terrifying world

00:38:08.340 | to live in than a world that is based on YouTube videos,

00:38:10.880 | but yeah, I think the answer,

00:38:13.100 | I just kind of don't think that'll quite,

00:38:15.280 | well, it won't work that easily.

00:38:16.880 | You will get somewhere and you will learn something,

00:38:20.100 | which means it's probably worth it,

00:38:21.700 | but you won't get there.

00:38:23.580 | You won't solve the, you know, here's the thing.

00:38:25.980 | We build these things and we say we want them to learn,

00:38:30.700 | but what actually happens, and let's say they do learn.

00:38:33.420 | I mean, certainly every paper I've gotten published

00:38:35.220 | the things learn, I don't know about anyone else,

00:38:37.500 | but they actually change us, right?

00:38:39.820 | We react to it differently, right?

00:38:41.280 | So we keep redefining what it means to be successful,

00:38:44.280 | both in the negative in the case,

00:38:45.780 | but also in the positive in that,

00:38:47.060 | oh, well, this is an accomplishment.

00:38:50.160 | I'll give you an example,

00:38:51.980 | which is like the one you just described with YouTube.

00:38:53.580 | Let's get completely out of machine learning.

00:38:55.340 | Well, not completely, but mostly out of machine learning.

00:38:57.460 | Think about Google.

00:38:58.420 | People were trying to solve information retrieval,

00:39:02.180 | the ad hoc information retrieval problem forever.

00:39:04.420 | I mean, first major book I ever read about it was what,

00:39:07.600 | '71, I think was when it came out.

00:39:09.760 | Anyway, it's, you know, we'll treat everything as a vector

00:39:12.600 | and we'll do these vector space models and whatever.

00:39:15.360 | And that was all great.

00:39:16.560 | And we made very little progress.

00:39:18.920 | I mean, we made some progress.

00:39:20.160 | And then Google comes and makes the ad hoc problem

00:39:23.680 | seem pretty easy.

00:39:24.560 | I mean, it's not, there's lots of computers

00:39:26.000 | and databases involved, but you know,

00:39:27.960 | and there's some brilliant algorithmic stuff behind it too,

00:39:31.240 | and some systems building.

00:39:32.540 | But the problem changed, right?

00:39:36.540 | If you've got a world that's that connected

00:39:40.100 | so that you have, you know, there are 10 million answers

00:39:43.140 | quite literally to the question that you're asking,

00:39:46.540 | then the problem wasn't give me the things that are

00:39:49.620 | relevant, the problem is don't give me anything

00:39:51.580 | that's irrelevant, at least in the first page,

00:39:53.900 | because nothing else matters.

00:39:56.260 | So Google is not solving the information retrieval problem,

00:39:59.920 | at least not on this webpage.

00:40:01.700 | Google is minimizing false positives,

00:40:05.940 | which is not the same thing as getting an answer.

00:40:08.180 | It turns out it's good enough for what it is

00:40:10.000 | we wanna use Google for, but it also changes

00:40:12.300 | what the problem was we thought we were trying to solve

00:40:14.680 | in the first place.

00:40:15.520 | You thought you were trying to find an answer,

00:40:17.420 | but you're not, we're trying to find the answer,

00:40:19.060 | but it turns out you're just trying to find an answer.

00:40:21.100 | Now, yes, it is true, it's also very good

00:40:22.580 | at finding you exactly that webpage.

00:40:24.220 | Of course, you trained yourself to figure out

00:40:25.980 | what the keywords were to get you that webpage.

00:40:29.060 | But in the end, by having that much data,

00:40:31.980 | you've just changed the problem into something else.

00:40:33.820 | You haven't actually learned what you set out to learn.

00:40:35.620 | Now, the counter to that would be

00:40:37.180 | maybe we're not doing that either, we just think we are,

00:40:40.120 | because we're in our own heads, maybe we're learning

00:40:43.180 | the wrong problem in the first place,

00:40:44.420 | but I don't think that matters.

00:40:45.520 | I think the point is is that Google has not solved

00:40:48.560 | information retrieval, Google has done amazing service.

00:40:51.180 | I have nothing bad to say about what they've done.

00:40:52.820 | Lord knows my entire life is better because Google exists,

00:40:55.820 | in foreign for Google Maps, I don't think I've ever

00:40:57.320 | found this place.

00:40:58.160 | (laughing)

00:40:59.220 | - Where is this?

00:41:00.060 | - Like 95, I see 110 and I see, but where did 95 go?

00:41:04.360 | So I'm very grateful for Google, but they just have

00:41:08.580 | to make certain the first five things are right.

00:41:11.260 | And everything after that is wrong.

00:41:12.780 | Look, we're going off on a totally different topic here,

00:41:15.580 | but think about the way we hire faculty.

00:41:19.780 | It's exactly the same thing.

00:41:21.460 | - Are you getting controversial?

00:41:22.860 | - I'm not getting controversial.

00:41:24.500 | It's exactly the same problem, right?

00:41:27.420 | It's minimizing false positives.

00:41:29.500 | We say things like we want to find the best person

00:41:34.280 | to be an assistant professor at MIT

00:41:36.280 | in the new College of Computing,

00:41:39.280 | which I will point out was founded 30 years after

00:41:42.440 | the College of Computing I'm a part of.

00:41:44.480 | Both of my alma mater--

00:41:46.480 | - No more fighting words.

00:41:47.480 | - I'm just saying I appreciate all that they did

00:41:50.000 | and all that they're doing.

00:41:51.400 | Anyway, so we're gonna try to hire the best person

00:41:57.300 | to be the best professor.

00:41:58.120 | That's what we say, the best person for this job.

00:41:59.620 | But that's not what we do at all, right?

00:42:02.540 | Do you know which percentage of faculty in the top four

00:42:06.020 | earn their PhDs from the top four?

00:42:07.720 | Say in 2017, which is the most recent year

00:42:12.000 | for which I have data.

00:42:12.840 | - Maybe a large percentage.

00:42:14.260 | - About 60%.

00:42:15.100 | - 60.

00:42:15.940 | - 60% of the faculty in the top four earn their PhDs

00:42:18.140 | in the top four.

00:42:18.980 | This is computer science for which there is no top five.

00:42:21.220 | There's only a top four, right?

00:42:22.280 | 'Cause they're all tied for one.

00:42:23.460 | - For people who don't know, by the way,

00:42:24.700 | that would be MIT Stanford, Berkeley, CMU.

00:42:27.220 | - Yep.

00:42:28.060 | - Georgia Tech.

00:42:30.780 | - Number eight.

00:42:31.620 | - Number eight, you're keeping track.

00:42:34.260 | - Oh yes, it's a large part of my job.

00:42:35.700 | Number five is Illinois.

00:42:37.140 | Number six is a tie with UW and Cornell.

00:42:40.560 | And Princeton and Georgia Tech are tied for eight

00:42:42.460 | and UT Austin is number 10.

00:42:43.960 | Michigan's number 11, by the way.

00:42:47.180 | So if you look at the top 10,

00:42:49.020 | you know what percentage of faculty in the top 10

00:42:51.660 | earn their PhDs from the top 10?

00:42:54.100 | 65, roughly, 65%.

00:42:57.120 | If you look at the top 55 ranked departments,

00:43:00.540 | 50% of the faculty earn their PhDs from the top 10.

00:43:04.300 | There's no universe in which all the best faculty,

00:43:09.300 | even just for R1 universities,

00:43:12.420 | the majority of them come from 10 places.

00:43:14.580 | There's just no way that's true,

00:43:15.420 | especially when you consider how small

00:43:16.980 | some of those universities are

00:43:18.140 | in terms of the number of PhDs they produce.

00:43:20.580 | Now, that's not a negative.

00:43:21.980 | I mean, it is a negative.

00:43:23.060 | It also has a habit of entrenching

00:43:24.940 | certain historical inequities and accidents.

00:43:28.640 | But what it tells you is, well, ask yourself the question,

00:43:32.920 | why is it like that?

00:43:36.020 | Well, because it's easier.

00:43:37.320 | If we go all the way back to the 1980s,

00:43:41.160 | you know, there was a saying that, you know,

00:43:42.520 | nobody ever lost his job buying a computer from IBM.

00:43:45.720 | And it was true.

00:43:47.080 | And nobody ever lost their job hiring a PhD from MIT, right?

00:43:51.920 | If the person turned out to be terrible, well, you know,

00:43:54.200 | they came from MIT, what did you expect me to know?

00:43:56.060 | However, that same person coming from,

00:43:58.880 | pick whichever is your least favorite place

00:44:01.900 | that produces PhDs in say, computer science,

00:44:04.320 | well, you took a risk, right?

00:44:06.240 | So all the incentives,

00:44:07.400 | particularly because you're only gonna hire one this year,

00:44:09.240 | well, now we're hiring 10, but you know,

00:44:10.920 | you're only gonna hire one or two or three this year.

00:44:12.920 | And by the way, when they come in,

00:44:14.240 | you're stuck with them for at least seven years

00:44:16.120 | in most places, because that's before you know

00:44:17.600 | whether they're getting tenure or not.

00:44:18.560 | And if they get tenure, you're stuck with them

00:44:19.960 | for a good 30 years, unless they decide to leave.

00:44:22.240 | That means the pressure to get this right is very high.

00:44:23.960 | So what are you gonna do?

00:44:24.800 | You're gonna minimize false positives.

00:44:27.820 | You don't care about saying no inappropriately.

00:44:30.480 | You only care about saying yes inappropriately.

00:44:33.280 | So all the pressure drives you

00:44:34.840 | into that particular direction.

00:44:36.340 | Google, not to put too fine a point on it,

00:44:40.000 | was in exactly the same situation with their search.

00:44:42.000 | It turns out you just don't wanna give people

00:44:43.560 | the wrong page in the first three or four pages.

00:44:47.120 | And if there's 10 million right answers

00:44:48.940 | and a hundred bazillion wrong answers,

00:44:50.680 | just make certain the wrong answers don't get up there.

00:44:52.280 | And who cares if the right answer

00:44:54.160 | was actually the 13th page?

00:44:55.880 | A right answer, a satisficing answer,

00:44:58.440 | is number one, two, three, or four.

00:44:59.800 | So who cares?

00:45:00.680 | - Or an answer that will make you discover

00:45:02.240 | something beautiful, profound to your question.

00:45:05.440 | - Well, that's a different problem, right?

00:45:06.960 | - But isn't that the problem?

00:45:08.240 | Can we linger on this topic

00:45:09.800 | without sort of walking with grace?

00:45:14.800 | How do we get, for hiring faculty,

00:45:19.440 | how do we get that 13th page with a truly special person?

00:45:24.440 | I mean, it depends on the department.

00:45:27.440 | Computer science probably has those kinds of people.

00:45:32.140 | Like you have the Russian guy, Grigori Perelman.

00:45:35.120 | Just these awkward, strange minds

00:45:40.600 | that don't know how to play the little game of etiquette

00:45:44.240 | that faculty have all agreed somehow converged

00:45:48.320 | over the decades how to play with each other.

00:45:50.600 | And also is not, on top of that,

00:45:53.480 | is not from the top four, top whatever numbers, the schools.

00:45:57.520 | And maybe actually just says a few every once in a while

00:46:01.080 | to the traditions of old

00:46:04.400 | within the computer science community.

00:46:06.000 | Maybe talks trash about machine learning

00:46:09.200 | is a total waste of time.

00:46:10.680 | And that's there on their resume.

00:46:13.760 | So how do you allow the system

00:46:17.680 | to give those folks a chance?

00:46:19.680 | - Well, you have to be willing to take a certain kind of,

00:46:21.000 | without taking a particular position

00:46:22.200 | on any particular person,

00:46:23.120 | you'd have to take, you have to be willing to take risk.

00:46:26.640 | A small amount of risk.

00:46:27.480 | I mean, if we were treating this as a,

00:46:29.640 | well, as a machine learning problem,

00:46:31.480 | as a search problem, which is what it is,

00:46:32.760 | it's a search problem.

00:46:33.600 | If we were treating it that way,

00:46:34.640 | you would say, oh, well, the main thing is you want,

00:46:36.560 | you know, you've got a prior,

00:46:38.080 | you want some data 'cause I'm Bayesian.

00:46:40.120 | If you don't wanna do it that way,

00:46:41.280 | we'll just inject some randomness in and it'll be okay.

00:46:44.760 | The problem is that feels very, very hard to do with people.

00:46:48.600 | All the incentives are wrong there,

00:46:51.800 | but it turns out, and let's say,

00:46:53.760 | let's say that's the right answer.

00:46:54.640 | Let's just give, for the sake of argument,

00:46:56.440 | that, you know, injecting randomness in the system

00:46:58.600 | at that level for who you hire is just not worth doing

00:47:01.840 | because the price is too high or the cost is too high.

00:47:04.320 | We had infinite resources, sure, but we don't.

00:47:05.760 | And also you've gotta teach people.

00:47:07.000 | So, you know, you're ruining other people's lives

00:47:08.800 | if you get it too wrong.

00:47:10.840 | But we've taken that principle, even if I grant it,

00:47:14.400 | and pushed it all the way back, right?

00:47:17.240 | So we could have a better pool than we have

00:47:22.240 | of people we look at and give an opportunity to.

00:47:24.960 | If we do that, then we have a better chance of finding that.

00:47:28.320 | Of course, that just pushes the problem back

00:47:30.280 | another level, but let me tell you something else.

00:47:32.200 | You know, I did a sort of study, I call it a study.

00:47:35.080 | I called up eight of my friends and asked them

00:47:36.400 | for all of their data for graduate admissions,

00:47:38.280 | but then someone else followed up and did an actual study.

00:47:41.320 | And it turns out that I can tell you

00:47:43.360 | how everybody gets into grad school, more or less.

00:47:46.440 | More or less.

00:47:48.060 | You basically admit everyone from places

00:47:49.640 | higher ranked than you.

00:47:50.720 | You admit most people from places ranked around you,

00:47:52.600 | and you admit almost no one from places ranked below you,

00:47:54.600 | with the exception of the small liberal arts colleges

00:47:56.560 | that aren't ranked at all, like Harvey Mudd,

00:47:58.160 | 'cause they don't, they don't have PhDs,

00:47:59.280 | so they aren't ranked.

00:48:00.320 | This is all CS.

00:48:01.160 | Which means the decision of whether, you know,

00:48:06.200 | you become a professor at Cornell

00:48:09.360 | was determined when you were 17, right?

00:48:11.760 | By what you knew to go to undergrad to do whatever, right?

00:48:15.720 | So if we can push these things back a little bit

00:48:17.920 | and just make the pool a little bit bigger,

00:48:19.440 | at least you raise the probability

00:48:20.680 | that you will be able to see someone interesting

00:48:23.120 | and take the risk.

00:48:25.320 | The other answer to that question, by the way,

00:48:28.200 | which you could argue is the same as,

00:48:30.000 | you either adjust the pool so the probabilities go up,

00:48:32.520 | that's a way of injecting a little bit of uniformity.

00:48:36.040 | Uniform noise in the system, as it were,

00:48:38.080 | is you change your loss function.

00:48:40.240 | You just let yourself be measured by something

00:48:42.840 | other than whatever it is

00:48:44.200 | that we're measuring ourselves by now.

00:48:46.640 | I mean, US News and World Report,

00:48:48.680 | every time they change their formula

00:48:50.600 | for determining rankings,

00:48:51.840 | move entire universities to behave differently,

00:48:54.860 | because rankings matter.

00:48:56.440 | - Can you talk trash about those rankings for a second?

00:48:59.520 | No, I'm joking about talking trash.

00:49:01.280 | I actually, it's so funny how, from my perspective,

00:49:05.400 | from a very shallow perspective,

00:49:07.320 | how dogmatic, like how much I trust those rankings.

00:49:11.560 | They're almost ingrained in my head.

00:49:14.040 | I mean, at MIT, everybody kind of,

00:49:16.820 | it's a propagated, mutually agreed upon idea

00:49:22.880 | that those rankings matter.

00:49:25.080 | And I don't think anyone knows what they're,

00:49:27.040 | like most people don't know what they're based on.

00:49:30.320 | And what are they exactly based on

00:49:33.040 | and what are the flaws in that?

00:49:34.600 | - Well, so it depends on which rankings you're talking about.

00:49:38.200 | Do you wanna talk about computer science

00:49:39.320 | or you wanna talk about universities?

00:49:40.280 | - Computer science, US News, isn't that the main one?

00:49:43.480 | - Yeah, it's US News.

00:49:44.320 | The only one that matters is US News.

00:49:45.440 | Nothing else matters.

00:49:46.480 | Sorry, csrankings.org, but nothing else matters

00:49:48.800 | but US News.

00:49:50.200 | So US News has formula that it uses for many things,

00:49:54.320 | but not for computer science,

00:49:55.280 | because computer science is considered a science,

00:49:57.600 | which is absurd.

00:49:58.480 | So the rankings for computer science is 100% reputation.

00:50:04.480 | So two people at each department,

00:50:09.160 | it's not really a department, whatever,

00:50:10.280 | at each department basically rank everybody.

00:50:14.000 | Slightly more complicated than that.

00:50:15.320 | But whatever, they rank everyone.

00:50:16.960 | And then those things are put together

00:50:18.240 | and somehow-- - Oh no.

00:50:20.000 | So that means, how do you improve reputation?

00:50:22.120 | How do you move up and down the space of reputation?

00:50:25.760 | - Yes, that's exactly the question.

00:50:27.320 | - Twitter?

00:50:28.160 | (Dave laughs)

00:50:29.000 | - It can help.

00:50:29.840 | I can tell you how Georgia Tech did it,

00:50:31.960 | or at least how I think Georgia Tech did it,

00:50:33.120 | because Georgia Tech is actually the case to look at.

00:50:36.360 | Not just because I'm at Georgia Tech,

00:50:37.640 | but because Georgia Tech is the only computing unit

00:50:40.440 | that was not in the top 20 that has made it into the top 10.

00:50:43.120 | It's also the only one in the last two decades, I think,

00:50:46.720 | that moved up in the top 10,

00:50:50.960 | as opposed to having someone else move down.

00:50:53.280 | So we used to be number 10,

00:50:54.920 | and then we became number nine,

00:50:56.160 | because UT Austin went down slightly

00:50:58.480 | and now we were tied for ninth,

00:50:59.560 | 'cause that's how rankings work.

00:51:01.400 | And we moved from nine to eight,

00:51:03.040 | because our raw score moved up a point.

00:51:06.680 | So Georgia, something about Georgia Tech,

00:51:09.360 | computer science, or computing anyway.

00:51:11.560 | I think it's because we have shown leadership

00:51:15.640 | at every crisis level, right?

00:51:17.000 | So we created a college, first public university to do it,

00:51:19.200 | second college, second university to do it

00:51:21.200 | after CMU is number one.

00:51:22.320 | I also think it's no accident that CMU is the largest,

00:51:26.000 | and we're, depending upon how you count

00:51:27.800 | and depending on exactly where MIT ends up

00:51:29.640 | with its final college of computing,

00:51:31.200 | second or third largest.

00:51:32.160 | I don't think that's an accident.

00:51:33.080 | We've been doing this for a long time.

00:51:34.600 | But in the 2000s,

00:51:36.040 | when there was a crisis about undergraduate education,

00:51:39.400 | Georgia Tech took a big risk

00:51:41.000 | and succeeded at rethinking undergrad education

00:51:43.680 | and computing.

00:51:45.160 | I think we created these schools at a time

00:51:47.840 | when most public universities anyway were afraid to do it.

00:51:50.320 | We did the online masters,

00:51:51.820 | and that mattered because people were trying to figure out

00:51:55.880 | what to do with MOOCs and so on.

00:51:57.200 | I think it's about being observed by your peers

00:52:00.760 | and having an impact.

00:52:02.200 | So, I mean, that is what reputation is, right?

00:52:04.240 | So the way you move up in the reputation rankings

00:52:07.140 | is by doing something that makes people turn

00:52:08.960 | and look at you and say, "That's good.

00:52:11.640 | They're better than I thought."

00:52:13.600 | - Yeah. - Beyond that,

00:52:14.440 | it's just inertia.

00:52:15.260 | And there's huge hysteria in the system, right?

00:52:16.800 | Like, I mean, there was these,

00:52:18.280 | I can't remember this, this may be apocryphal,

00:52:19.760 | but there's a major or a department

00:52:23.360 | that like MIT was ranked number one in,

00:52:25.600 | and they didn't have it.

00:52:27.120 | Right, it's just about what you,

00:52:28.560 | I don't know if that's true,

00:52:29.400 | but someone said that to me anyway.

00:52:31.120 | But it's a thing, right?

00:52:33.160 | It's all about reputation.

00:52:34.240 | Of course MIT is great because MIT is great.

00:52:36.520 | It's always been great.

00:52:37.340 | By the way, because MIT is great,

00:52:39.280 | the best students come, which keeps it being great.

00:52:42.960 | I mean, it's just a positive feedback loop.

00:52:44.840 | It's not surprising.

00:52:45.680 | I don't think it's wrong.

00:52:47.240 | - Yeah, but it's almost like a narrative.

00:52:48.920 | Like, it doesn't actually have to be backed by reality.

00:52:52.080 | And it's, you know, not to say anything about MIT,

00:52:54.480 | but like, it does feel like we're playing

00:52:59.320 | in the space of narratives,

00:53:00.840 | not the space of something grounded in,

00:53:03.840 | like one of the surprising things when I showed up at MIT

00:53:06.280 | and just all the students I've worked with

00:53:08.760 | and all the research I've done is it like,

00:53:13.360 | they're the same people as I've met other places.

00:53:16.640 | - I mean, what MIT is going for,

00:53:20.440 | well, MIT has many things going for it,

00:53:21.280 | but one of the things MIT is going for is--

00:53:23.120 | - Nice logo.

00:53:24.000 | - It has a nice logo.

00:53:25.000 | It's a lot better than it was when I was here.

00:53:28.120 | Nice colors too.

00:53:29.200 | Terrible, terrible name for a mascot.

00:53:31.400 | But the thing that MIT has going for it

00:53:34.460 | is it really does get the best students.

00:53:36.520 | It just doesn't get all of the best students.

00:53:39.040 | There are many more best students out there, right?

00:53:41.480 | And the best students want to be here

00:53:42.800 | 'cause it's the best place to be

00:53:44.720 | or one of the best places to be.

00:53:45.800 | And it just kind of, it's a sort of positive feedback.

00:53:47.880 | But you said something earlier,

00:53:49.380 | which I think is worth examining for a moment, right?

00:53:52.820 | You said it's, I forget the word you used.

00:53:54.800 | You said, "We're living in the space of narrative

00:53:58.080 | "as opposed to something objective."

00:54:00.080 | Narrative is objective.

00:54:01.260 | I mean, one could argue that the only thing

00:54:03.400 | that we do as humans is narrative.

00:54:04.920 | We just build stories to explain why we do what we do.

00:54:08.060 | Someone once asked me, "But wait, there's nothing objective."

00:54:10.240 | No, it's completely an objective measure.

00:54:12.920 | It's an objective measure of the opinions of everybody else.

00:54:16.340 | Now, is that physics?

00:54:19.480 | I don't know.

00:54:20.600 | But, you know, what, I mean, tell me something

00:54:23.180 | you think is actually objective and measurable

00:54:24.820 | in a way that makes sense.

00:54:25.780 | Like cameras, they don't, do you know that,

00:54:29.900 | I mean, you're getting me off on something here,

00:54:31.180 | but do you know that cameras,

00:54:34.900 | which are just reflecting light and putting them on film,

00:54:37.780 | like did not work for dark-skinned people

00:54:40.300 | until like the 1970s?

00:54:42.800 | You know why?

00:54:43.700 | Because you were building cameras

00:54:45.340 | for the people who were gonna buy cameras,

00:54:47.260 | who all, at least in the United States

00:54:49.100 | and Western Europe, were relatively light-skinned.

00:54:52.140 | Turns out it took terrible pictures

00:54:53.440 | of people who look like me.

00:54:55.620 | That got fixed with better film and whole processes.

00:54:58.700 | Do you know why?

00:55:00.240 | Because furniture manufacturers

00:55:02.980 | wanted to be able to take pictures of mahogany furniture.

00:55:06.340 | Right, because candy manufacturers

00:55:08.540 | wanted to be able to take pictures of chocolate.

00:55:11.780 | Now, the reason I bring that up

00:55:13.660 | is because you might think that cameras--

00:55:17.900 | - Are objective.

00:55:18.740 | - Are objective, they're just capturing light.

00:55:20.140 | No, they're made, they are doing the things

00:55:23.000 | that they are doing based upon decisions

00:55:24.780 | by real human beings to privilege,

00:55:27.300 | if I may use that word, some physics over others,

00:55:29.900 | because it's an engineering problem,

00:55:31.260 | there are trade-offs, right?

00:55:32.260 | So I can either worry about this part of the spectrum

00:55:34.740 | or this part of the spectrum.

00:55:35.920 | This costs more, that costs less,

00:55:37.340 | this costs the same,

00:55:38.300 | but I have more people paying money over here, right?

00:55:40.380 | And it turns out that if a giant conglomerate

00:55:43.860 | wants, demands that you do something different

00:55:45.900 | and it's gonna involve all kinds of money for you,

00:55:47.700 | suddenly the trade-offs change, right?

00:55:49.460 | And so there you go.

00:55:51.020 | I actually don't know how I ended up there.

00:55:52.220 | Oh, it's because of this notion of objectiveness, right?

00:55:54.020 | So even the objective isn't objective

00:55:57.060 | because at the end you've gotta tell a story,

00:55:58.460 | you've gotta make decisions,

00:55:59.300 | you've gotta make trade-off,

00:56:00.140 | and what else is engineering other than that?

00:56:01.900 | So I think that the rankings capture something.

00:56:05.340 | They just don't necessarily capture

00:56:07.820 | what people assume they capture.

00:56:09.860 | - Just to linger on this idea,

00:56:15.300 | why is there not more people who just play

00:56:20.020 | with whatever that narrative is, have fun with it,

00:56:22.380 | have like excite the world,

00:56:23.940 | whether it's in the Carl Sagan style

00:56:26.540 | of that calm, sexy voice of explaining the stars

00:56:30.460 | and all the romantic stuff,

00:56:31.660 | or the Elon Musk, dare I even say Donald Trump,

00:56:35.180 | where you're like trolling and shaking up the system

00:56:37.560 | and just saying controversial things.

00:56:39.560 | I talked to Lisa Feldman Barrett,

00:56:43.100 | who's a neuroscientist who just enjoys

00:56:45.820 | playing the controversy,

00:56:47.580 | thinks like, finds the counterintuitive ideas

00:56:51.700 | in the particular science and throws them out there

00:56:54.820 | and sees how they play in the public discourse.

00:56:57.660 | Like why don't we see more of that?

00:57:00.420 | And why doesn't academia attract an Elon Musk type?

00:57:03.620 | - Well, tenure is a powerful thing

00:57:06.480 | that allows you to do whatever you want,

00:57:08.140 | but getting tenure typically requires

00:57:10.660 | you to be relatively narrow, right?

00:57:13.220 | Because people are judging you.

00:57:14.420 | Well, I think the answer is we have told ourselves a story,

00:57:18.660 | a narrative, that that is vulgar,

00:57:22.220 | which you just described as vulgar.

00:57:23.900 | It's certainly unscientific, right?

00:57:26.620 | And it is easy to convince yourself

00:57:31.620 | that in some ways you're the mathematician, right?

00:57:35.100 | The fewer there are in your major,

00:57:38.120 | the more that proves your purity, right?

00:57:40.540 | - Yeah.

00:57:41.380 | - So once you tell yourself that story,

00:57:44.180 | then it is beneath you to do that kind of thing, right?

00:57:49.180 | I think that's wrong.

00:57:51.340 | I think that, and by the way, everyone doesn't have to do this.

00:57:53.380 | Everyone's not good at it,

00:57:54.220 | and everyone, even if they would be good at it,

00:57:55.260 | would enjoy it.

00:57:56.340 | So it's fine.

00:57:57.700 | But I do think you need some diversity

00:57:59.740 | in the way that people choose to relate

00:58:03.060 | to the world as academics,

00:58:05.620 | because I think the great universities

00:58:08.900 | are ones that engage with the rest of the world.

00:58:12.260 | It is a home for public intellectuals.

00:58:14.940 | And in 2020, being a public intellectual

00:58:18.260 | probably means being on Twitter,

00:58:20.700 | whereas of course that wasn't true 20 years ago,

00:58:23.140 | 'cause Twitter wasn't around 20 years ago.

00:58:25.300 | And if it was, it wasn't around in a meaningful way.

00:58:26.660 | I don't actually know how long Twitter's been around.

00:58:28.740 | As I get older, I find that my notion of time

00:58:31.980 | has gotten worse and worse.

00:58:33.100 | Like Google really has been around that long?

00:58:34.540 | Anyway, the point is that I think that

00:58:38.140 | I think that we sometimes forget

00:58:42.180 | that a part of our job is to impact the people

00:58:44.900 | who aren't in the world that we're in,

00:58:46.380 | and that that's the point of being at a great place

00:58:48.500 | and being a great person, frankly.

00:58:50.340 | - There's an interesting force

00:58:51.660 | in terms of public intellectuals.

00:58:53.660 | Forget Twitter, we could look at just online courses

00:58:56.740 | that are public-facing in some part.

00:58:59.220 | Like there is a kind of force that pulls you back.

00:59:05.860 | I would, let me just call it out

00:59:07.380 | 'cause I don't give a damn at this point.

00:59:09.740 | There's a little bit of, all of us have this,

00:59:11.780 | but certainly faculty have this, which is jealousy.

00:59:15.100 | It's whoever's popular at being a good communicator,

00:59:20.100 | exciting the world with their science.

00:59:23.020 | And of course, when you excite the world with the science,

00:59:27.480 | it's not peer-reviewed, clean.

00:59:30.480 | It all sounds like bullshit.

00:59:33.740 | It's like a TED Talk.

00:59:35.540 | And people roll their eyes,

00:59:37.100 | and they hate that a TED Talk gets millions of views

00:59:40.360 | or something like that.

00:59:41.220 | And then everybody pulls each other back.

00:59:43.300 | There's this force that just kind of,

00:59:45.500 | it's hard to stand out

00:59:46.940 | unless you like win a Nobel Prize or whatever.

00:59:48.740 | Like it's only when you like get senior enough

00:59:51.940 | where you just stop giving a damn.

00:59:54.940 | But just like you said, even when you get tenure,

00:59:57.900 | that was always the surprising thing to me.

01:00:00.380 | I have many colleagues and friends who have gotten tenure,

01:00:04.260 | but there's not a switch.

01:00:06.600 | There's not an F-you money switch

01:00:11.220 | where you're like, you know what?

01:00:13.600 | Now I'm going to be more bold.

01:00:15.300 | It doesn't, I don't see it.

01:00:17.220 | - Well, there's a reason for that.

01:00:18.860 | Tenure isn't a test.

01:00:19.780 | It's a training process.

01:00:20.980 | It teaches you to behave in a certain way,

01:00:23.560 | to think in a certain way, to accept certain values,

01:00:26.200 | and to react accordingly.

01:00:27.940 | And the better you are at that,

01:00:29.260 | the more likely you are to earn tenure.

01:00:30.880 | And by the way, this is not a bad thing.

01:00:32.620 | Most things are like that.

01:00:34.700 | And I think most of my colleagues

01:00:36.400 | are interested in doing great work,

01:00:38.340 | and they're just having impact

01:00:39.380 | in the way that they want to have impact.

01:00:41.340 | I do think that as a field, not just as a field,

01:00:45.420 | as a profession, we have a habit of belittling those

01:00:50.420 | who are popular, as it were,

01:00:55.680 | as if the word itself is a kind of scarlet A, right?

01:00:58.980 | I think it's easy to convince yourself,

01:01:03.980 | and no one is immune to this,

01:01:07.820 | that the people who are better known

01:01:09.540 | are better known for bad reasons.

01:01:11.740 | The people who are out there dumbing it down

01:01:15.260 | are not being pure to whatever the values and ethos is

01:01:19.900 | for your field.

01:01:20.740 | And it's just very easy to do.

01:01:22.900 | Now, having said that, I think that ultimately,

01:01:26.580 | people who are able to be popular and out there

01:01:30.260 | and are touching the world and making a difference,

01:01:32.760 | our colleagues do, in fact,

01:01:35.340 | appreciate that in the long run.

01:01:37.260 | It's just, you have to be very good at it,

01:01:39.380 | or you have to be very interested in pursuing it.

01:01:41.660 | And once you get past a certain level,

01:01:42.940 | I think people accept that for who it is.

01:01:44.820 | I mean, I don't know.

01:01:45.860 | I'd be really interested in how Rod Brooks felt

01:01:47.700 | about how people were interacting with him

01:01:50.500 | when he did "Fast, Cheap, and Out of Control"

01:01:52.860 | way, way, way back when.

01:01:54.300 | - What's "Fast, Cheap, and Out of Control"?

01:01:56.420 | - It was a documentary that involved four people.

01:01:59.820 | I remember nothing about it other than Rod Brooks was in it

01:02:01.980 | and something about naked mole rats.

01:02:04.820 | Can't remember what the other two things were.

01:02:06.460 | It was robots, naked mole rats, and then two other--

01:02:08.580 | - By the way, Rod Brooks used to be the head

01:02:10.260 | of the Artificial Intelligence Laboratory at MIT,

01:02:13.460 | and then launched, I think, iRobot,

01:02:16.860 | and then Think Robotics, Rethink Robotics?

01:02:19.380 | - Yes, sir, yes.

01:02:21.140 | - Think is in the word.

01:02:23.380 | And also is a little bit of a rock star personality

01:02:27.180 | in the AI world, very opinionated, very intelligent.

01:02:30.260 | Anyway, sorry, mole rats and naked.

01:02:32.580 | - Naked mole rats.

01:02:33.420 | Also, he was one of my two advisors for my PhD.

01:02:37.140 | - This explains a lot.

01:02:38.220 | (laughing)

01:02:39.060 | - I don't know how it explains.

01:02:40.940 | I love Rod.

01:02:41.780 | But I also love my other advisor, Paul.

01:02:43.740 | Paul, if you're listening, I love you too.

01:02:45.140 | Both very, very different people.

01:02:46.700 | - Paul Viola.

01:02:47.540 | - Paul Viola, both very interesting people,

01:02:48.820 | very different in many ways.

01:02:50.900 | But I don't know what Rod would say to you

01:02:52.860 | about what the reaction was.

01:02:54.980 | I know that for the students at the time,

01:02:56.980 | 'cause I was a student at the time, it was amazing.

01:02:59.500 | This guy was in a movie, being very much himself.

01:03:04.340 | Actually, the movie version of him

01:03:06.260 | is a little bit more Rod than Rod.

01:03:09.620 | I mean, I think they edited it appropriately for him.

01:03:12.860 | But it was very much Rod, and he did all this

01:03:14.700 | while doing great work.

01:03:15.700 | I mean, he was running, was he running the iLab

01:03:17.340 | at that point or not?

01:03:18.180 | I don't know.

01:03:19.000 | I think the iLab would be soon.

01:03:21.080 | He was a giant in the field.

01:03:22.040 | He did amazing things, made a lot of his bones

01:03:23.720 | by doing the kind of counterintuitive science, right?

01:03:27.760 | And saying, no, you're doing this all wrong.

01:03:29.680 | Representation is crazy.

01:03:30.800 | The world is your own representation.

01:03:32.120 | You just react to it.

01:03:32.960 | I mean, he did amazing things,

01:03:34.080 | and continues to do those sorts of things

01:03:36.480 | as he's moved on.

01:03:37.880 | I have, I think he might tell you,

01:03:40.880 | I don't know if he would tell you it was good or bad,

01:03:42.080 | but I know that for everyone else out there in the world,

01:03:45.480 | it was a good thing,

01:03:46.320 | and certainly he continued to be respected.

01:03:48.340 | So it's not as if it destroyed his career by being popular.

01:03:52.080 | - All right, let's go into a topic where I'm on thin ice,

01:03:56.280 | because I grew up in the Soviet Union and Russia.

01:03:58.080 | My knowledge of music, this American thing you guys do,

01:04:02.160 | is quite foreign.

01:04:07.160 | So your research group is called,

01:04:10.960 | as we've talked about,

01:04:12.180 | the Lab for Interactive Artificial Intelligence,

01:04:14.660 | but also there's just a bunch of mystery around this.

01:04:17.540 | My research fails me.

01:04:19.200 | Also called PFUNK.

01:04:21.220 | P stands for probabilistic.

01:04:24.020 | And what does FUNK stand for?

01:04:27.880 | - So a lot of my life is about making acronyms.

01:04:30.440 | So if I have one quirk, it's that people will say words,

01:04:32.880 | and I see if they make acronyms.

01:04:34.600 | And if they do, then I'm happy,

01:04:36.920 | and then if they don't, I try to change it

01:04:38.760 | so that they make acronyms.

01:04:39.760 | It's just a thing that I do.

01:04:41.640 | So PFUNK is an acronym.

01:04:43.240 | It has three or four different meanings.

01:04:45.840 | But finally I decided that the P stands for probabilistic

01:04:48.480 | because at the end of the day,

01:04:49.880 | it's machine learning and it's randomness

01:04:51.440 | and it's uncertainty, which is the important thing here.

01:04:54.300 | And the FUNK can be lots of different things,

01:04:56.580 | but I decided I should leave it up to the individual

01:04:58.640 | to figure out exactly what it is.

01:05:00.680 | But I will tell you that when my students graduate,

01:05:03.600 | when they get out, as we say, at Tech,

01:05:06.460 | I hand them, they put on a hat and star glasses

01:05:11.660 | and a medallion from the PFUNK era,

01:05:15.820 | and we take a picture,

01:05:17.380 | and I hand them a pair of fuzzy dice,

01:05:21.040 | which they get to keep.

01:05:22.320 | - So there's a sense to it which is not an acronym,

01:05:25.660 | like literally FUNK.

01:05:27.160 | You have a dark, mysterious past.

01:05:30.780 | Oh, it's not dark, it's just fun,

01:05:34.460 | as in hip hop and funk.

01:05:38.240 | - Yep.

01:05:39.480 | - So can you educate a Soviet-born Russian

01:05:43.400 | about this thing called hip hop?

01:05:45.840 | Like if you were to give me,

01:05:48.680 | like if we went on a journey together

01:05:50.860 | and you were trying to educate me about,

01:05:53.020 | especially the past couple of decades in the '90s

01:05:58.020 | about hip hop or funk,

01:05:59.800 | what records or artists would you introduce me to?

01:06:04.800 | Would you tell me about,

01:06:09.740 | or maybe what influenced you in your journey,

01:06:12.320 | or what you just love?

01:06:13.520 | Like when the family's gone and you just sit back

01:06:16.600 | and just blast some stuff these days,

01:06:19.440 | what do you listen to?

01:06:20.440 | - Well, so I listen to a lot,

01:06:21.640 | but I will tell you, well, first off,

01:06:22.960 | all great music was made when I was 14,

01:06:24.800 | and that statement is true for all people,

01:06:26.520 | no matter how old they are or where they live.

01:06:28.560 | But for me, the first thing that's worth pointing out

01:06:31.420 | is that hip hop and rap aren't the same thing,

01:06:33.820 | so depending on who you talk to about this,

01:06:35.080 | and there are people who feel very strongly about this,

01:06:37.960 | much more strongly than I do.

01:06:38.800 | - You're offending everybody in this conversation,

01:06:40.440 | so this is great, let's keep going.

01:06:41.840 | - Hip hop is a culture.

01:06:44.040 | - Yeah, I take that.

01:06:44.880 | - It's a whole set of things, of which rap is a part.

01:06:46.680 | So tagging is a part of hip hop.

01:06:48.320 | I don't know why that's true,

01:06:49.160 | but people tell me it's true,

01:06:50.080 | and I'm willing to go along with it,

01:06:51.280 | 'cause they get very angry about it.

01:06:52.600 | But hip hop is-- - Tagging is like graffiti.

01:06:54.040 | - Tagging is like graffiti.

01:06:55.860 | And there's all these, including the popping and the locking

01:06:57.880 | and all the dancing and all those things,

01:06:59.040 | that's all a part of hip hop.

01:07:00.400 | It's a way of life, which I think is true.

01:07:03.040 | And then there's rap, which is this particular--

01:07:05.240 | - It's the music part.

01:07:06.080 | - Yes, or a music part.

01:07:07.800 | - A music part, yeah.

01:07:09.000 | I mean, you wouldn't call the stuff that DJs do

01:07:10.600 | the (imitates scratching) scratching.

01:07:12.000 | That's not rap, right?

01:07:12.920 | But it's a part of hip hop, right?

01:07:14.280 | So given that we understand that hip hop is this whole thing,

01:07:17.480 | what are the rap albums that best touch that for me?

01:07:20.360 | Well, if I were gonna educate you,

01:07:22.040 | I would try to figure out what you liked,

01:07:23.280 | and then I would work you there.

01:07:24.840 | - Leonard Skinner.

01:07:26.640 | - Oh my God.

01:07:27.480 | - Yes. (laughs)

01:07:28.320 | - I would probably start with--

01:07:29.800 | (Leonard laughs)

01:07:32.400 | - Led Zeppelin.

01:07:33.640 | - There's a fascinating, oh, it's okay.

01:07:35.080 | There's a fascinating exercise one can do

01:07:37.320 | by watching old episodes of "I Love the '70s,"

01:07:40.960 | "I Love the '80s," "I Love the '90s"

01:07:43.160 | with a bunch of friends,

01:07:44.200 | and just see where people come in and out of pop culture.

01:07:47.280 | So if you're talking about those people,

01:07:51.080 | then I would actually start you with

01:07:53.680 | where I would hope to start you with anyway,

01:07:55.120 | which is "Public Enemy."

01:07:56.680 | Particularly, it takes a nation of millions to hold us back,

01:07:59.840 | which is clearly the best album ever produced,

01:08:03.560 | and certainly the best hip hop album ever produced,

01:08:06.120 | in part because it was so much

01:08:08.280 | of what was great about the time.

01:08:09.920 | Fantastic lyrics, 'cause to me, it's all about the lyrics.

01:08:12.720 | Amazing music that was coming from,

01:08:14.360 | Rick Rubin was the producer of that,

01:08:16.520 | and he did a lot, very kind of heavy metal-ish,

01:08:19.080 | at least in the '80s sense at the time,

01:08:21.840 | and it was focused on politics in the 1980s,

01:08:26.240 | which was what made hip hop so great then.

01:08:28.320 | I would start you there,

01:08:29.520 | then I would move you up through things

01:08:31.560 | that have been happening more recently.

01:08:33.280 | I'd probably get you to someone like a Mos Def.

01:08:35.280 | I would give you a history lesson, basically.

01:08:37.080 | Mos Def's amazing.

01:08:37.920 | - So he hosted a poetry jam thing on HBO

01:08:40.120 | or something like that?

01:08:41.080 | - Probably, I don't think I've seen it,

01:08:41.920 | but I wouldn't be surprised.

01:08:43.000 | - Yeah, spoken poetry, that guy.

01:08:44.200 | - Yeah, he's amazing.

01:08:45.640 | He's amazing.

01:08:46.880 | And then, after I got you there,

01:08:48.920 | I'd work you back to EPMD,

01:08:50.640 | and eventually, I would take you back to "The Last Poets,"

01:08:56.000 | and particularly their first album, "The Last Poets,"

01:08:58.080 | which was 1970, to give you a sense of history,

01:09:01.400 | and that it actually has been building up

01:09:02.960 | over a very, very long time.

01:09:05.160 | So we would start there,

01:09:06.840 | because that's where your music aligns,

01:09:08.480 | and then we would cycle out,

01:09:09.920 | and I'd move you to the present,

01:09:11.000 | and then I'd take you back to the past.

01:09:12.680 | Because I think a large part of people

01:09:14.280 | who are kind of confused about any kind of music,

01:09:17.040 | you know, the truth is, this is the same thing

01:09:18.680 | we've always been talking about, right?

01:09:19.720 | It's about narrative and being a part of something

01:09:21.400 | and being immersed in something,

01:09:22.960 | so you understand it, right?

01:09:24.560 | Jazz, which I also like,

01:09:27.640 | is, one of the things that's cool about jazz

01:09:29.880 | is that you come and you meet someone

01:09:30.880 | who's talking to you about jazz,

01:09:31.960 | and you have no idea what they're talking about.

01:09:33.480 | And then one day it all clicks,

01:09:35.440 | and you've been so immersed in it,

01:09:36.600 | you go, "Oh yeah, that's a Charlie Parker,"

01:09:38.640 | you start using words that nobody else understands, right?

01:09:40.520 | And it becomes part of, hip-hop's the same way,

01:09:42.680 | everything's the same way, they're all cultural artifacts.

01:09:44.860 | But I would help you to see that there's a history of it,

01:09:46.920 | and how it connects to other genres of music

01:09:48.840 | that you might like, to bring you in,

01:09:50.680 | so that you could kind of see how it connects

01:09:53.680 | to what you already like,

01:09:54.520 | including some of the good work that's been done

01:09:58.640 | with fusions of hip-hop and bluegrass.

01:10:02.640 | - Oh no. - Yes.

01:10:04.040 | Some of it's even good.

01:10:07.000 | Not all of it, but some of it is good.

01:10:09.720 | But I'd start you with "It Takes a Nation to Make Us All"

01:10:11.720 | as a back.

01:10:12.560 | - There's an interesting tradition in more modern hip-hop

01:10:16.000 | of integrating almost like classic rock songs or whatever,

01:10:20.000 | like integrating into their music,

01:10:23.820 | into the beat, into the whatever.

01:10:25.580 | It's kind of interesting.

01:10:26.520 | It gives a whole new, not just classic rock,

01:10:30.360 | but what is it, Kanye, Gold Digger, the--

01:10:33.800 | - Mm-hmm, old R&B.

01:10:36.200 | - Taking and pulling old R&B, right.

01:10:38.200 | - Well, that's been true since the beginning.

01:10:39.640 | I mean, in fact, that's in some ways,

01:10:41.360 | that's why the DJ used to get top billing,

01:10:45.480 | 'cause it was the DJ that brought all the records together

01:10:47.760 | and made it work so that people could dance.

01:10:49.440 | If you go back to those days, mostly in New York,

01:10:53.040 | though not exclusively, but mostly in New York,

01:10:54.660 | where it sort of came out of,

01:10:56.600 | it was the DJ that brought all the music together

01:10:58.080 | and the beats and showed that basically music

01:11:00.280 | is itself an instrument, very meta,

01:11:02.360 | and you can bring it together

01:11:03.400 | and then you sort of wrap over it and so on,

01:11:05.120 | and it sort of moved that way.

01:11:07.000 | So that's going way, way back.

01:11:07.960 | Now, in the period of time where I grew up,

01:11:10.400 | when I became really into it, which was most of the '80s,

01:11:14.600 | it was more funk was the back for a lot of the stuff,

01:11:17.960 | public enemy at that time notwithstanding,

01:11:21.240 | and so, which is very nice,

01:11:22.960 | 'cause it tied into what my parents listened to

01:11:25.200 | and what I vaguely remember listening to

01:11:26.600 | when I was very small.

01:11:28.160 | And by the way, complete revival of George Clinton

01:11:32.080 | and Parliament and Funkadelic and all of those things

01:11:34.560 | to bring it sort of back into the '80s and into the '90s.

01:11:37.280 | And as we go on, you're gonna see the last decade

01:11:40.600 | and the decade before that being brought in.

01:11:42.460 | And when you don't think

01:11:43.600 | that you're hearing something you've heard,

01:11:44.760 | it's probably because it's being sampled by someone

01:11:47.560 | who referring to something they remembered

01:11:52.560 | when they were young,

01:11:54.120 | perhaps from somewhere else altogether,

01:11:55.880 | and you just didn't realize what it was

01:11:58.200 | because it wasn't a popular song

01:11:59.840 | where you happened to grow up.

01:12:00.840 | So this stuff's been going on for a long time.

01:12:02.280 | It's one of the things that I think is beautiful.

01:12:04.640 | Run DMC, Jam Master Jay used to play, he played piano.

01:12:09.320 | He would record himself playing piano

01:12:11.280 | and then sample that to make it a part

01:12:13.000 | of what was going on rather than play the piano.

01:12:15.360 | - That's how his mind can think more.

01:12:17.080 | - Well, it's pieces, you're putting pieces together.

01:12:18.400 | You're putting pieces of music together

01:12:19.440 | to create new music, right?

01:12:20.680 | Now, that doesn't mean that the root,

01:12:21.760 | I mean, the roots are doing their own thing, right?

01:12:24.640 | - Yeah, those are, that's a whole.

01:12:26.040 | - Yeah, but still, it's the right attitude.

01:12:29.680 | And what else is jazz, right?

01:12:31.760 | Jazz is about putting pieces together

01:12:33.120 | and then putting your own spin on it.

01:12:34.120 | It's all the same, it's all the same thing.

01:12:36.040 | It's all the same thing.

01:12:37.240 | - 'Cause you mentioned lyrics.

01:12:38.240 | It does make me sad, again,

01:12:40.440 | this is me talking trash about modern hip hop.

01:12:42.640 | I haven't investigated, I'm sure people will correct me

01:12:46.320 | that there's a lot of great artists.

01:12:47.680 | That's part of the reason I'm saying it

01:12:49.120 | is they'll leave it in the comments

01:12:50.680 | that you should listen to this person,

01:12:52.360 | is the lyrics went away from talking about

01:12:56.000 | maybe not just politics, but life and so on.

01:13:01.040 | The kind of protest songs,

01:13:03.920 | even if you look at a Bob Marley,

01:13:06.040 | or you see Public Enemy, or Rage Against the Machine

01:13:08.160 | more on the rock side,

01:13:09.360 | that's the place where we go to those lyrics.

01:13:14.520 | Classic rock is all about my woman left me,

01:13:18.720 | or I'm really happy that she's still with me,

01:13:22.200 | or the flip side, it's like love songs of different kinds.

01:13:24.680 | It's all love, but it's less political,

01:13:26.880 | like less interesting, I would say,

01:13:28.720 | in terms of deep, profound knowledge.

01:13:31.240 | It seems like rap is the place where you would find that,

01:13:35.320 | and it's sad that for the most part,

01:13:38.160 | what I see, you look at like mumble rap or whatever,

01:13:41.000 | they're moving away from lyrics

01:13:42.520 | and more towards the beat and the musicality of it.

01:13:45.680 | - I've always been a fan of the lyrics.

01:13:46.880 | In fact, if you go back and you read my reviews,

01:13:49.320 | which I recently was rereading,

01:13:51.320 | man, fuck, I wrote my last review

01:13:54.640 | the month I graduated, when I got my PhD,

01:13:57.360 | which says something about something.

01:13:58.520 | I'm not sure what, though.

01:14:00.120 | I always would, I don't always,

01:14:01.360 | but I often would start with, it's all about the lyrics.

01:14:03.200 | For me, it's about the lyrics.

01:14:04.840 | Someone has already written in the comments

01:14:08.240 | before I've even finished having this conversation

01:14:10.160 | that neither of us knows what we're talking about,

01:14:12.080 | and it's all in the underground hip hop,

01:14:14.120 | and here's who you should go listen to.

01:14:15.640 | And that is true.

01:14:16.480 | Every time I despair for popular rap,

01:14:19.960 | someone points me to where I discover

01:14:22.800 | some underground hip hop song,

01:14:24.240 | and I am made happy and whole again.

01:14:26.560 | So I know it's out there.

01:14:28.640 | I don't listen to it as much as I used to,

01:14:30.640 | because I'm listening to podcasts

01:14:32.000 | and old music from the 1980s and '80s.

01:14:34.000 | - Kind of rap, no beat, though.

01:14:34.960 | - It's a kind of, no, no beat at all,

01:14:36.800 | but there's a little bit of sampling here and there,

01:14:38.600 | I'm sure. (laughing)

01:14:40.840 | - By the way, James Brown is funk or no?

01:14:42.600 | - Yes, and so is Junior Wells, by the way.

01:14:45.120 | - Who's that?

01:14:45.960 | - Ah, Junior Wells, Chicago Blues.

01:14:48.400 | He was James Brown before James Brown was.

01:14:50.440 | - It's hard to imagine somebody being James Brown.

01:14:52.840 | - Go look up Hoodoo Man Blues, Junior Wells,

01:14:55.740 | and just listen to Snatch It Back and Hold It,

01:15:00.060 | and you'll see it.

01:15:02.200 | And they were contemporaries.

01:15:03.440 | - Where do you put Little Richard

01:15:04.840 | or all that kind of stuff, like Ray Charles,

01:15:07.280 | like when they get hit the road, jack,

01:15:10.720 | and don't you come back?

01:15:11.960 | Isn't that like, there's a funkiness in it.

01:15:14.080 | - Oh, there's definitely a funkiness in it.

01:15:15.640 | I mean, it's all, I mean, it's all a lie.

01:15:18.240 | I mean, it's all, there's all a line

01:15:19.880 | that carries it all together.

01:15:21.240 | You know, it's, I guess I would answer your question

01:15:23.600 | depending on whether I'm thinking about it in 2020

01:15:25.320 | or I'm thinking about it in 1960.

01:15:27.680 | I'd probably give a different answer.

01:15:28.920 | I'm just thinking in terms of, you know, that was rock,

01:15:31.120 | but when you look back on it, it was funky.

01:15:35.080 | But we didn't use those words, or maybe we did,

01:15:36.960 | I wasn't around, but you know,

01:15:38.400 | I don't think we used the word 1960, funk.

01:15:40.760 | Certainly not the way we used it in the '70s and the '80s.

01:15:43.160 | - Do you reject disco?

01:15:44.720 | - I do not reject disco.

01:15:45.640 | I appreciate all the mistakes that we have made

01:15:47.600 | to get to where we are now.

01:15:49.000 | Actually, some of the disco is actually really, really good.

01:15:51.120 | - John Travolta, oh boy, he regrets it probably.

01:15:54.520 | Maybe not.

01:15:55.360 | Well, like it's the mistakes thing.

01:15:56.840 | - Yeah, and it got him to where he's going, where he is.

01:15:59.640 | - Oh, well, thank you for taking that detour.

01:16:03.820 | You've talked about computing,

01:16:05.720 | we've already talked about computing a little bit,

01:16:08.480 | but can you try to describe how you think

01:16:11.320 | about the world of computing,

01:16:13.440 | where it fits into the sets of different disciplines?

01:16:16.120 | We mentioned College of Computing.

01:16:18.360 | What should people, how should they think about computing,

01:16:21.880 | especially from an educational perspective,

01:16:24.220 | of like what is the perfect curriculum

01:16:27.020 | that defines for a young mind what computing is?

01:16:32.900 | - So I don't know about a perfect curriculum,

01:16:34.280 | although that's an important question,

01:16:35.400 | because at the end of the day,

01:16:36.680 | without the curriculum, you don't get anywhere.

01:16:38.520 | Curriculum, to me, is the fundamental data structure.

01:16:41.040 | It's not even the classroom.

01:16:42.080 | - Data structure, I love it.

01:16:43.680 | - Right?

01:16:44.520 | So I think the curriculum is where I like to play.

01:16:48.000 | So I spend a lot of time thinking about this.

01:16:50.400 | But I will tell you, I'll answer your question

01:16:51.840 | by answering a slightly different question first

01:16:53.240 | and getting back to this, which is,

01:16:54.760 | you talked about disciplines,

01:16:56.180 | and what does it mean to be a discipline?

01:16:58.560 | The truth is, what we really educate people in

01:17:01.880 | from the beginning, but certainly through college,

01:17:04.600 | you've sort of failed if you don't think about it this way,

01:17:06.700 | I think, is the world,

01:17:09.420 | people often think about tools and tool sets,

01:17:13.400 | and when you're really trying to be good,

01:17:14.520 | you think about skills and skill sets.

01:17:16.660 | But disciplines are about mindsets, right?

01:17:18.760 | They're about fundamental ways of thinking,

01:17:21.500 | not just the hammer that you pick up,

01:17:23.940 | whatever that is, to hit the nail,

01:17:26.640 | not just the skill of learning how to hammer well,

01:17:29.160 | or whatever, it's the mindset of, like,

01:17:31.040 | what's the fundamental way to think about the world, right?

01:17:36.040 | And disciplines, different disciplines,

01:17:38.080 | give you different mindsets.

01:17:39.680 | They give you different ways of sort of thinking through.

01:17:41.600 | So, with that in mind, I think that computing,

01:17:44.760 | to even ask the question whether it's a discipline,

01:17:46.240 | you have to decide, does it have a mindset,

01:17:48.040 | does it have a way of thinking about the world

01:17:49.640 | that is different from the scientist who is doing discovery

01:17:53.480 | and using the scientific method as a way of doing it,

01:17:55.280 | or the mathematician who builds abstractions

01:17:57.440 | and tries to find sort of steady-state truth

01:17:59.800 | about the abstractions that may be artificial, but whatever.

01:18:03.720 | Or is it the engineer who's all about

01:18:06.440 | building demonstrably superior technology

01:18:08.460 | with respect to some notion of trade-offs,

01:18:11.080 | whatever that means, right?

01:18:11.920 | That's sort of the world that you live in.

01:18:14.480 | What is computing?

01:18:15.640 | How is computing different?

01:18:16.800 | So, I've thought about this for a long time,

01:18:18.040 | and I've come to a view about what computing actually is,

01:18:22.160 | what the mindset is, and it's a little abstract,

01:18:25.460 | but that would be appropriate for computing.

01:18:27.280 | I think that what distinguishes the computationalist

01:18:30.200 | from others is that he or she understands

01:18:36.200 | that models, languages, and machines are equivalent.

01:18:39.960 | They're the same thing.

01:18:41.520 | And because it's not just a model,

01:18:43.960 | but it's a machine that is an executable thing

01:18:46.520 | that can be described as a language,

01:18:49.140 | that means that it's dynamic.

01:18:51.300 | So, it is mathematical in some sense,

01:18:54.100 | in the kind of sense of abstraction,

01:18:55.960 | but it is fundamentally dynamic and executable.

01:18:58.320 | The mathematician is not necessarily worried

01:19:00.000 | about either the dynamic part.

01:19:01.780 | In fact, whenever I tried to write something

01:19:04.100 | for mathematicians, they invariably demand

01:19:07.040 | that I make it static, and that's not a bad thing.

01:19:09.320 | It's just, it's a way of viewing the world,

01:19:11.240 | that truth is a thing, right?

01:19:12.640 | It's not a process that continually runs, right?

01:19:16.000 | So, that dynamic thing matters,

01:19:18.000 | that self-reflection of the system itself matters,

01:19:20.960 | and that is what computing brought us.

01:19:23.360 | So, it is a science, because the models fundamentally

01:19:26.260 | represent truths in the world.

01:19:27.960 | Information is a scientific thing to discover, right?

01:19:30.900 | Not just a mathematical conceit that gets created.

01:19:33.920 | But of course, it's engineering,

01:19:35.280 | because you're actually dealing with constraints

01:19:37.180 | in the world and trying to execute machines

01:19:39.640 | that actually run.

01:19:41.420 | But it's also a math, because you're actually worrying

01:19:44.120 | about these languages that describe what's happening.

01:19:47.400 | But the fact that regular expressions

01:19:52.400 | and finite state automata, one of which feels like a machine,

01:19:56.160 | or at least an abstraction machine,

01:19:57.360 | and the other is a language,

01:19:58.280 | that they're actually the equivalent thing.

01:19:59.720 | I mean, that is not a small thing,

01:20:01.360 | and it permeates everything that we do,

01:20:03.280 | even when we're just trying to figure out

01:20:04.560 | how to do debugging.

01:20:05.720 | So, that idea, I think, is fundamental,

01:20:08.320 | and we would do better if we made that more explicit.

01:20:12.220 | How my life has changed in my thinking about this

01:20:15.300 | in the 10 or 15 years it's been

01:20:18.100 | since I tried to put that to paper with some colleagues

01:20:20.680 | is the realization, which comes to a question

01:20:24.360 | you actually asked me earlier,

01:20:27.560 | which has to do with trees falling down

01:20:29.000 | and whether it matters,

01:20:30.160 | is this sort of triangle of equality,

01:20:34.660 | it only matters because there's a person

01:20:36.540 | inside the triangle, right?

01:20:38.840 | That what's changed about computing,

01:20:42.020 | computer science, whatever you want to call it,

01:20:44.140 | is we now have so much data

01:20:46.580 | and so much computational power,

01:20:48.820 | we're able to do really, really interesting,

01:20:50.780 | promising things.

01:20:52.040 | But the interesting and the promising

01:20:55.060 | kind of only matters with respect to human beings

01:20:56.960 | and their relationship to it.

01:20:58.340 | So, the triangle exists, that is fundamentally computing.

01:21:01.820 | What makes it worthwhile and interesting,

01:21:03.440 | and potentially world species changing,

01:21:08.140 | is that there are human beings inside of it,

01:21:10.100 | and intelligence that has to interact with it

01:21:11.980 | that changes the data, the information that makes sense

01:21:14.540 | and gives meaning to the models,

01:21:17.460 | the languages, and the machines.

01:21:19.180 | So, if the curriculum can convey that

01:21:23.020 | while conveying the tools and the skills

01:21:24.660 | that you need in order to succeed,

01:21:26.740 | then it is a big win.

01:21:28.740 | That's what I think you have to do.

01:21:30.940 | - Do you pull psychology,

01:21:33.100 | it's like these human things into that,

01:21:35.900 | into the idea, into this framework of computing,

01:21:39.100 | do you pull in psychology, neuroscience,

01:21:41.340 | like parts of psychology, parts of neuroscience,

01:21:43.220 | parts of sociology?

01:21:44.780 | What about philosophy, like studies of human nature

01:21:48.180 | from different perspectives?

01:21:49.260 | - Absolutely.

01:21:50.100 | And by the way, it works both ways.

01:21:51.300 | So, let's take biology for a moment.

01:21:53.380 | It turns out a cell is basically

01:21:54.740 | a bunch of if-then statements,

01:21:56.220 | if you look at it the right way,

01:21:57.860 | which is nice because I understand if-then statements.

01:22:00.540 | I never really enjoyed biology,

01:22:01.860 | but I do understand if-then statements.

01:22:03.380 | And if you tell the biologist that

01:22:05.020 | and they begin to understand that,

01:22:06.140 | it actually helps them to think about

01:22:08.220 | a bunch of really cool things.

01:22:09.700 | There'll still be biology involved, but whatever.

01:22:11.980 | On the other hand, the fact of biology is,

01:22:15.180 | in fact, the cell is a bunch of if-then statements

01:22:17.420 | or whatever, allows the computationalists

01:22:19.420 | to think differently about the language

01:22:20.980 | and the way that we, well,

01:22:22.220 | certainly the way we would do AI and machine learning,

01:22:23.900 | but there's just even the way that we think about,

01:22:25.960 | we think about computation.

01:22:27.180 | So, the important thing to me is,

01:22:29.740 | as my engineering colleagues

01:22:31.440 | who are not in computer science worry about computer science

01:22:33.780 | eating up engineering colleges

01:22:35.540 | where computer science is trapped,

01:22:37.400 | it's not a worry.

01:22:40.380 | You shouldn't worry about that at all.

01:22:41.520 | Computing is, computer science, computing,

01:22:43.680 | it's not, it's central,

01:22:45.020 | but it's not the most important thing in the world.

01:22:46.880 | It's not more important.

01:22:47.720 | It is just key to helping others

01:22:50.900 | do other cool things they're gonna do.

01:22:52.380 | You're not gonna be a historian in 2030.

01:22:54.740 | You're not gonna get a PhD in history

01:22:55.860 | without understanding some data science and computing

01:22:57.860 | because the way you're gonna get history done, in part,

01:23:00.940 | and I say done,

01:23:01.820 | the way you're gonna get it done

01:23:03.040 | is you're going to look at data

01:23:04.320 | and you're gonna let,

01:23:05.280 | you're gonna have the system

01:23:06.200 | that's gonna help you to analyze things,

01:23:07.480 | to help you to think about a better way to describe history

01:23:10.060 | and to understand what's going on

01:23:11.020 | and what it tells us about where we might be going.

01:23:13.480 | The same is true for psychology,

01:23:14.460 | same is true for all of these things.

01:23:16.120 | The reason I brought that up

01:23:17.040 | is because the philosopher has a lot to say about computing.

01:23:20.200 | The psychologist has a lot to say

01:23:22.060 | about the way humans interact with computing, right?

01:23:24.800 | And certainly a lot about intelligence,

01:23:26.400 | which, at least for me,

01:23:28.960 | ultimately is kind of the goal

01:23:30.660 | of building these computational devices

01:23:32.300 | is to build something intelligent.

01:23:33.620 | - Did you think computing will eat everything

01:23:36.380 | in some certain sense

01:23:37.500 | or almost like disappear because it's part of everything?

01:23:40.460 | - It's so funny you say this.

01:23:41.300 | I wouldn't say it's gonna metastasize,

01:23:42.860 | but there's kind of two ways that fields destroy themselves.

01:23:46.620 | One is they become super narrow,

01:23:48.680 | and I think we can think of fields that might be that way.

01:23:53.500 | They become pure.

01:23:54.860 | And we have that instinct, we have that impulse.

01:23:57.100 | I'm sure you can think of several people

01:23:58.660 | who want computer science to be this pure thing.

01:24:01.860 | The other way is you become everywhere

01:24:04.240 | and you become everything and nothing.

01:24:06.060 | And so everyone says,

01:24:07.440 | I'm gonna teach Fortran for engineers or whatever,

01:24:10.320 | I'm gonna do this.

01:24:11.420 | And then you lose the thing

01:24:12.660 | that makes it worth studying in and of itself.

01:24:15.460 | The thing about computing,

01:24:16.500 | and this is not unique to computing,

01:24:18.380 | though at this point in time,

01:24:19.620 | it is distinctive about computing,

01:24:21.820 | where we happen to be in 2020,

01:24:23.600 | is we are both a thriving major.

01:24:26.660 | In fact, the thriving major, almost every place.

01:24:29.260 | And we're a service unit,

01:24:33.340 | because people need to know the things we need to know.

01:24:36.220 | And our job, much as the mathematician's job,

01:24:38.600 | is to help this person over here

01:24:40.820 | to think like a mathematician,

01:24:41.880 | much the way the point of you taking chemistry as a freshman

01:24:45.840 | is not to learn chemistry,

01:24:47.000 | it's to learn to think like a scientist, right?

01:24:48.820 | Our job is to help them to think like a computationalist,

01:24:51.980 | and we have to take both of those things very seriously.

01:24:54.460 | And I'm not sure that as a field,

01:24:56.820 | we have historically certainly taken the second thing,

01:24:59.180 | that our job is to help them to think a certain way.

01:25:01.940 | People who aren't gonna be our major,

01:25:02.780 | I don't think we've taken that very seriously at all.

01:25:06.020 | - I don't know if you know who Dan Carlin is,

01:25:07.700 | he has this podcast called "Hardcore History."

01:25:09.780 | - Yes.

01:25:10.620 | - I've just did an amazing four-hour conversation with him,

01:25:14.380 | mostly about Hitler.

01:25:15.780 | But I bring him up because he talks about this idea

01:25:20.240 | that it's possible that history as a field will become,

01:25:24.340 | like currently, most people study history a little bit,

01:25:29.340 | kind of are aware of it, we have a conversation about it,

01:25:33.140 | different parts of it, I mean,

01:25:34.300 | there's a lot of criticism to say

01:25:35.700 | that some parts of history are being ignored,

01:25:37.300 | blah, blah, blah, so on.

01:25:38.780 | But most people are able to have a curiosity

01:25:42.620 | and able to learn it.

01:25:44.000 | His thought is it's possible,

01:25:47.380 | given the way social media works,

01:25:49.100 | the current way we communicate,

01:25:51.600 | that history becomes a niche field

01:25:53.900 | where literally most people just ignore,

01:25:57.260 | 'cause everything is happening so fast

01:25:59.300 | that the history starts losing its meaning

01:26:01.860 | and then it starts being a thing that only,

01:26:04.880 | like the theoretical computer science part

01:26:08.300 | of computer science, it becomes a niche thing

01:26:10.620 | that only the rare holders of the World Wars

01:26:14.980 | and all the history, the founding of the United States,

01:26:19.500 | all those kinds of things, the Civil Wars.

01:26:22.420 | And it's a kind of profound thing to think about,

01:26:26.220 | how we can lose track, how we can lose these fields

01:26:31.220 | when they're best, like in the case of history,

01:26:34.340 | it's best for that to be a pervasive thing

01:26:36.940 | that everybody learns and thinks about and so on.

01:26:40.700 | And I would say computing is quite obviously

01:26:44.180 | similar to history in the sense that it seems

01:26:47.100 | like it should be a part of everybody's life to some degree,

01:26:51.540 | especially as we move into the later parts

01:26:54.380 | of the 21st century.

01:26:56.300 | And it's not obvious that that's the way it'll go.

01:26:59.140 | It might be in the hands of the few still.

01:27:02.020 | Depending if it's machine learning,

01:27:04.780 | it's unclear that computing will win out.

01:27:09.100 | It's currently very successful, but it's not,

01:27:11.300 | I would say that's something,

01:27:13.060 | I mean, you're at the leadership level of this,

01:27:15.420 | you're defining the future, so it's in your hands.

01:27:18.540 | - No pressure.

01:27:19.380 | - But it feels like there's multiple ways

01:27:21.420 | this can go, and there's this kind of conversation

01:27:25.020 | of everybody should learn to code, right?

01:27:27.900 | The changing nature of jobs and so on.

01:27:30.160 | Do you have a sense of what your role

01:27:36.100 | in education of computing is here?

01:27:40.860 | Like what's the hopeful path forward?

01:27:42.940 | - There's a lot there.

01:27:43.780 | I will say that, well, first off,

01:27:45.620 | it would be an absolute shame if no one studied history.

01:27:49.740 | On the other hand, as T approaches infinity,

01:27:51.220 | the amount of history is presumably also growing,

01:27:53.740 | at least linearly, and so you have to forget

01:27:58.620 | more and more of history, but history

01:28:00.780 | needs to always be there.

01:28:01.620 | I mean, I can imagine a world where,

01:28:03.340 | if you think of your brains as being outside of your head,

01:28:06.400 | that you can kind of learn the history you need to know

01:28:08.740 | when you need to know it.

01:28:09.900 | That seems fanciful, but it's a kind of way of,

01:28:12.620 | is there a sufficient statistic of history?

01:28:15.720 | No, and there certainly, but there may be

01:28:17.740 | for the particular thing you have to care about,

01:28:19.260 | but those who do not remember.

01:28:20.820 | - For our objective camera discussion, right?

01:28:23.220 | - Yeah, right, and we've already lost lots of history,

01:28:26.060 | and of course, you have your own history,

01:28:27.420 | that some of which will be, it's even lost to you.

01:28:29.980 | You don't even remember whatever it was

01:28:31.780 | you were doing 17 years ago.

01:28:33.420 | - All the ex-girlfriends.

01:28:34.540 | - Yeah.

01:28:35.380 | - Gone.

01:28:36.820 | - Exactly, so history's being lost anyway,

01:28:39.660 | but the big lessons of history shouldn't be,

01:28:41.820 | and I think to take it to the question

01:28:44.620 | of computing and sort of education,

01:28:46.260 | the point is you have to get across those lessons.

01:28:48.340 | You have to get across the way of thinking.

01:28:50.960 | And you have to be able to go back,

01:28:52.640 | and you don't wanna lose the data,

01:28:54.600 | even if you don't necessarily have the information

01:28:56.600 | at your fingertips.

01:28:57.880 | With computing, I think it's somewhat different.

01:29:00.640 | Everyone doesn't have to learn how to code,

01:29:02.800 | but everyone needs to learn how to think in the way

01:29:06.040 | that you can be precise, and I mean precise

01:29:08.240 | in the sense of repeatable, not just in the sense of,

01:29:12.600 | not resolution in the sense of get the right number of bits.

01:29:15.600 | In saying what it is you want the machine to do,

01:29:19.520 | and being able to describe a problem in such a way

01:29:22.280 | that it is executable, which we are not,

01:29:26.040 | human beings are not very good at that.

01:29:27.760 | In fact, I think we spend much of our time

01:29:29.240 | talking back and forth just to kind of vaguely understand

01:29:31.480 | what the other person means, and hope we get it good enough

01:29:33.760 | that we can act accordingly.

01:29:35.360 | You can't do that with machines, at least not yet.

01:29:38.320 | And so having to think that precisely

01:29:42.320 | about things is quite important,

01:29:45.240 | and that's somewhat different from coding.

01:29:47.600 | Coding is a crude means to an end.

01:29:51.160 | On the other hand, the idea of coding,

01:29:53.720 | what that means, that it's a programming language

01:29:55.760 | and it has these sort of things that you fiddle with

01:29:58.160 | in these ways that you express,

01:29:59.560 | that is an incredibly important point.

01:30:01.080 | In fact, I would argue that one of the big holes

01:30:03.520 | in machine learning right now and in AI

01:30:05.080 | is that we forget that we are basically doing

01:30:08.720 | software engineering.

01:30:09.560 | We forget that we are doing, we are using programming.

01:30:13.440 | Like, we're using languages to express what we're doing.

01:30:15.400 | We get just so all caught up in the deep network,

01:30:17.280 | or we get all caught up in whatever,

01:30:18.360 | that we forget that we're making decisions

01:30:22.800 | based upon a set of parameters that we made up.

01:30:25.440 | And if we did slightly different parameters,

01:30:27.080 | we'd have completely different outcomes.

01:30:29.080 | And so the lesson of computing, computer science education,

01:30:32.440 | is to be able to think like that,

01:30:34.760 | and to be aware of it when you're doing it.

01:30:36.760 | Basically, at the end of the day,

01:30:38.200 | it's a way of surfacing your assumptions.

01:30:41.360 | I mean, we call them parameters,

01:30:42.720 | or we call them if-then statements or whatever,

01:30:45.960 | but you're forced to surface those assumptions.

01:30:48.520 | That's the key thing that you should get out

01:30:50.720 | of a computing education, that and that the models,

01:30:52.600 | the languages, and the machines are equivalent.

01:30:54.040 | But it actually follows from that,

01:30:56.140 | that you have to be explicit about

01:30:58.800 | what it is you're trying to do,

01:30:59.760 | because the model you're building

01:31:02.000 | is something you will one day run.

01:31:04.360 | So you better get it right, or at least understand it,

01:31:06.720 | and be able to express roughly what you want to express.

01:31:09.740 | So I think it is key that we figure out

01:31:14.240 | how to educate everyone to think that way,

01:31:19.040 | because at the end, it will not only make them better

01:31:22.400 | at whatever it is that they are doing,

01:31:24.460 | and I emphasize doing,

01:31:27.480 | it'll also make them better citizens.

01:31:29.680 | It'll help them to understand what others are doing to them

01:31:33.080 | so that they can react accordingly,

01:31:35.440 | 'cause you're not gonna solve the problem of social media

01:31:37.720 | insofar as you think of social media as a problem

01:31:40.800 | by just making slightly better code, right?

01:31:43.920 | It only works if people react to it appropriately

01:31:47.640 | and know what's happening,

01:31:49.160 | and therefore take control over what they're doing.

01:31:52.560 | I mean, that's my take on it.

01:31:54.240 | - Okay, let me try to proceed awkwardly

01:31:58.920 | into the topic of race.

01:32:00.440 | - Okay.

01:32:01.600 | - One is because it's a fascinating part of your story,

01:32:03.800 | and you're just eloquent and fun about it,

01:32:05.560 | and then the second is because we're living

01:32:07.720 | through a pretty tense time in terms of race tensions

01:32:13.320 | and discussions and ideas in this time in America.

01:32:17.840 | You grew up in Atlanta, not born in Atlanta.

01:32:22.600 | Is some southern state somewhere, Tennessee,

01:32:24.760 | something like that? - Tennessee.

01:32:25.660 | - Nice, okay.

01:32:26.680 | But early on, you moved, you basically,

01:32:31.480 | you identify as an Atlanta native, yeah,

01:32:35.920 | and you've mentioned that you grew up

01:32:39.640 | in a predominantly black neighborhood.

01:32:42.520 | By the way, black, African-American, personal color.

01:32:45.080 | - I prefer black. - Black.

01:32:46.560 | - With a capital B. - With a capital B.

01:32:49.400 | The other letters are-- - The rest of them,

01:32:51.320 | no matter, it's a capital B.

01:32:52.960 | (Lex laughing)

01:32:54.680 | - Okay, so predominantly black neighborhood,

01:32:57.360 | and so you didn't almost see race.

01:32:59.360 | Maybe you can correct me on that.

01:33:01.720 | And then just in the video you talked about

01:33:04.680 | when you showed up to Georgia Tech for your undergrad,

01:33:08.040 | you were one of the only black folks there,

01:33:12.000 | and that was like, oh, that was a new experience.

01:33:14.520 | So can you take me from just a human perspective,

01:33:19.400 | but also from a race perspective,

01:33:21.100 | your journey growing up in Atlanta,

01:33:23.240 | and then showing up at Georgia Tech?

01:33:24.960 | - All right, that's easy.

01:33:25.840 | And by the way, that story continues through MIT as well.

01:33:28.800 | In fact, it was quite a bit more stark at MIT and Boston.

01:33:33.080 | - So maybe just a quick pause,

01:33:34.800 | Georgia Tech was undergrad, MIT was graduate school.

01:33:38.080 | - Mm-hmm, mm-hmm, and I went directly

01:33:39.640 | to grad school from undergrad,

01:33:40.800 | so I had no distractions in between

01:33:43.880 | my bachelor's and my master's and PhD.

01:33:45.760 | - You didn't go on a backpacking trip in Europe.

01:33:47.880 | - Didn't do any of that, didn't do it.

01:33:48.960 | In fact, I literally went to IBM for three months,

01:33:51.920 | got in a car, and drove straight to Boston with my mother,

01:33:54.720 | or Cambridge.

01:33:55.560 | Moved into an apartment I'd never seen over the Royal East.

01:34:00.840 | Anyway, that's another story.

01:34:02.320 | So let me tell you a little bit about--

01:34:03.640 | - You miss MIT?

01:34:05.120 | - Oh, I loved MIT.

01:34:06.280 | I don't miss Boston at all, but I loved MIT.

01:34:09.200 | - And then Miss Fighting Warrants, okay.

01:34:10.360 | - Oh, so let's back up to this.

01:34:12.440 | So as you said, I was born in Chattanooga, Tennessee.

01:34:14.960 | My earliest memory is arriving in Atlanta

01:34:16.680 | in a moving truck at the age of three and a half,

01:34:18.280 | so I think of myself as being from Atlanta.

01:34:20.360 | I have a very distinct memory of that.

01:34:21.520 | So I grew up in Atlanta,

01:34:22.640 | it's the only place I ever knew as a kid.

01:34:24.960 | I loved it.

01:34:25.800 | Like much of the country, and certainly much of Atlanta

01:34:29.560 | in the '70s and '80s, it was deeply, highly segregated,

01:34:33.240 | though not in a way that I think was obvious to you

01:34:36.200 | unless you were looking at it,

01:34:37.600 | or were old enough to have noticed it,

01:34:39.880 | but you could divide up Atlanta,

01:34:41.040 | and Atlanta's hardly unique in this way, by highway,

01:34:43.540 | and you could get race and class that way.

01:34:45.220 | So I grew up not only in a predominantly black area,

01:34:48.600 | to say the very least, I grew up on the poor side of that.

01:34:52.060 | But I was very much aware of race for a bunch of reasons,

01:34:57.640 | one, that people made certain that I was, my family did,

01:35:00.940 | but also that it would come up.

01:35:02.120 | So in first grade, I had a girlfriend.

01:35:07.120 | I say I had a girlfriend, I didn't have a girlfriend,

01:35:08.960 | I wasn't even entirely sure what girls were

01:35:10.760 | in the first grade, but I do remember she decided

01:35:13.880 | I was her girlfriend, this little white girl named Heather.

01:35:16.640 | And we had a long discussion about how it was okay

01:35:18.920 | for us to be boyfriend and girlfriend,

01:35:20.320 | despite the fact that she was white and I was black.

01:35:22.880 | - Between the two of you, or your parents?

01:35:24.760 | Did your parents know about this?

01:35:26.520 | - Yes, but being a girlfriend and boyfriend in first grade

01:35:29.800 | just basically meant that you spent

01:35:31.280 | slightly more time together during recess.

01:35:33.720 | It had no, I think we Eskimo kissed once.

01:35:35.720 | It doesn't mean, it didn't mean anything.

01:35:38.320 | It was, at the time, it felt very scandalous

01:35:40.400 | 'cause everyone was watching.

01:35:41.300 | I was like, ah, my life is, now my life has changed

01:35:44.080 | in first grade, no one told me elementary school

01:35:45.880 | would be like this.

01:35:46.720 | - Did you write poetry or?

01:35:47.840 | - Not in first grade, that would come later.

01:35:50.240 | That would come during puberty,

01:35:52.080 | when I wrote lots and lots of poetry.

01:35:54.280 | Anyway, so I was aware of it.

01:35:56.320 | I didn't think too much about it, but I was aware of it.

01:35:59.520 | But I was surrounded.

01:36:00.520 | It wasn't that I wasn't aware of race,

01:36:02.220 | it's that I wasn't aware that I was a minority.

01:36:05.480 | Different, and it's because I wasn't.

01:36:09.040 | As far as my world was concerned,

01:36:10.240 | I mean, I'm six years old, five years old in first grade.

01:36:13.120 | The world is the seven people I see every day.

01:36:15.640 | So it didn't feel that way at all.

01:36:18.040 | And by the way, this being Atlanta,

01:36:20.200 | home of the Civil Rights Movement and all the rest,

01:36:22.000 | it meant that when I looked at TV,

01:36:23.480 | which back then one did,

01:36:24.720 | 'cause there were only three, four, five channels, right?

01:36:27.040 | And I saw the news, which my mother might make me watch.

01:36:29.840 | Monica Kaufman was on TV telling me the news.

01:36:35.280 | And they were all black, and the mayor was black,

01:36:37.320 | and always been black.

01:36:38.160 | And so it just never occurred to me.

01:36:39.920 | When I went to Georgia Tech,

01:36:42.200 | I remember the first day walking across campus,

01:36:45.040 | from West Campus to East Campus,

01:36:47.360 | and realizing along the way

01:36:48.960 | that of the hundreds and hundreds and hundreds

01:36:50.560 | and hundreds of students that I was seeing,

01:36:52.400 | I was the only black one.

01:36:54.060 | That was enlightening and very off-putting,

01:36:58.200 | because it occurred to me.

01:36:59.560 | And then of course it continued that way for,

01:37:02.240 | well, for much of the rest of my career at Georgia Tech.

01:37:05.840 | Of course I found lots of other students.

01:37:07.400 | And I met people, 'cause in Atlanta,

01:37:09.100 | you're either black or you're white.

01:37:10.320 | There was nothing else.

01:37:12.200 | So I began to meet students of Asian descent,

01:37:14.480 | and I met students who we would call Hispanic,

01:37:16.680 | and so on and so forth.

01:37:17.640 | And so my world, this is what college is supposed to do,

01:37:19.680 | right, it's supposed to open you up to people, and it did.

01:37:23.040 | But it was a very strange thing to be in the minority.

01:37:28.040 | When I came to Boston, I will tell you a story.

01:37:32.160 | I applied to one place as an undergrad, Georgia Tech,

01:37:36.840 | because I was stupid, I didn't know any better.

01:37:38.800 | I just didn't know any better, right?

01:37:41.320 | No one told me.

01:37:42.640 | When I went to grad school, I applied to three places,

01:37:44.920 | Georgia Tech, because that's where I was, MIT, and CMU.

01:37:47.680 | When I got in to MIT, I got into CMU,

01:37:54.720 | but I had a friend who went to CMU.

01:37:57.920 | And so I asked him what he thought about it.

01:37:59.560 | He spent his time explaining to me about Pittsburgh,

01:38:01.880 | much less about CMU, but more about Pittsburgh,

01:38:04.720 | of which I developed a strong opinion

01:38:06.520 | based upon his strong opinion,

01:38:07.860 | something about the sun coming out two days out of the year.

01:38:11.080 | And I didn't get a chance to go there

01:38:12.360 | because the timing was wrong.

01:38:14.240 | I think it was because the timing was wrong.

01:38:16.700 | At MIT, I asked 20 people I knew,

01:38:20.240 | either when I visited or I had already known

01:38:23.600 | for a variety of reasons whether they liked Boston.

01:38:26.160 | And 10 of them loved it, and 10 of them hated it.

01:38:29.560 | The 10 who loved it were all white.

01:38:31.120 | The 10 who hated it were all black.

01:38:33.480 | And they explained to me very much why that was the case.

01:38:36.280 | Both dads told me why.

01:38:38.840 | And the stories were remarkably the same

01:38:40.920 | for the two clusters.

01:38:42.240 | And I came up here, and I could see it immediately,

01:38:45.840 | why people would love it and why people would not.

01:38:48.000 | - Why people tell you about the nice coffee shops.

01:38:51.400 | - Well, it wasn't coffee shops.

01:38:52.240 | It was CD, used CD places.

01:38:54.080 | But yeah, it was that kind of a thing.

01:38:56.000 | Nice shops, oh, there's all these students here.

01:38:57.840 | Harvard Square is beautiful.

01:38:59.520 | You can do all these things,

01:39:00.400 | and you can walk in something about the outdoors,

01:39:01.960 | which I wasn't the slightest bit interested in.

01:39:03.480 | The outdoors is for the bugs, it's not for humans.

01:39:05.960 | And the-- (Lyle laughs)

01:39:08.320 | - That should be a T-shirt.

01:39:09.560 | - Yeah, I mean, it's the way I feel about it.

01:39:12.520 | And the black folk told me completely different stories

01:39:14.960 | about which part of town you did not wanna be caught in

01:39:18.040 | after dark, and I heard all, but that was nothing new.

01:39:22.060 | So I decided that MIT was a great place to be

01:39:25.480 | as a university, and I believed it then, I believe it now.

01:39:29.680 | And that whatever it is I wanted to do,

01:39:31.280 | I thought I knew what I wanted to do,

01:39:32.360 | but what if I was wrong?

01:39:34.180 | Someone there would know how to do it.

01:39:36.160 | Of course, then I would pick the one topic

01:39:37.680 | that nobody was working on at the time, but that's okay.

01:39:41.520 | It was great, and so I thought that I would be fine.

01:39:43.200 | And I'd only be there for like four or five years.

01:39:45.640 | I told myself, which turned out not to be true at all.

01:39:48.620 | But I enjoyed my time, I enjoyed my time there.

01:39:50.520 | But I did see a lot of,

01:39:52.060 | I ran across a lot of things

01:39:55.800 | that were driven by what I look like while I was here.

01:39:59.560 | I got asked a lot of questions, I ran into a lot of cops.

01:40:03.000 | I saw a lot about the city.

01:40:06.320 | But at the time, I mean, I haven't been here a long time,

01:40:08.480 | these are the things that I remember.

01:40:09.580 | So this is 1990.

01:40:11.180 | There was not a single black radio station.

01:40:16.360 | Now this is 1990.

01:40:17.800 | I don't know if there are any radio stations anymore.

01:40:19.720 | I'm sure there are, but I don't listen to the radio anymore,

01:40:22.320 | and almost no one does,

01:40:23.480 | at least if you're under a certain age.

01:40:26.800 | But the idea is you could be in a major metropolitan area

01:40:28.840 | and there wasn't a single black radio station,

01:40:30.400 | by which I mean a radio station

01:40:31.560 | that played what we would call black music then,

01:40:33.960 | was absurd, but somehow captured

01:40:37.760 | kind of everything about the city.

01:40:39.840 | I grew up in Atlanta,

01:40:41.160 | and you've heard me tell you about Atlanta.

01:40:43.440 | Boston had no economically viable

01:40:48.240 | or socially cohesive black middle class.

01:40:50.560 | Insofar as it existed, it was uniformly distributed

01:40:55.160 | throughout large parts, not all parts,

01:40:56.600 | but large parts of the city.

01:40:57.940 | And where you had concentrations of black Bostonians,

01:41:02.320 | they tended to be poor.

01:41:04.400 | It was very different from where I grew up.

01:41:06.120 | I grew up on the poor side of town, sure,

01:41:07.960 | but then in high school, well, in ninth grade,

01:41:11.040 | we didn't have middle school.

01:41:12.960 | I went to an eighth grade school

01:41:14.120 | where there was a lot of, let's just say,

01:41:16.280 | we had a riot the year that I was there.

01:41:17.920 | There was at least one major fight every week.

01:41:19.820 | It was an amazing experience.

01:41:24.820 | But when I went to ninth grade, I went to academy.

01:41:27.680 | - Math and-- - Math and Science Academy,

01:41:30.120 | Mays High, it was a public school.

01:41:31.880 | It was a magnet school, that's why I was able to go there.

01:41:34.640 | It was the first school, high school,

01:41:36.400 | I think, in the state of Georgia

01:41:37.480 | to sweep the state math and science fairs.

01:41:39.840 | It was great.

01:41:40.680 | It had 385 students, all but four of whom were black.

01:41:45.680 | I went to school with the daughter

01:41:50.360 | of the former mayor of Atlanta, Michael Jackson's cousin.

01:41:54.240 | I mean, you know, it was an upper middle class--

01:41:56.760 | - Dropping names. - Whatever.

01:41:58.120 | I just dropped names occasionally.

01:41:59.680 | You know, I dropped the mic, dropped some names,

01:42:01.920 | just to let you know I used to hang out

01:42:02.960 | with Michael Jackson's cousin.

01:42:04.440 | 12th cousin, nine times removed, I don't know.

01:42:06.200 | The point is, we had a parking problem

01:42:08.760 | 'cause the kids had cars.

01:42:10.080 | I did not come from a place where you had cars.

01:42:12.360 | I had my first car when I came to MIT, actually.

01:42:14.840 | So it was just a very different experience for me.

01:42:19.840 | But I'd been to places where, whether you were rich

01:42:24.660 | or whether you were poor, you could be black and rich

01:42:27.320 | or black and poor, and it was there,

01:42:28.440 | and there were places, and they were segregated

01:42:31.000 | by class as well as by race, but that existed.

01:42:34.240 | Here, at least when I was here,

01:42:35.720 | it didn't feel that way at all,

01:42:36.840 | and it felt like a bunch of,

01:42:38.120 | a really interesting contradiction.

01:42:40.260 | It felt like it was the interracial dating capital

01:42:44.240 | of the country.

01:42:47.080 | It really felt that way.

01:42:48.320 | But it also felt like the most racist place

01:42:51.440 | I ever spent any time.

01:42:54.240 | You couldn't go up the Orange Line, at that time.

01:42:56.580 | I mean, again, that was 30 years ago.

01:42:58.940 | I don't know what it's like now.

01:43:00.180 | But there were places you couldn't go,

01:43:01.780 | and you knew it, everybody knew it.

01:43:04.740 | And there were places you couldn't live,

01:43:07.580 | and everybody knew that.

01:43:08.780 | And that was just the greater Boston area in 1992.

01:43:12.820 | - Subtle racism or explicit racism?

01:43:14.900 | - Both, both.

01:43:16.860 | - In terms of within the institutions,

01:43:18.900 | did you feel, was there levels in which you were empowered

01:43:24.140 | to be first, or one of the first black people

01:43:28.080 | in a particular discipline,

01:43:29.620 | in some of these great institutions that you were a part of,

01:43:33.060 | you know, Georgia Tech or MIT?

01:43:35.640 | And was there a part where it was, it felt limiting?

01:43:38.920 | - I always felt empowered.

01:43:41.520 | Some of that was my own delusion, I think.

01:43:44.580 | But it worked out.

01:43:46.380 | So I never felt, in fact, quite the opposite.

01:43:50.260 | Not only did I not feel as if no one was trying to stop me,

01:43:54.340 | I had the distinct impression

01:43:55.420 | that people wanted me to succeed.

01:43:57.580 | By people, I meant the people in power.

01:44:00.900 | Not my fellow students, not that they didn't want me

01:44:02.460 | to succeed, but I felt supported,

01:44:07.020 | or at least that people were happy to see me succeed

01:44:10.420 | at least as much as anyone else.

01:44:12.260 | But you know, 1990, you're dealing with a different set

01:44:14.900 | of problems, right?

01:44:15.740 | You're very early, at least in computer science,

01:44:18.620 | you're very early in the sort of Jackie Robinson period.

01:44:21.540 | You know, there's this thing called

01:44:22.540 | Jackie Robinson syndrome, which is that you have to,

01:44:25.940 | you know, the first one has to be perfect,

01:44:27.860 | or has to be sure to succeed because if that person fails,

01:44:30.700 | no one else comes after for a long time.

01:44:32.440 | So, you know, it was kind of in everyone's best interest.

01:44:35.860 | But I think it came from a sincere place.

01:44:37.460 | I'm completely sure that people went out of their way

01:44:40.700 | to try to make certain that the environment would be good.

01:44:45.220 | Not just for me, but for the other people

01:44:46.780 | who of course were around.

01:44:47.620 | And I was hardly the only, I was the only person

01:44:49.300 | in the AI lab, but I wasn't the only person

01:44:52.220 | at MIT by a long shot.

01:44:54.060 | On the other hand, we're what?

01:44:55.860 | At that point, we would have been what?

01:44:57.220 | Less than 20 years away from the first black PhD

01:45:00.460 | to graduate from MIT, right?

01:45:02.780 | Shirley Jackson, right?

01:45:03.620 | 1971, something like that, somewhere around then.

01:45:07.020 | So we weren't that far away from the first first.

01:45:10.820 | And we were still another eight years away

01:45:12.700 | from the first black PhD computer science, right?

01:45:15.900 | So we were in a, it was a sort of interesting time.

01:45:19.620 | But I did not feel as if the institutions

01:45:21.880 | of the university were against any of that.

01:45:26.880 | And furthermore, I felt as if there was enough

01:45:29.160 | of a critical mass across the institute

01:45:31.580 | from students and probably faculty,

01:45:34.120 | though I didn't know them, who wanted to make certain

01:45:37.060 | that the right thing happened.

01:45:38.580 | That's very different from the institutions

01:45:40.180 | of the rest of the city, which I think were designed

01:45:44.140 | in such a way that they felt no need to be supportive.

01:45:47.540 | - Let me ask a touchy question on that.

01:45:51.280 | So you kind of said that you didn't feel,

01:45:56.160 | you felt empowered.

01:46:00.180 | Is there some lesson, advice, in the sense

01:46:04.220 | that no matter what, you should feel empowered?

01:46:07.860 | You said, you used the word, I think, illusion or delusion.

01:46:12.060 | Is there a sense from the individual perspective

01:46:15.140 | where you should always kind of ignore, you know,

01:46:20.800 | the, ignore your own eyes,

01:46:23.160 | ignore the little forces that you are able

01:46:32.260 | to observe around you that are like trying

01:46:35.340 | to mess with you of whether it's jealousy,

01:46:37.400 | whether it's hatred in its pure form,

01:46:39.880 | whether it's just hatred in its like deluded form,

01:46:43.820 | all that kind of stuff, and just kind of see yourself

01:46:46.780 | as empowered and confident, all those kinds of things?

01:46:50.340 | - I mean, it certainly helps, but there's a trade-off, right?

01:46:52.540 | You have to be deluded enough to think

01:46:53.880 | that you can succeed.

01:46:55.020 | I mean, you can't get a PhD unless you're crazy enough

01:46:57.020 | to think you can invent something

01:46:58.820 | that no one else has come up with.

01:47:00.140 | I mean, that kind of massive delusion is that.

01:47:02.340 | So you have to be deluded enough to believe

01:47:03.700 | that you can succeed despite whatever odds you see

01:47:05.760 | in front of you, but you can't be so deluded

01:47:07.580 | that you don't think that you need to step out

01:47:09.020 | of the way of the oncoming train.

01:47:10.740 | - Right. - Right.

01:47:11.580 | So it's all a trade-off, right?

01:47:12.940 | You have to kind of believe in yourself.

01:47:14.220 | It helps to have a support group around you

01:47:16.540 | in some way or another.

01:47:18.060 | I was able to find that.

01:47:19.500 | I've been able to find that wherever I've gone,

01:47:21.860 | even if it wasn't necessarily on the floor that I was in.

01:47:24.100 | I had lots of friends when I was here.

01:47:25.500 | Many of them still live here,

01:47:27.660 | and I've kept up with many of them.

01:47:29.140 | So I felt supported, and certainly I had my mother

01:47:31.940 | and my family and those people back home

01:47:34.500 | that I could always lean back on,

01:47:36.980 | even if it were a long-distance call that cost money,

01:47:39.420 | which is not something that any of the kids today

01:47:42.100 | even know what I'm talking about.

01:47:43.100 | But back then it mattered.

01:47:44.580 | Calling my mom was an expensive proposition.

01:47:46.980 | But you have that, and it's fine.

01:47:48.300 | I think it helps.

01:47:49.340 | But you cannot be so deluded that you miss the obvious

01:47:52.540 | because it makes things slower,

01:47:54.140 | and it makes you think you're doing better than you are,

01:47:56.620 | and it will hurt you in the long run.

01:47:58.520 | - You mentioned cops.

01:48:02.540 | You tell a story of being pulled over.

01:48:05.060 | Perhaps it happened more than once.

01:48:07.100 | - More than once, for sure.

01:48:09.020 | - One, could you tell that story?

01:48:10.700 | And in general, can you give me a sense

01:48:14.180 | of what the world looks like

01:48:17.580 | when the law doesn't always look at you

01:48:20.140 | with the blank slate,

01:48:24.920 | with the objective eyes?

01:48:31.700 | I don't know how to say it more poetically.

01:48:33.960 | - Well, I guess the, I don't either.

01:48:35.940 | I guess the answer is it looks exactly the way it looks now

01:48:39.940 | because this is the world that we happen to live in.

01:48:42.620 | It's people clustering and doing the things that they do

01:48:46.220 | and making decisions based on one or two bits

01:48:49.620 | of information they find relevant,

01:48:51.460 | which, by the way, are all positive feedback loops,

01:48:54.340 | which makes it easier for you to believe

01:48:56.500 | what you believed before

01:48:57.340 | because you behave in a certain way that makes it true,

01:48:59.540 | and it goes on and circles and it cycles

01:49:01.100 | and it cycles and it cycles.

01:49:02.460 | So it's just about being on edge.

01:49:06.260 | I do not, despite having made it over 50 now.

01:49:10.160 | Despite-- - Congratulations, by the way.

01:49:13.620 | - God, I have a few gray hairs here and there.

01:49:16.260 | - You did pretty good.

01:49:17.420 | - I think, you know, I don't imagine I will ever

01:49:22.180 | see a police officer and not get very, very tense.

01:49:25.660 | Now, everyone gets a little tense

01:49:28.480 | because it probably means you're being pulled over

01:49:30.460 | for speeding or something,

01:49:32.140 | or you're gonna get a ticket or whatever, right?

01:49:34.340 | I mean, the interesting thing about the law in general

01:49:36.620 | is that most human beings' experience of it

01:49:38.980 | is fundamentally negative, right?

01:49:41.820 | You're only dealing with a lawyer

01:49:43.160 | if you're in trouble,

01:49:44.540 | except in a few very small circumstances, right?

01:49:47.620 | But, so that's just, that's an underlying reality.

01:49:50.220 | Now imagine that that's also

01:49:51.420 | at the hands of the police officer.

01:49:53.060 | I remember the time when I was,

01:49:54.580 | when I got pulled over that time,

01:49:57.540 | halfway between Boston and Wellesley, actually.

01:49:59.940 | I remember thinking,

01:50:02.940 | as he, when he pulled his gun on me,

01:50:06.380 | that if he shot me right now, he'd get away with it.

01:50:11.540 | That was the worst thing that I felt

01:50:13.920 | about that particular moment,

01:50:15.360 | is that if he shoots me now, he will get away with it.

01:50:18.760 | It would be years later when I realized,

01:50:21.400 | actually much worse than that,

01:50:22.960 | is that he'd get away with it,

01:50:27.680 | and if anyone, if it became a thing

01:50:29.760 | that other people knew about,

01:50:31.000 | odds were, would be, of course, that it wouldn't,

01:50:32.940 | but if it became a thing that other people knew about,

01:50:34.400 | if I was living in today's world

01:50:35.840 | as opposed to the world 30 years ago,

01:50:38.380 | that not only would he get away with it,

01:50:40.400 | but that I would be painted a villain.

01:50:43.840 | I was probably big and scary,

01:50:45.300 | and I probably moved too fast,

01:50:46.500 | and if only I'd done what he said,

01:50:47.740 | and da, da, da, da, da, da,

01:50:49.320 | which is somehow worse, right?

01:50:51.240 | You know, that hurts not just you, you're dead,

01:50:53.700 | but your family, and the way people look at you,

01:50:57.420 | and look at your legacy or your history,

01:51:00.020 | that's terrible, and it would work.

01:51:02.300 | I absolutely believe it would've worked had he done it.

01:51:04.860 | Now, he didn't.

01:51:05.700 | I don't think he wanted to shoot me.

01:51:06.540 | I don't think he felt like killing anybody.

01:51:08.180 | He did not go out that night expecting to do that,

01:51:10.260 | or planning on doing it,

01:51:11.240 | and I wouldn't be surprised if he never, ever did that,

01:51:14.280 | or ever even pulled his gun again.

01:51:15.640 | I don't know the man's name.

01:51:16.600 | I don't remember anything about him.

01:51:17.600 | I do remember the gun.

01:51:18.680 | Guns are very big when they're in your face.

01:51:20.120 | I can tell you this much.

01:51:20.960 | They're much larger than they seem.

01:51:23.160 | - And you were basically speeding or something like that?

01:51:25.680 | - He said I ran a light, I think.

01:51:26.840 | - Ran a light, sure.

01:51:27.680 | - I don't think I ran a light,

01:51:29.080 | but in fact, I may not have even gotten a ticket.

01:51:31.740 | I may have just gotten a warning.

01:51:33.280 | I think he was a little spooked, too.

01:51:34.120 | - But he pulled a gun.

01:51:35.240 | - Yeah, apparently I moved too fast or something.

01:51:38.160 | Rolled my window down before I should've.

01:51:39.800 | It was unclear.

01:51:40.640 | I think he thought I was gonna do something,

01:51:42.720 | or at least that's how he behaved.

01:51:44.800 | - So how, if we can take a little walk around your brain,

01:51:49.380 | how do you feel about that guy,

01:51:55.280 | and how do you feel about cops after that experience?

01:51:59.160 | - Well, I don't remember that guy,

01:52:01.160 | but my views on police officers

01:52:02.880 | is the same view I have about lots of things.

01:52:06.480 | Fire is an important and necessary thing in the world,

01:52:10.320 | but you must respect fire because it will burn you.

01:52:13.860 | Fire is a necessary evil in the sense that it can burn you,

01:52:19.000 | necessary in the sense that, you know, heat,

01:52:21.360 | and all the other things that we use fire for.

01:52:23.340 | So when I see a cop, I see a giant ball of flame,

01:52:28.340 | and I just try to avoid it.

01:52:31.880 | - And then some people might see a nice place,

01:52:34.120 | a nice thing to roast marshmallows with a family over.

01:52:37.960 | - Which is fine, I don't roast marshmallows.

01:52:40.120 | - Okay, so let me go a little darker, and I apologize.

01:52:43.120 | Just talked to Dan Carlin about it, he left for four hours.

01:52:45.120 | So sorry if I go dark here a little bit,

01:52:48.640 | but is it easy for this experience

01:52:53.600 | of just being careful with the fire

01:52:55.760 | and avoiding it to turn to hatred?

01:52:58.600 | - Yeah, of course.

01:52:59.760 | And one might even argue that it is a illogical conclusion.

01:53:05.100 | On the other hand, you've got to live in the world,

01:53:07.620 | and I don't think it's helpful.

01:53:12.140 | Hate is something one should,

01:53:13.420 | I mean, hate is something that takes a lot of energy.

01:53:18.020 | So one should reserve it for when it is useful

01:53:21.940 | and not carried around with you all the time.

01:53:24.280 | Again, there's a big difference between the happy delusion

01:53:28.140 | that convinces you that you can actually get out of bed

01:53:30.140 | and make it to work today without getting hit by a car,

01:53:33.660 | and the sad delusion that means you can not worry

01:53:37.260 | about this car that is barreling towards you, right?

01:53:39.780 | So we all have to be a little deluded

01:53:41.900 | because otherwise we're paralyzed, right?

01:53:44.060 | But one should not be ridiculous.

01:53:46.460 | If we go all the way back to something you said earlier

01:53:48.580 | about empathy, I think what I would ask other people

01:53:53.580 | to get out of this one of many, many, many stories

01:54:00.140 | is to recognize that it is real.

01:54:03.160 | People would ask me to empathize with the police officer.

01:54:07.260 | I would quote back statistics saying that,

01:54:09.580 | you know, being a police officer isn't even

01:54:12.700 | in the top 10 most dangerous jobs in the United States,

01:54:14.860 | you're much more likely to get killed in a taxi cab.

01:54:17.540 | Half of police officers are actually killed by suicide.

01:54:21.940 | But that means their lives are something,

01:54:25.420 | something's going on there with them.

01:54:27.860 | And I would more than happy to be empathetic

01:54:30.100 | about what it is they go through

01:54:31.140 | and how they see the world.

01:54:32.780 | I think though that if we step back from what I feel,

01:54:37.220 | and we step back from what an individual police officer

01:54:39.500 | feels, you step up a level, and all this,

01:54:42.300 | because all things tie back into interactive AI.

01:54:45.500 | The real problem here is that we've built a narrative,

01:54:47.460 | we built a big structure that has made it easy

01:54:50.340 | for people to put themselves into different pots

01:54:53.240 | in the different clusters, and to basically

01:54:56.820 | forget that the people in the other clusters

01:54:59.540 | are ultimately like them.

01:55:02.300 | It is a useful exercise to ask yourself sometimes,

01:55:04.540 | I think, that if I had grown up in a completely

01:55:07.140 | different house, in a completely different household,

01:55:09.340 | as a completely different person, if I had been a woman,

01:55:12.140 | would I see the world differently?

01:55:13.660 | Would I believe what that crazy person over there believes?

01:55:16.700 | And the answer's probably yes,

01:55:19.820 | because after all, they believe it.

01:55:21.740 | And fundamentally, they're the same as you.

01:55:24.460 | So then what can you possibly do to fix it?

01:55:27.620 | How do you fix Twitter, if you think Twitter

01:55:29.460 | needs to be, is broken, or Facebook,

01:55:31.180 | if you think Facebook is broken, how do you fix racism?

01:55:34.340 | How do you fix any of these things?

01:55:36.740 | It's all structural, right?

01:55:38.740 | It's not, I mean, individual conversations matter a lot,

01:55:41.720 | but you have to create structures that allow people

01:55:45.420 | to have those individual conversations all the time

01:55:47.720 | in a way that is relatively safe,

01:55:50.460 | and that allows them to understand that other people

01:55:52.700 | have had different experiences, but that ultimately

01:55:54.660 | we're the same, which sounds very,

01:55:56.420 | I don't even know what the right word is.

01:55:59.660 | I'm trying to avoid a word like saccharine.

01:56:01.060 | But it feels very optimistic.

01:56:04.900 | But I think that's okay.

01:56:08.620 | I think that's a part of the delusion,

01:56:09.980 | is you wanna be a little optimistic,

01:56:11.580 | and then recognize that the hard problem

01:56:13.100 | is actually setting up the structures in the first place,

01:56:14.900 | because it's in almost no one's interest

01:56:17.900 | to change the infrastructure.

01:56:20.540 | - Right, I tend to believe that leaders

01:56:22.860 | have a big role to that, of selling that optimistic delusion

01:56:27.100 | to everybody, and that eventually leads

01:56:29.620 | to the building of the structures.

01:56:31.340 | But that requires a leader that unites,

01:56:34.520 | sort of unites everybody on a vision,

01:56:36.740 | as opposed to divides on a vision,

01:56:38.580 | which is this particular moment in history feels

01:56:43.460 | like there's a non-zero probability, if we go to the P,

01:56:49.380 | of something akin to a violent or a non-violent civil war.

01:56:53.620 | This is one of the most divisive periods

01:56:57.100 | of American history in recent,

01:56:58.700 | you can speak to this from perhaps a more knowledgeable

01:57:03.700 | and deeper perspective than me,

01:57:05.340 | but from my naive perspective,

01:57:06.980 | this seems like a very strange time.

01:57:09.740 | There's a lot of anger, and it has to do with people,

01:57:13.920 | I mean, for many reasons.

01:57:15.500 | One, the thing that's not spoken about,

01:57:17.660 | I think, much is the quiet economic pain of millions

01:57:22.660 | that's growing because of COVID,

01:57:28.500 | because of closed businesses, because of lost dreams.

01:57:32.100 | So that's building, whatever that tension is building.

01:57:35.700 | The other is, there seems to be

01:57:38.300 | an elevated level of emotion.

01:57:39.940 | I'm not sure if you can psychoanalyze

01:57:41.940 | where that's coming from, but this sort of,

01:57:44.620 | from which the protests and so on percolated.

01:57:47.300 | It's like, why now?

01:57:48.420 | Why this particular moment in history?

01:57:50.340 | - Oh, because enough time has passed.

01:57:52.220 | I mean, the very first race riots were in Boston,

01:57:54.780 | not to draw anything-- - Really?

01:57:57.380 | When?

01:57:58.220 | - Oh. - This is before--

01:57:59.060 | - Going way, I mean, like the 1700s or whatever, right?

01:58:01.500 | I mean, there was a massive one in New York.

01:58:03.700 | I mean, I'm talking way, way, way back when.

01:58:05.380 | So Boston used to be the hotbed of riots.

01:58:07.900 | It's just what Boston was all about,

01:58:10.220 | or so I'm told from history class.

01:58:12.420 | There's an interesting one in New York.

01:58:14.860 | I don't remember when that was.

01:58:15.700 | Anyway, the point is,

01:58:17.380 | basically you gotta get another generation,

01:58:23.140 | old enough to be angry, but not so old

01:58:25.260 | to remember what happened the last time, right?

01:58:28.140 | And that's sort of what happens.

01:58:29.380 | But you said like two completely,

01:58:32.420 | you said two things there that I think are worth unpacking.

01:58:35.300 | One has to do with this sort of moment in time,

01:58:38.740 | and why, why is this sort of upbuilt?

01:58:43.620 | And the other has to do with a kind of,

01:58:45.620 | sort of the economic reality of COVID.

01:58:47.980 | So I'm actually, I want to separate those things,

01:58:49.940 | because for example, you know,

01:58:51.700 | this happened before COVID happened, right?

01:58:55.420 | So let's separate these two things for a moment.

01:58:57.820 | Now, let me preface all this by saying

01:59:01.140 | that although I am interested in history,

01:59:04.600 | one of my three minors as an undergrad was history,

01:59:07.660 | specifically history of the 1960s.

01:59:09.360 | - Interesting. - The other was Spanish.

01:59:12.260 | - Okay, that's a mistake.

01:59:14.060 | - Oh, I loved, I loved. - Okay.

01:59:16.380 | - And history of, and Spanish history, actually,

01:59:18.180 | but Spanish, and the other was

01:59:19.820 | what we would now call cognitive science,

01:59:21.140 | but at the time--

01:59:22.820 | - Oh, that's fascinating, interesting.

01:59:25.260 | - I minored in cocci here for grad school.

01:59:28.300 | That was really, that was really fascinating.

01:59:30.820 | It was a very different experience

01:59:32.060 | from all the computer science classes I'd been taking,

01:59:34.260 | even the cocci classes I was taking at an undergrad.

01:59:37.780 | Anyway, I'm not, I am a, I'm interested in history,

01:59:42.780 | but I'm hardly a historian, right?

01:59:45.060 | So, you know, forgive my, I will ask the audience

01:59:48.660 | to forgive my simplification.

01:59:50.500 | But I think the question that's always worth asking

01:59:56.620 | as opposed to, it's the same question,

01:59:59.320 | but a little different.

02:00:01.220 | Not why now, but why not before, right?

02:00:05.220 | So why the 1950s, '60s civil rights movement

02:00:09.940 | as opposed to the 1930s, 1940s?

02:00:11.620 | Well, first off, there was a civil rights movement

02:00:12.980 | in the '30s and '40s, it just wasn't of the same character

02:00:16.220 | or quite as well-known.

02:00:17.380 | Post-World War II, lots of interesting things

02:00:19.820 | were happening.

02:00:20.660 | It's not as if a switch was turned on

02:00:22.700 | and Brown v. the Board of Education

02:00:24.940 | or the Montgomery Bus Boycott,

02:00:27.500 | and that's when it happened.

02:00:28.340 | These things have been building up forever

02:00:29.500 | and go all the way back and all the way back

02:00:30.980 | and all the way back, and, you know,

02:00:32.100 | Harriet Tubman was not born in 1950, right?

02:00:34.580 | So, you know, we can take these things--

02:00:36.100 | - It could have easily happened right after World War II.

02:00:38.900 | - Yes, I think, and again, I am not a scholar,

02:00:43.900 | I think that the big difference was TV.

02:00:48.100 | These things are visible.

02:00:50.860 | People can see them.

02:00:52.580 | It's hard to avoid, right?

02:00:55.580 | Why not James Farmer?

02:00:56.860 | Why Martin Luther King?

02:00:57.980 | 'Cause one was born 20 years after the other, whatever.

02:01:02.700 | I think it turns out that, you know what King's

02:01:06.180 | biggest failure was in the early days?

02:01:08.780 | It was in Georgia.

02:01:09.680 | You know, they were doing the usual thing,

02:01:14.320 | trying to integrate, and I forget the guy's name,

02:01:18.460 | but you can look this up, but he, a cop,

02:01:21.820 | he was a sheriff, made a deal with the whole state of Georgia.

02:01:24.980 | We're gonna take people and we are going to

02:01:26.500 | non-violently put them in trucks,

02:01:28.820 | and then we are going to take them

02:01:30.140 | and put them in jails very far away from here.

02:01:33.280 | And we're gonna do that, and we're not gonna,

02:01:35.200 | there'll be no reason for the press to hang around.

02:01:37.720 | And they did that, and it worked.

02:01:40.160 | And the press left, and nothing changed.

02:01:43.200 | So, next they went to Birmingham, Alabama,

02:01:45.240 | and Bull O'Connor, and you got to see on TV

02:01:49.240 | little boys and girls being hit with fire hoses

02:01:52.500 | and being knocked down, and there was outrage,

02:01:54.560 | and things changed, right?

02:01:56.260 | Part of the delusion is pretending that nothing bad

02:01:59.820 | is happening that might force you to do something big

02:02:01.680 | you don't want to do, but sometimes it gets put in your face

02:02:04.200 | and then you kind of can't ignore it.

02:02:05.760 | And a large part, in my view, of what happened, right,

02:02:09.840 | was that it was too public to ignore.

02:02:12.000 | Now, we created other ways of ignoring it.

02:02:14.640 | Lots of change happened in the South,

02:02:16.080 | but part of that delusion was that it wasn't gonna affect

02:02:17.860 | the West or the Northeast, and of course it did,

02:02:20.280 | and that caused its own set of problems,

02:02:21.740 | which went into the late '60s into the '70s,

02:02:23.880 | and in some ways we're living with that legacy now,

02:02:26.440 | and so on.

02:02:28.420 | So, why not, what's happening now?

02:02:30.600 | Why it didn't happen 10 years ago?

02:02:32.600 | I think it's people have more voices,

02:02:35.240 | there's not just more TV, there's social media,

02:02:37.480 | it's very easy for these things

02:02:38.760 | to kind of build on themselves,

02:02:40.320 | and things are just quite visible.

02:02:43.500 | And there's demographic change,

02:02:45.920 | I mean, the world is changing rapidly, right?

02:02:47.600 | And so it's very difficult.

02:02:49.200 | You're now seeing people you could have avoided seeing

02:02:50.920 | most of your life growing up in a particular time,

02:02:53.960 | and it's happening, it's dispersing at a speed

02:02:56.920 | that is fast enough to cause concern for some people,

02:03:00.080 | but not so fast to cause massive negative reaction.

02:03:03.240 | So that's that.

02:03:06.120 | On the other hand, and again,

02:03:07.720 | that's a massive oversimplification,

02:03:09.360 | but I think there's something there anyway,

02:03:11.000 | at least something worth exploring.

02:03:11.880 | I'm happy to be yelled at by a real historian.

02:03:14.080 | - Oh yeah, I mean, there's just the obvious thing,

02:03:17.400 | I mean, I guess you're implying, but not saying this,

02:03:20.480 | I mean, it seemed to have percolated the most

02:03:22.760 | with just a single video, for example,

02:03:24.600 | the George Floyd video.

02:03:26.160 | - It makes a huge difference.

02:03:27.960 | It's fascinating to think that whatever the mechanisms

02:03:32.600 | that put injustice in front of our face,

02:03:36.480 | not like directly in front of our face,

02:03:40.320 | those mechanisms are the mechanisms of change.

02:03:43.000 | - Yeah, on the other hand, Rodney King.

02:03:44.640 | So no one remembers this.

02:03:46.160 | I seem to be the only person who remembers this,

02:03:47.680 | but sometime before the Rodney King incident,

02:03:50.560 | there was a guy who was a police officer

02:03:52.640 | who was saying that things were really bad

02:03:55.600 | in Southern California, and he was gonna prove it

02:03:59.040 | by having some news, some camera people follow him around.

02:04:02.840 | And he says, "I'm gonna go into these towns

02:04:04.320 | "and just follow me for a week,

02:04:05.400 | "and you will see that I'll get harassed."

02:04:06.920 | And like the first night, he goes out there,

02:04:09.480 | he crosses into the city, some cops pull him over,

02:04:11.920 | and he's a police officer, remember.

02:04:14.420 | They don't know that, of course.

02:04:15.400 | They like shove his face through a glass window.

02:04:17.800 | This was on the news,

02:04:18.640 | like I distinctly remember watching this as a kid.

02:04:21.620 | Actually, I guess I wasn't a kid,

02:04:22.520 | I was in college at the time, I was in grad school at the time.

02:04:25.200 | - So that's not enough, like just--

02:04:27.880 | - Well, it disappeared.

02:04:28.720 | Like a day later, it didn't go viral.

02:04:30.740 | - Whatever that is, whatever that magic thing is.

02:04:34.040 | - And whatever it was in '92,

02:04:35.320 | it was harder to go viral in '92, right?

02:04:37.680 | Or '91, actually it must have been '90 or '91.

02:04:40.160 | But that happened, and like two days later,

02:04:42.160 | it's like it never happened.

02:04:43.000 | Like nobody, again, nobody remembers this,

02:04:44.720 | but I'm like the only person.

02:04:45.560 | Sometimes I think I must have dreamed it.

02:04:46.920 | Anyway, Rodney King happens, it goes viral,

02:04:50.200 | or the moral equivalent thereof at the time.

02:04:53.160 | And eventually, we get April 29th, right?

02:04:56.240 | And I don't know what the difference was

02:04:58.600 | between the two things,

02:04:59.440 | other than one thing caught on and one thing didn't.

02:05:01.440 | Maybe what's happening now

02:05:03.280 | is two things are feeding onto one another.

02:05:06.000 | One is more people are willing to believe.

02:05:09.200 | And the other is there's easier and easier ways

02:05:11.800 | to give evidence.

02:05:13.060 | Cameras, body cams, or whatever.

02:05:15.480 | But we're still finding ourselves telling the same story.

02:05:17.440 | It's the same thing over and over again.

02:05:18.520 | I would invite you to go back and read the op-eds

02:05:21.520 | from what people were saying about

02:05:23.640 | the violence is not the right answer after Rodney King.

02:05:27.280 | And then go back to 1980 and the big riots

02:05:29.040 | that were happening around then,

02:05:30.320 | and read the same op-ed.

02:05:31.720 | It's the same words over and over and over again.

02:05:35.200 | I mean, there's your remembering history right there.

02:05:37.720 | I mean, it's like literally the same words.

02:05:39.080 | Like you could have just caught it,

02:05:39.920 | and I'm surprised no one got flagged for plagiarism.

02:05:43.160 | - It's interesting if you have an opinion

02:05:45.000 | on the question of violence,

02:05:46.720 | and the popular, perhaps, caricature

02:05:49.040 | of Malcolm X versus Martin Luther King.

02:05:53.320 | - You know Malcolm X was older than Martin Luther King?

02:05:55.800 | People kind of have it in their head that he's younger.

02:05:58.040 | Well, he died sooner, right?

02:06:00.960 | But only by a few years, right?

02:06:03.040 | People think of MLK as the older statesman,

02:06:05.640 | and they think of Malcolm X as the young, angry, whatever.

02:06:09.080 | But that's more of a narrative device.

02:06:11.280 | It's not true at all.

02:06:12.500 | I don't, I just, I reject the choice.

02:06:18.480 | I think it's a false choice.

02:06:19.640 | I think they're just things that happen.

02:06:20.880 | You just do, as I said, hatred is not,

02:06:23.540 | it takes a lot of energy.

02:06:25.240 | But every once in a while you have to fight.

02:06:27.320 | One thing I will say, without taking a moral position,

02:06:31.620 | which I will not take on this matter,

02:06:34.440 | violence has worked.

02:06:39.040 | - Yeah, that's the annoying thing.

02:06:42.480 | - That's the annoying thing.

02:06:43.680 | - It seems like over-the-top anger works.

02:06:48.440 | Outrage works.

02:06:50.000 | So you can say being calm and rational,

02:06:53.760 | just talking it out is gonna lead to progress,

02:06:57.120 | but it seems like if you just look through history,

02:07:00.740 | being irrationally upset is the way you make progress.

02:07:06.740 | - Well, it's certainly the way that you

02:07:08.600 | get someone to notice you.

02:07:09.800 | - Yeah, and that's--

02:07:11.240 | - And if they don't notice you,

02:07:12.080 | I mean, what's the difference between that and what,

02:07:13.880 | again, without taking a moral position on this,

02:07:15.600 | I'm just trying to observe history here.

02:07:17.120 | If you, maybe if television didn't exist,

02:07:20.520 | the civil rights movement doesn't happen,

02:07:22.360 | or it takes longer, or it takes a very different form.

02:07:25.000 | Maybe if social media doesn't exist,

02:07:27.220 | a whole host of things, positive and negative,

02:07:29.640 | don't happen, right?

02:07:31.160 | So, and what do any of those things do

02:07:33.720 | other than expose things to people?

02:07:38.320 | Violence is a way of shouting.

02:07:40.800 | I mean, many people far more talented and thoughtful

02:07:43.400 | than I have have said this in one form or another, right?

02:07:46.000 | That violence is the voice of the unheard, right?

02:07:49.880 | I mean, it's a thing that people do

02:07:53.400 | when they feel as if they have no other option.

02:07:56.000 | And sometimes we agree, and sometimes we disagree.

02:07:59.680 | Sometimes we think they're justified.

02:08:01.040 | Sometimes we think they are not.

02:08:03.460 | But regardless, it is a way of shouting.

02:08:06.440 | And when you shout, people tend to hear you,

02:08:08.800 | even if they don't necessarily hear

02:08:10.000 | the words that you're saying.

02:08:10.880 | They hear that you were shouting.

02:08:12.800 | I see no way.

02:08:14.360 | So another way of putting it, which I think is less,

02:08:16.760 | let us just say, provocative, but I think is true,

02:08:21.760 | is that all change, particularly change

02:08:27.160 | that impacts power, requires struggle.

02:08:30.060 | The struggle doesn't have to be violent.

02:08:33.320 | But it's a struggle nonetheless.

02:08:37.440 | - The powerful don't give up power easily.

02:08:40.320 | - I mean, why should they?

02:08:43.440 | But even so, it still has to be a struggle.

02:08:45.000 | And by the way, this isn't just about violent, political,

02:08:48.180 | whatever, nonviolent political change, right?

02:08:49.920 | This is true for understanding calculus, right?

02:08:52.240 | I mean, everything requires a struggle.

02:08:53.880 | - We're back to talking about faculty hiring.

02:08:55.560 | - At the end of the day, in the end of the day,

02:08:57.160 | it all comes down to faculty hiring.

02:08:58.520 | - And the godfather scene.

02:09:00.360 | - All a metaphor.

02:09:01.400 | Faculty hiring is a metaphor for all of life.

02:09:04.120 | - Let me ask a strange question.

02:09:07.040 | Do you think everything is gonna be okay in the next year?

02:09:13.240 | Do you have a hope that we're gonna be okay?

02:09:16.760 | - I tend to think that everything's gonna be okay,

02:09:18.860 | because I just tend to think

02:09:20.120 | that everything's gonna be okay.

02:09:21.800 | My mother says something to me a lot, and always has,

02:09:25.400 | and I find it quite comforting, which is,

02:09:27.000 | this too shall pass.

02:09:28.600 | And this too shall pass.

02:09:30.480 | Now, this too shall pass is not just

02:09:32.520 | this bad thing is going away.

02:09:35.760 | Everything passes.

02:09:36.760 | I mean, I have a 16-year-old daughter

02:09:38.560 | who's going to go to college,

02:09:41.100 | probably in about 15 minutes,

02:09:42.480 | given how fast she seems to be growing up.

02:09:44.600 | And I get to hang out with her now,

02:09:46.040 | but one day I won't.

02:09:47.560 | She'll ignore me just as much as I ignored my parents

02:09:50.160 | when I was in college and went to grad school.

02:09:51.560 | This too shall pass.

02:09:52.400 | But I think that one day, if we're all lucky,

02:09:55.920 | you live long enough to look back on something

02:09:57.640 | that happened a while ago, even if it was painful,

02:09:59.520 | and mostly, it's a memory.

02:10:02.260 | So yes, I think it'll be okay.

02:10:06.260 | - What about humans?

02:10:07.320 | Do you think we'll live into the 21st century?

02:10:11.520 | - I certainly hope so.

02:10:12.720 | - Are you worried about,

02:10:14.880 | are you worried that we might destroy ourselves

02:10:16.720 | with nuclear weapons, with AGI, with engineering?

02:10:20.080 | - I'm not worried about AGI doing it,

02:10:21.400 | but I am worried, I mean, at any given moment, right?

02:10:23.840 | Also, but you know, at any given moment, a comet could,

02:10:26.040 | I mean, you know, whatever.

02:10:27.320 | I tend to think that outside of things

02:10:30.440 | completely beyond our control,

02:10:32.300 | we have a better chance than not of making it.

02:10:36.960 | - You know, I talked to Alex Villepenco from Berkeley.

02:10:40.360 | He was talking about comets

02:10:42.000 | and that they can come out of nowhere,

02:10:43.540 | and that was a realization to me.

02:10:46.840 | Wow, we're just watching this darkness,

02:10:50.100 | and they can just enter, and then we have less than a month.

02:10:53.320 | - Yeah, and yet, you make it from day to day.

02:10:56.260 | - That one shall not pass.

02:10:59.340 | Well, maybe for Earth it'll pass, but not for humans.

02:11:02.240 | - But I'm just choosing to believe

02:11:04.320 | that it's going to be okay,

02:11:07.480 | and we're not gonna get hit by an asteroid,

02:11:09.620 | at least not while I'm around, and if we are,

02:11:11.840 | well, there's very little I can do about it,

02:11:13.560 | so I might as well assume it's not going to happen.

02:11:16.280 | - It makes food taste better.

02:11:17.800 | - It makes food taste better.

02:11:19.720 | - So you, out of the millions of things

02:11:22.960 | you've done in your life, you've also began

02:11:25.400 | the This Week in Black History calendar of facts.

02:11:28.860 | There's like a million questions I can ask here.

02:11:33.360 | You said you're not a historian, but is there,

02:11:38.200 | let's start at the big history question of,

02:11:41.680 | is there somebody in history, in black history,

02:11:45.260 | that you draw a lot of philosophical

02:11:49.220 | or personal inspiration from, or you just find interesting,

02:11:52.820 | or a moment in history you find interesting?

02:11:55.180 | - Well, I find the entirety of the '40s and the '60s

02:11:59.060 | and the civil rights movement that didn't happen

02:12:01.500 | and did happen at the same time during then

02:12:03.460 | quite inspirational.

02:12:04.540 | I mean, I've read quite a bit of the time period,

02:12:08.020 | at least I did in my younger days

02:12:10.000 | when I had more time to read as many things as I wanted to.

02:12:12.900 | What was quirky about This Week in Black History

02:12:17.140 | when I started in the '80s was how focused it was.

02:12:22.140 | It was because of the sources I was stealing from,

02:12:24.480 | and I was very much stealing from,

02:12:25.640 | so I'd take calendars, anything I could find,

02:12:28.160 | Google didn't exist, right, and I just pulled

02:12:29.600 | as much as I could and just put it together

02:12:30.960 | in one place for other people.

02:12:32.560 | What ended up being quirky about it,

02:12:33.760 | and I started getting people sending me information,

02:12:36.160 | was the inventors, people who,

02:12:40.240 | Garrett Morgan to Benjamin Banneker,

02:12:42.520 | people who were inventing things

02:12:45.040 | at a time when, how in the world

02:12:52.460 | did they manage to invent anything?

02:12:54.160 | Like, all these other things were happening,

02:12:56.280 | mother necessity, right, all these other things

02:12:57.720 | were happening, and there were so many

02:12:59.460 | terrible things happening around them,

02:13:00.760 | and they went to the wrong state at the wrong time,

02:13:02.680 | they may never come back,

02:13:04.160 | but they were inventing things we use, right?

02:13:06.360 | And it was always inspiring to me

02:13:09.080 | that people would still create,

02:13:11.400 | even under those circumstances.

02:13:14.560 | I got a lot out of that.

02:13:15.400 | I also learned a few lessons, I think,

02:13:17.440 | you know, the Charles Richard Drews of the world.

02:13:19.200 | You know, you create things that impact people,

02:13:24.200 | you don't necessarily get credit for them,

02:13:26.500 | and that's not right, but it's also okay.

02:13:30.460 | - You're okay with that?

02:13:31.640 | - Up to a point, yeah.

02:13:33.740 | I mean, look, in our world, all we really have is credit.

02:13:38.740 | - I was always bothered by how much value credit is given.

02:13:44.060 | - That's the only thing you got.

02:13:45.400 | I mean, if you're an academic in some sense,

02:13:47.140 | well, it isn't the only thing you've got,

02:13:48.240 | but it feels that way sometimes.

02:13:49.780 | - But you got the actual, we're all gonna be dead soon.

02:13:53.800 | You got the joy of having created.

02:13:56.520 | You know, the credit with Jan,

02:14:00.780 | I've talked to Jorgen Schmidhuber, right?

02:14:02.980 | The Turing Award given to three people for deep learning,

02:14:07.820 | and you could say that a lot of other people

02:14:10.580 | should be on that list.

02:14:11.660 | It's the Nobel Prize question.

02:14:13.700 | Yeah, it's sad.

02:14:14.820 | It's sad, and people like talking about it,

02:14:16.560 | but I feel like in the long arc of history,

02:14:20.300 | the only person who'll be remembered

02:14:21.860 | is Einstein, Hitler, maybe Elon Musk.

02:14:24.300 | And the rest of us are just like.

02:14:27.500 | - Well, you know, someone asked me about immortality once,

02:14:29.660 | and I said, and I stole this from somebody else,

02:14:32.460 | I don't remember who, but it was, you know,

02:14:35.420 | I asked him, "What's your great-grandfather's name?"

02:14:37.940 | Any of them.

02:14:38.780 | Of course, they don't know.

02:14:40.780 | Most of us do not know.

02:14:42.540 | I mean, I'm not entirely sure I know my grandparents' names,

02:14:45.020 | all my grandparents' names.

02:14:45.940 | I know what I called them, right?

02:14:47.700 | I don't know their middle names, for example.

02:14:49.800 | Didn't live in living memory, so I could find out.

02:14:53.900 | Actually, my grandfather didn't know when he was born.

02:14:56.540 | Had no idea how old he was, right?

02:14:59.180 | But I definitely don't know

02:15:00.620 | who any of my great-grandparents are.

02:15:02.740 | So in some sense, immortality is doing something,

02:15:06.020 | preferably positive, so that your great-grandchildren

02:15:08.500 | know who you are, right?

02:15:10.740 | And that's kind of what you can hope for,

02:15:12.260 | which is very depressing in some ways.

02:15:14.100 | I could turn it into something uplifting

02:15:16.860 | if you need me to, but it's--

02:15:17.860 | - Yeah, can you do the work here?

02:15:19.260 | - Yeah, it's simple, right?

02:15:21.180 | It doesn't matter.

02:15:22.100 | I don't have to know who my great-grandfather was

02:15:23.980 | to know that I wouldn't be here without him.

02:15:25.980 | - Yeah.

02:15:26.820 | - And I don't know who my great-grandchildren are,

02:15:29.300 | certainly who my great-great-grandchildren are,

02:15:31.180 | and I'll probably never meet them,

02:15:32.900 | although I would very much like to.

02:15:35.060 | But hopefully I'll set the world in motion

02:15:37.900 | in such a way that their lives will be better

02:15:39.700 | than they would have been if I hadn't done that.

02:15:41.340 | Well, certainly they wouldn't have existed

02:15:42.820 | if I hadn't done the things that I did.

02:15:44.420 | So I think that's a good positive thing.

02:15:47.060 | You live on through other people.

02:15:49.500 | - Are you afraid of death?

02:15:51.380 | - I don't know if I'm afraid of death, but I don't like it.

02:15:53.620 | (laughing)

02:15:55.380 | - Another T-shirt.

02:15:56.780 | (laughing)

02:15:58.500 | I mean, do you ponder it?

02:15:59.820 | Do you think about the--

02:16:01.660 | - Yes, the inevitability of oblivion?

02:16:04.420 | - Yes.

02:16:05.260 | - I do occasionally.

02:16:06.500 | This feels like a very Russian conversation, actually.

02:16:08.340 | - It's very, yeah.

02:16:09.740 | - I will tell you a story,

02:16:10.660 | a very, something that happened to me recently.

02:16:13.460 | If you look very carefully, you will see I have a scar.

02:16:15.980 | - Yes.

02:16:17.580 | - Which, by the way, is an interesting story of its own

02:16:20.000 | about why people who have half of their thyroid taken out,

02:16:22.020 | some people get scars and some don't.

02:16:24.300 | But anyway, I had half my thyroid taken out.

02:16:28.100 | The way I got there, by the way,

02:16:29.060 | is its own interesting story, but I won't go into it.

02:16:30.780 | Just suffice it to say, I did what I keep telling people

02:16:33.020 | you should never do, which is never go to the doctor

02:16:34.500 | unless you have to, because there's nothing good

02:16:36.280 | that's ever gonna come out of a doctor's visit, right?

02:16:38.000 | So I went to the doctor to look at one thing,

02:16:40.660 | this little bump I had on the side

02:16:41.940 | that I thought might be something bad

02:16:43.980 | because my mother made me, and I went there,

02:16:45.740 | and he's like, "Oh, it's nothing,

02:16:46.780 | "but by the way, your thyroid is huge.

02:16:48.380 | "Can you breathe?"

02:16:49.220 | "Yes, I can breathe."

02:16:50.040 | "Are you sure?

02:16:50.880 | "'Cause it's pushing on your windpipe.

02:16:51.720 | "You should be dead."

02:16:52.540 | "Ah!"

02:16:53.380 | So I was sitting there, and to look at my thyroid,

02:16:57.740 | it was growing.

02:16:58.820 | I had what's called a goiter,

02:17:00.220 | and he said, "We're gonna have to take it out

02:17:02.060 | "at some point."

02:17:02.900 | "When?"

02:17:03.720 | "Sometime before you're 85, probably,

02:17:05.420 | "but if you wait 'til you're 85, that'll be really bad

02:17:08.260 | "because you don't wanna have surgery

02:17:10.420 | "when you're 85 years old, if you can help it."

02:17:13.160 | Certainly not the kind of surgery it takes

02:17:14.600 | to take out your thyroid.

02:17:16.080 | So I went there, and we decided,

02:17:19.340 | I would decide I would put it off until December 19th

02:17:23.500 | because my birthday's December 18th,

02:17:25.180 | and I wouldn't be able to say I made it to 49 or whatever,

02:17:27.940 | so I said, "I'll wait 'til after my birthday."

02:17:30.220 | In the first six months of that, nothing changed.

02:17:34.980 | Apparently, in the next three months,

02:17:37.000 | it had grown, I hadn't noticed this at all.

02:17:40.380 | I went and had surgery.

02:17:42.660 | They took out half of it.

02:17:43.840 | The other half is still there,

02:17:44.860 | and it's working fine, by the way.

02:17:46.260 | I don't have to take a pill or anything like that.

02:17:48.060 | It's great.

02:17:49.740 | I'm in the hospital room, and the doctor comes in.

02:17:54.740 | I've got these things on my arm.

02:17:56.900 | They're gonna do whatever.

02:17:58.060 | They're talking to me, and the anesthesiologist says,

02:18:01.200 | "Huh, your blood pressure's through the roof.

02:18:02.500 | "Do you have high blood pressure?"

02:18:04.260 | I said, "No, but I'm terrified if that helps you at all."

02:18:08.100 | And the anesthetist, who's the nurse

02:18:10.220 | who supports the anesthesiologist, if I got that right,

02:18:13.460 | said, "Oh, don't worry about it.

02:18:14.300 | "I just put some stuff in your IV.

02:18:16.000 | "You're gonna be feeling pretty good in a couple minutes."

02:18:17.660 | And I remember turning and saying,

02:18:20.240 | "Well, I'm gonna feel pretty good in a couple minutes."

02:18:21.940 | Next thing I know, there's this guy,

02:18:24.700 | and he's moving my bed.

02:18:26.140 | And he's talking to me.

02:18:28.260 | I have this distinct impression that I've met this guy,

02:18:31.260 | and I should know what he's talking about,

02:18:34.060 | but I kind of just don't remember what just happened.

02:18:37.860 | And I look up, and I see the tiles going by,

02:18:39.700 | and I'm like, "Oh, it's just like in the movies

02:18:42.260 | "where you see the tiles go by."

02:18:43.980 | And then I have this brief thought

02:18:47.660 | that I'm in an infinitely long warehouse,

02:18:49.860 | and there's someone sitting next to me.

02:18:52.840 | And I remember thinking, "Oh, she's not talking to me."

02:18:54.980 | And then I'm back in the hospital bed.

02:18:57.260 | And in between the time where the tiles were going by

02:19:02.340 | and I got in the hospital bed,

02:19:03.700 | something like five hours had passed.

02:19:05.580 | Apparently, it had grown so much

02:19:07.020 | that it was a four and a half hour procedure

02:19:08.740 | instead of an hour long procedure.

02:19:10.380 | I lost a neck size and a half.

02:19:14.420 | It was pretty big.

02:19:15.260 | Apparently, it was as big as my heart.

02:19:17.380 | Why am I telling you this?

02:19:18.500 | I'm telling you this because--

02:19:20.460 | - It's a hell of a story already, so--

02:19:21.780 | - Between the tiles going by

02:19:24.660 | and me waking up in my hospital bed, no time passed.

02:19:29.260 | There was no sensation of time passing.

02:19:32.060 | When I go to sleep and I wake up in the morning,

02:19:35.020 | I have this feeling that time has passed,

02:19:36.780 | or this feeling that something

02:19:38.060 | has physically changed about me.

02:19:39.900 | Nothing happened between the time

02:19:42.180 | they put the magic juice in me

02:19:44.180 | and the time that I woke up, nothing.

02:19:46.500 | By the way, my wife was there with me talking.

02:19:48.860 | Apparently, I was also talking.

02:19:50.860 | I don't remember any of this,

02:19:52.340 | but luckily I didn't say anything I wouldn't normally say.

02:19:54.940 | My memory of it is I would talk to her

02:19:57.380 | and she would teleport around the room.

02:19:59.340 | And then I accused her of witchcraft

02:20:01.940 | and that was the end of that.

02:20:02.780 | But her point of view is I would start talking

02:20:05.820 | and then I would fall asleep

02:20:07.220 | and then I would wake up and leave off where I was before.

02:20:09.500 | I had no notion of any time passing.

02:20:11.300 | I kind of imagine that that's death.

02:20:14.380 | - Yeah.

02:20:16.660 | - Is the lack of sensation of time passing.

02:20:19.060 | And on the one hand, I am, I don't know,

02:20:23.700 | soothed by the idea that I won't notice.

02:20:27.060 | On the other hand, I'm very unhappy

02:20:28.780 | at the idea that I won't notice.

02:20:30.860 | So I don't know if I'm afraid of death,

02:20:34.620 | but I'm completely sure that I don't like it

02:20:36.820 | and that I particularly would prefer

02:20:39.180 | to discover on my own whether immortality sucks

02:20:42.860 | and be able to make a decision about it.

02:20:44.780 | That's what I would prefer.

02:20:46.260 | - You'd like to have a choice in the matter.

02:20:48.420 | - I would like to have a choice in the matter.

02:20:49.900 | - Well, again, on the Russian thing,

02:20:51.500 | I think the finiteness of it is the thing

02:20:54.060 | that gives it a little flavor, a little spice.

02:20:56.980 | So, immortality.

02:20:57.820 | - Well, in reinforcement learning, we believe that.

02:20:59.260 | That's why we have discount factors.

02:21:00.420 | Otherwise, it doesn't matter what you do.

02:21:02.740 | - Amen.

02:21:03.620 | Well, let me, one last question

02:21:06.580 | sticking on the Russian theme.

02:21:09.360 | You talked about your great-grandparents

02:21:13.680 | not remembering their name.

02:21:15.760 | What do you think is the,

02:21:17.000 | in this kind of Markov chain that is life,

02:21:23.160 | what do you think is the meaning of it all?

02:21:27.000 | What's the meaning of life?

02:21:28.440 | - Well, in a world where eventually

02:21:31.120 | you won't know who your great-grandchildren are,

02:21:35.240 | I'm reminded of something I heard once,

02:21:39.240 | or I read once that I really like,

02:21:42.120 | which is it is well worth remembering

02:21:45.320 | that the entire universe, save for one trifling exception,

02:21:52.480 | is composed entirely of others.

02:21:54.440 | I think that's the meaning of life.

02:21:58.880 | - Charles, this was one of the best conversations

02:22:03.600 | I've ever had, and I get to see you tomorrow again

02:22:06.640 | to hang out with who looks to be one of the most,

02:22:11.640 | how should I say, interesting personalities

02:22:15.760 | that I'll ever get to meet with Michael Lipman.

02:22:17.520 | So I can't wait.

02:22:19.180 | I'm excited to have had this opportunity.

02:22:21.440 | Thank you for traveling all the way here.

02:22:23.160 | It was amazing.

02:22:24.780 | I'm excited.

02:22:25.620 | I always loved Georgia Tech.

02:22:26.760 | I'm excited to see with you being involved there

02:22:29.800 | what the future holds.

02:22:30.640 | So thank you for talking today.

02:22:31.800 | - Thank you for having me.

02:22:32.640 | I appreciate every minute of it.

02:22:34.280 | - Thanks for listening to this conversation

02:22:35.720 | with Charles Isbell, and thank you to our sponsors,

02:22:39.000 | Neuro, the maker of functional sugar-free gum and mints

02:22:42.600 | that I use to give my brain a quick caffeine boost,

02:22:46.480 | Decoding Digital, a podcast on tech and entrepreneurship

02:22:49.960 | that I listen to and enjoy, Masterclass,

02:22:53.000 | online courses that I watch from some

02:22:55.620 | of the most amazing humans in history,

02:22:57.900 | and Cash App, the app I use to send money

02:23:00.680 | to friends for food and drinks.

02:23:03.680 | Please check out these sponsors in the description

02:23:06.000 | to get a discount and to support this podcast.

02:23:09.360 | If you enjoy this thing, subscribe on YouTube,

02:23:11.680 | review it with Five Stars on Apple Podcasts,

02:23:13.800 | follow on Spotify, support on Patreon,

02:23:16.360 | or connect with me on Twitter @LexFriedman.

02:23:19.600 | And now let me leave you with some poetic words

02:23:22.820 | from Martin Luther King Jr.

02:23:24.540 | There comes a time when people get tired

02:23:28.600 | of being pushed out of the glittering sunlight

02:23:30.840 | of life's July and left standing amid the piercing chill

02:23:35.680 | of an alpine November.

02:23:37.620 | Thank you for listening, and hope to see you next time.

02:23:41.720 | (upbeat music)

02:23:44.300 | (upbeat music)

02:23:46.880 | [BLANK_AUDIO]

Charles Isbell: Computing, Interactive AI, and Race in America | Lex Fridman Podcast #135

Chapters