AMA: career advice given AGI, how I research ft. Sholto & Trenton

1. 0:00 – 4:57

   Book launch announcement

   1. DPDwarkesh Patel0:00

      Today, this is going to be an Ask Me Anything episode. I'm joined with my friends, Trenton Brickin and Sholto Douglas. You guys do some AI stuff, right? (laughs)

   2. TBTrenton Bricken0:08

      Yeah. A dabble. (laughs)

   3. DPDwarkesh Patel0:12

      (laughs) Um, they're researchers at Anthropic. Other news, I have a book launching today, it's called The Scaling Era.

   4. TBTrenton Bricken0:17

      Yeah.

   5. DPDwarkesh Patel0:17

      Um, I hope one of the questions ends up being, why you should buy this book. (laughs)

   6. TBTrenton Bricken0:23

      (laughs)

   7. DPDwarkesh Patel0:23

      But we can kill two birds with one stone. But, um, okay, let's just get at it. What's the first question that we gotta answer?

   8. TBTrenton Bricken0:28

      Take us away.

   9. SDSholto Douglas0:29

      So, I wanna ask the flyball question that I heard before of, why should ordinary people care about this book? Like, like, why should my mom buy and read the book?

   10. DPDwarkesh Patel0:38

       Yeah. First, let me tell you about the book, what it is. So, you know, these last few years, I've been interviewing AI lab CEOs, researchers, people like you guys obviously, but also scholars from all kinds of different fields, economists, philosophers. And they've been addressing I think what are basically the gnarliest, most interesting, most important questions we've ever had to ask ourselves. Like, what is the fundamental nature of intelligence? What will happen when we have billions of extra workers? How do we eco- how do we model out the economics of that? Um, how do we think about an intelligence that is greater than the rest of humanity combined? Is that even a coherent concept? And so, what- what I'm super delighted with is that with Stripe Press, we made this book where we compiled and curated the best, most insightful snippets across all these interviews. And you can read Dario addressing, why does scaling work? And then on the sa- next page is Demis explaining, um, DeepMind's plans for whether they're gonna go with the RL route and how much of the AlphaZero stuff will play into the next generation of LLMs. And on the next pages is of course, you guys going through the technical details of how these models work. Um, and then there's so many different fields that are implicated. I mean, I feel like AI is one of the most multi-disciplinary fields that one can imagine because there's no field, no domain of human knowledge that is not relevant to understanding what a future society of different kinds of beings will look like. Um, there's, you know, you're gonna have like Carl Schulman talk about how the scaling hypothesis shows up in primate brain scaling from chimpanzees to humans. On the next page might be an economist trying to argue with like Tyler Cowen, explaining why he doesn't expect explosive economic growth, and why the bottlenecks will eat all that up. Um, anyways, so that's why your mom should buy this book. It just like, it is the distillation of all these different fields of human knowledge applied to the most important questions that humanity is facing right now.

   11. SDSholto Douglas2:40

       I- I do like how the book is sliced up by different topics-

   12. DPDwarkesh Patel2:43

       Yes.

   13. SDSholto Douglas2:43

       ... and across interviews.

   14. DPDwarkesh Patel2:45

       Yeah. Yeah.

   15. SDSholto Douglas2:45

       So, it- it does seem like a nice way to listen to all of the interviews-

   16. DPDwarkesh Patel2:49

       Yeah.

   17. SDSholto Douglas2:50

       ... in- in one digestible way.

   18. DPDwarkesh Patel2:52

       Yeah. There's, um, two interviews I've done that haven't been released publicly before that are in the book.

   19. SDSholto Douglas2:58

       Mm-hmm.

   20. DPDwarkesh Patel2:58

       So, one was that Jared Kaplan, um, uh, who's one of your co-founders, and this is another example where it's like th- he's like a physicist and he's explaining scaling from this like very mathematical perspective about data manifolds. Um, uh, and then on the same next page, you have like, a totally different perspective. It's like Goren talking about, you know, what- why- why- why can't we just have like distilled int- like, why did general intelligence actually evolve in the first place? What is the actual evolutionary purpose of it? Um, and it's like, page by page, right, you can just get addresses. Even for me, I mean, like the person who's been on the other end of these conversations, it was actually really cool to, um, like read it and just be like, "Oh, I actually like, I, now I realize how these insights connect to each other."

   21. SDSholto Douglas3:36

       Yeah. The, the only other thing that stood out to me as well is the introduction section-

   22. DPDwarkesh Patel3:39

       The only thing that stood out to you? (laughs)

   23. TBTrenton Bricken3:42

       (laughs)

   24. SDSholto Douglas3:42

       Like, like-

   25. DPDwarkesh Patel3:42

       Shh.

   26. TBTrenton Bricken3:42

       (laughs)

   27. SDSholto Douglas3:42

       Yeah, yeah, that was, that was really the only thing that was noteworthy.

   28. DPDwarkesh Patel3:45

       (laughs)

   29. SDSholto Douglas3:45

       Um, I just mean in ter- it stood out in accessibility.

   30. DPDwarkesh Patel3:49

       Yeah.
2. 4:57 – 10:52

   AI models not making connections across fields

   1. DPDwarkesh Patel4:57

      do it. Let's do it.

   2. TBTrenton Bricken4:57

      All right. Uh, so Brian Krav asks, "The issue you raised with Dario, and occasionally tweet about relating to models not making connections across disparate topics, some sort of combinatorial attention challenge. What are your thoughts on that now? Do you solve it with scale, thinking models or something else?"

   3. DPDwarkesh Patel5:11

      By the way, so the issue is, one of the questions I asked Dario is, look, these models have all of human knowledge memorized. And you would think if a human had this much stuff memorized, um, and they're moderately intelligent, they couldn't be making all these connections between different fields. And there are examples of humans doing this, by the way. Um, there's f- Donald Swann or something like this. This guy noticed that the- the way, what happens to a brain after magnesium deficiency is exactly, uh, the kinds of, I don't know, structure you see during a migraine. So, then he's like, "Okay, you take magnesium supplements and we're gonna cure a bunch of migraines." And it worked. And there's many other examples of things like this where you just like notice two different connections between pieces of knowledge. W- why, if these LLMs are intelligent, are they not able to use this unique advantage they have to make these kinds of discoveries?Um, I feel a little shy, like, me giving answers on AI shit- (laughs)

   4. NANarrator6:03

      (laughs)

   5. DPDwarkesh Patel6:03

      ... with you guys here. But, uh, uh, so actually, Scott Alexander addressed this question in one of his AMA threads, and he's like, "Look, humans also don't have this kind of logical omniscience," right? It's... He used the example of, in language, if you really thought about, like, why are two words connected? And it's like, oh, I understand, like, why, why rhyme has the same etymology as this other word, if you really thought about... But you just, like, don't think about it, right? There's, like, this combinatorial explosion. Um, I don't know if that addresses the fact that human, we know humans can do this, right? Like, the, humans have in fact done this, and I don't know of a single example of LLMs ever having done it. Um, actually, yeah, what, what is your answer to this?

   6. TBTrenton Bricken6:38

      I think my answer at the moment is that the sort of pre-training objective doesn't necessarily... Like, it imbues you with this, like, nice, flexible, general knowledge about the world, but doesn't necessarily imbue you with the, like the skill of making, like, con- novel connections or, like, research.

   7. DPDwarkesh Patel6:51

      Yeah.

   8. TBTrenton Bricken6:51

      The, the kinds of things that, uh, people are trained to do through PhD programs and through, like, sort of the process of exploring and interacting with the world. Um, and so I, I think, like, at a minimum, you need significant RL in at least similar things to, to be able to, like, approach, like, making novel discoveries. And so I would like to see some early evidence of this as we start to build models that are d- sort of interacting with one, trying to make scientific discoveries-

   9. DPDwarkesh Patel7:17

      Yeah.

   10. TBTrenton Bricken7:17

       ... and, like, and sort of, like, modeling the behaviors that we expect of people in these positions-

   11. DPDwarkesh Patel7:21

       Yeah.

   12. TBTrenton Bricken7:21

       ... because I don't actually think we've, we've done that in a, in like a s- meaningful or scaled way as a, as a field, so to speak.

   13. DPDwarkesh Patel7:27

       Mm-hmm.

   14. SDSholto Douglas7:27

       Yeah. R- riffing off that with respect to RL, I wonder if models currently just aren't good at knowing what memories they should be storing.

   15. DPDwarkesh Patel7:35

       Mm.

   16. SDSholto Douglas7:36

       Like, most of their training is just predicting the next word on the internet and remembering very specific facts from that. But if you were to teach me something new right now, I'm very aware of my own memory limitations, and so I would try to construct some summary-

   17. DPDwarkesh Patel7:51

       Mm-hmm.

   18. SDSholto Douglas7:51

       ... that would stick, and models cur- currently don't have the opportunity to do that. Memory scaffolding in general is just very-

   19. DPDwarkesh Patel7:58

       Hm.

   20. SDSholto Douglas7:59

       ... primitive right now. I-

   21. TBTrenton Bricken8:00

       Right, like, Claude plays Pokemon, right?

   22. DPDwarkesh Patel8:01

       Exactly, yeah.

   23. SDSholto Douglas8:02

       Yeah. Or, like, someone worked on it, it was awesome, it got far, but, uh, another excited Anthropic employee then, like, iterated on the memory scaffold and was able-

   24. DPDwarkesh Patel8:11

       Hm.

   25. SDSholto Douglas8:11

       ... to, like, very quickly improve on it.

   26. DPDwarkesh Patel8:13

       Interesting.

   27. SDSholto Douglas8:13

       Um, so yeah, that's one. I do also just wonder if models are idiot savants. Um, the best analogy might be to Kim Peek. So Kim Peek, uh, was born without a corpus callosum, if I recall correctly. Um, each hemisphere of his brain operated quite independently. Uh, he could read a page of a book. So he'd open a book, there'd be two pages visible. Each eye would read one of the pages.

   28. DPDwarkesh Patel8:37

       Huh.

   29. SDSholto Douglas8:38

       Uh, and he had, like, a perfect encyclopedic memory of, like, everything he'd ever read. Um, but at the same time, he had other debilitations.

   30. DPDwarkesh Patel8:46

       Yeah.
3. 10:52 – 15:20

   Career advice given AGI

   1. DPDwarkesh Patel10:52

   2. SDSholto Douglas10:52

      Uh, MaybeRabid, MaybeRabidMonkey, um-

   3. DPDwarkesh Patel10:58

      (laughs)

   4. SDSholto Douglas10:58

      ... asks, "Imagine you have a 17-year-old brother/nephew just starting college. What would you recommend he study given your AGI timelines?"

   5. DPDwarkesh Patel11:08

      I don't know. Become a podcaster? (laughs) I feel like that job's still gonna be around. It's funny because I, I studied computer science, and in retrospec... I mean, at the time, it was like, you could have become a software engineer or something. Instead, you became a podcaster. It's, like, kind of an irresponsible career move, but in retrospect, it's like... (laughs)

   6. TBTrenton Bricken11:23

      It's great. (laughs)

   7. DPDwarkesh Patel11:23

      It kinda worked out. Just as these guys are getting automated. (laughs)

   8. TBTrenton Bricken11:27

      I get asked this question all the time.

   9. DPDwarkesh Patel11:28

      Yes. Okay, go ahead.

   10. TBTrenton Bricken11:29

       Um, and one answer that I like to give is that you should think about the next couple of years as increasing your individual leverage by, like, a huge factor every year.

   11. DPDwarkesh Patel11:37

       Mm-hmm.

   12. TBTrenton Bricken11:37

       So, or, you know, already software engineers will come up and say, "You know, I'm two times faster," or, "In new languages, I'm five times faster than I was last year." Um, I expect that trend line to continue, basically-

   13. DPDwarkesh Patel11:46

       Right.

   14. TBTrenton Bricken11:46

       ... as you sort of go from this model of, well, I'm working with some model that's assisting me on my computer, and it's like, like, basically a pairing session, to I am managing a small team, uh, through to I'm managing, like, a division or a company-

   15. DPDwarkesh Patel12:00

       Yeah.

   16. TBTrenton Bricken12:00

       ... basically, that is, like, targeting a task. Um, and so I think that deep technical knowledge in fields will still matter-

   17. DPDwarkesh Patel12:08

       Yeah.

   18. TBTrenton Bricken12:08

       ... in four years.

   19. DPDwarkesh Patel12:09

       Yeah.

   20. TBTrenton Bricken12:09

       Like, it absolutely will-

   21. DPDwarkesh Patel12:10

       Yes.

   22. TBTrenton Bricken12:11

       ... because you will be in the position of managing dozens... Or, like, like, your sort of, your, your individual management bandwidth-... will be maxed out by trying to, to manage, like, teams of AIs-

   23. DPDwarkesh Patel12:21

       Yeah, yeah.

   24. TBTrenton Bricken12:21

       ... in this kind of thing. And maybe AIs... You know, maybe we end, end up in, like, a truly, like, singularity world, where you have AIs managing AIs, and this kind of stuff. But I think in a very wide part of the, like, possibility spectrum-

   25. DPDwarkesh Patel12:31

       Yes.

   26. TBTrenton Bricken12:31

       ... you are managing enormous, like, vastly more resources-

   27. DPDwarkesh Patel12:35

       Yeah.

   28. TBTrenton Bricken12:35

       ... than an individual could command today.

   29. DPDwarkesh Patel12:37

       Yeah, yeah.

   30. TBTrenton Bricken12:38

       Um, and you should be able to solve so many more things with that.
4. 15:20 – 17:19

   Guest selection criteria

   1. TBTrenton Bricken15:20

      What is your strategy, method, or criteria for choosing guests?

   2. DPDwarkesh Patel15:24

      The most important thing is, do I wanna spend one to two weeks reading every single thing you have ever written, every single interview you've ever recorded, talking to a bunch of other people about your research? Um, 'cause I get asked by people who are, like, quite influential often to be like, "Would you have me on your podcast?" Um, and more often than not, I say no, for two reasons. One is just, like... Like, wh- okay, you're influential or something, and just, like, wh- you're, it's not fundamentally that interesting as an interview prospect. N- not from, like, uh, I don't think about the hour that I'll spend with you. I think about, like, the two weeks, 'cause it's g- this is my life, right? The research is my life, and I, and I wanna have fun while doing it. Um, so just, like, is this gonna be an interesting two weeks to spend? Is it gonna help me with my future research or something? And the other is, big guests don't really matter that much in the, in the... Like, if you just look at what are the most popular episodes or, uh, what, in the long run, helps the podcast grow, um, by far, my most popular guest is Sarah Payne. And she, before I interviewed her, was just, like, a scholar, um, who was not publicly well- well-known at all, and I just found her books quite interesting. Same goes with... So my, my most popular guests are Sarah Payne, and then Sarah Payne, Sarah Payne, Sarah Payne, 'cause I did a lecture series with her.

   3. TBTrenton Bricken16:35

      (laughs)

   4. DPDwarkesh Patel16:35

      Um-

   5. SDSholto Douglas16:36

      (laughs) That's awesome.

   6. DPDwarkesh Patel16:37

      ... and by the way, from a viewer-minute adjusted basis, I host a Sarah Payne podcast where I occasionally talk about AI.

   7. TBTrenton Bricken16:44

      (laughs)

   8. DPDwarkesh Patel16:44

      (laughs)

   9. SDSholto Douglas16:44

      (laughs)

   10. DPDwarkesh Patel16:45

       Um, and then it's David Reich, who is a geneticist of ancient DNA. I mean, he- he's, like, somewhat well-known, but... He had a best-selling book, but he's not, like, he's not Satya Nadella or Mark Zuckerberg, who are the next people on the list. And then again, I think, at, like, pretty soon, it's, like, you guys or Leopold or something, and then you get to the last CEOs or something. So, big names just don't matter that much in the, like, for what I'm actually trying to do. And so it's just like, well... And it's also really hard to predict who's gonna be the David Reich or-

   11. TBTrenton Bricken17:09

       Mm-hmm.

   12. DPDwarkesh Patel17:09

       ... uh, Sarah Payne, so just, like, have fun. Who- talk to whoever you want to spend time researching, and it's a pretty good proxy for what will actually be popular.

   13. TBTrenton Bricken17:18

       What

5. 17:19 – 25:12

   Choosing to pursue the podcast long-term

   1. TBTrenton Bricken17:19

      was the specific moment, if there was one, that you realized you, that producing your podcast was a viable long-term strategy?

   2. DPDwarkesh Patel17:26

      I think when I was shopping around ad spots for a Mark Zuckerberg episode, and which were s- which are, you know, like, uh, and now when I look back on it, it's like, not in retrospect that mind-blowing, but at the time, I'm like, "Oh, I could actually hire an editor" (laughs)

   3. TBTrenton Bricken17:42

      (laughs)

   4. DPDwarkesh Patel17:42

      ... "full-time, or maybe more editors than one." Um, and from there, like, turning into a real business. Um, that's when I... 'Cause before I just, like, didn't... People would tell me, like, "Oh, these other podcasts are making whatever, whatever amount of money," and I'd be like, "How?" You know? (laughs)

   5. TBTrenton Bricken17:57

      (laughs)

   6. DPDwarkesh Patel17:57

      Um, I have this running joke with one of my friends that (laughs) ... I don't know if you've seen me do this, but every time I encounter, like, a young person who's like, "What should I do with my life?" I'm like, "You gotta start a blog. You gotta be the Matt Levine of AI. You can do this."

   7. TBTrenton Bricken18:09

      (laughs)

   8. DPDwarkesh Patel18:09

      That's like a-

   9. TBTrenton Bricken18:09

      Mm-hmm.

   10. DPDwarkesh Patel18:09

       ... a totally empty niche. You could, like, it's i- you, you can... um, and I have this running joke with them where they're like, "You're like a country bumpkin who's, like, won the lottery." And you're like-

   11. TBTrenton Bricken18:17

       (laughs)

   12. DPDwarkesh Patel18:17

       (laughs) You go up to everything, of course, and like-

   13. SDSholto Douglas18:19

       (laughs)

   14. TBTrenton Bricken18:19

       (laughs)

   15. DPDwarkesh Patel18:19

       ... "Guys, a scratch pad."

   16. TBTrenton Bricken18:20

       You have to love it.

   17. DPDwarkesh Patel18:20

       "Get the scratch pad." Yeah, yeah. (laughs)

   18. SDSholto Douglas18:22

       (laughs)

   19. TBTrenton Bricken18:22

       (laughs)

   20. DPDwarkesh Patel18:22

       Um...

   21. SDSholto Douglas18:24

       I, I, I, I do wanna press on that a bit more 'cause your immediate answer to the, uh, 17-year-old was to start a podcast.

   22. DPDwarkesh Patel18:30

       Yeah.

   23. SDSholto Douglas18:30

       So, like, what, what niches are there? What sort of things would you be excited to see-

   24. DPDwarkesh Patel18:34

       Yeah.

   25. SDSholto Douglas18:35

       ... in, like, new blogs podcasts?

   26. DPDwarkesh Patel18:36

       I mean, I wonder if you guys think this too, but I think this, like, Matt Levine of AI, um-

   27. TBTrenton Bricken18:41

       Yeah, absolutely.

   28. DPDwarkesh Patel18:42

       ... is, is like a totally open niche as, as far as I can tell. And I-Apologize to those who are trying to fill it in. (laughs)

   29. SDSholto Douglas18:48

       (laughs)

   30. TBTrenton Bricken18:48

       (laughs)
6. 25:12 – 31:10

   Reading habits

   1. DPDwarkesh Patel25:12

      what's your favorite history book? There's a wall of them behind you.

   2. TBTrenton Bricken25:15

      Mm-hmm.

   3. DPDwarkesh Patel25:17

      Oh, uh, I mean, obviously the, uh, Caro LBJ biographies.

   4. TBTrenton Bricken25:20

      Oh, okay, yeah.

   5. DPDwarkesh Patel25:21

      Yeah.

   6. TBTrenton Bricken25:21

      The, the-

   7. DPDwarkesh Patel25:22

      Sorry, the one, main thing I took away from those books is LBJ had this quote that he would tell his debate sc- he, in his early 20s, he taught debate to these, like, poor Mexican students in, um, in Texas. And he, he used to tell them, "If you do everything, you'll win." I think this is an underrated quote, so that's the main thing I took away. And you see it through his entire career, where there's a reasonable amount of effort, which, you know, goes by, like, 20/80. You do the 20 to get the 80% of the effect, and then if you go beyond that to get, like, "Oh, no, I'm not just gonna do the 20%, I'm gonna just, like, do the whole thing." And there's a level even beyond that, which is like, like this is, like, just, like, an unreasonable use of time. This is hav- gonna have no ultimate impact, and still try doing that. Um, uh, yeah.

   8. TBTrenton Bricken26:05

      You, you've shared on Twitter, uh, using Anki.

   9. DPDwarkesh Patel26:08

      Yeah, yeah.

   10. TBTrenton Bricken26:08

       And even, like, a Claude integration.

   11. DPDwarkesh Patel26:10

       Yeah.

   12. TBTrenton Bricken26:10

       Uh, do you do book clubs? Do you use Goodreads? And what are you reading right now?

   13. DPDwarkesh Patel26:16

       I don't use... I don't have, uh, book clubs. I do, the, um... But the Space Bar edition has just genuinely been a huge, like, uplift in my ability to learn. Mostly because it, it's not even the long-term impact over years, though I think that is part of it and I do regret all the episodes I did without using Space Bar because all the insights have just sort of faded away.

   14. TBTrenton Bricken26:37

       Mm-hmm.

   15. DPDwarkesh Patel26:38

       Um, the main thing is, if you're studying a complicated subject, for, at least for me, it's been super helpful to, um, consolidate. So it's like y- if you don't do it, you feel like a general, where you're like, "I'm gonna wage a campaign against this country," and then you, like, climb one hill, uh, and then the next day you gotta retreat, and then you climb the same hill. Um, there might be a sort of more kosher analogy (laughs) -

   16. TBTrenton Bricken27:02

       (laughs)

   17. DPDwarkesh Patel27:02

       ... uh, sorry. And the other question was what am I reading right now?

   18. TBTrenton Bricken27:05

       Yeah.

   19. DPDwarkesh Patel27:06

       Oh, uh, my friend Alvaro de Menard, author of Fantastic Anachronism. Can I, can I just pull it up, actually? It's right here.

   20. TBTrenton Bricken27:12

       Yeah.

   21. DPDwarkesh Patel27:13

       He... I hope he's, I hope he's okay with me sharing this, but he wrote, um, he made, like, 100 copies of this translation he did of his favorite Greek poet. And they're like... Yeah, uh, Cavafy. Hopefully I didn't mispronounce it.

   22. TBTrenton Bricken27:28

       Huh.

   23. DPDwarkesh Patel27:28

       Um, that one has a good inscription for Grin because that's his coffee, but... (laughs)

   24. TBTrenton Bricken27:32

       (laughs)

   25. DPDwarkesh Patel27:32

       Um, but it's, it's super delightful, and that's what I've been reading recently. Um-

   26. TBTrenton Bricken27:36

       Any insights from it so far?

   27. DPDwarkesh Patel27:38

       Poets will hate this framing. I feel like, uh, poetry is like, it's like TikTok, where... (laughs)

   28. TBTrenton Bricken27:49

       (laughs) Oh, boy.

   29. DPDwarkesh Patel27:49

       Um, where it's like you get this, like, quick vibe of a certain thing, and then you, like, uh, huh, and then like swipe, and then you get the next vibe, swipe. Alvaro (laughs) , I'm sorry.

   30. TBTrenton Bricken28:02

       (laughs)
7. 31:10 – 33:02

   Beard deepdive

   1. DPDwarkesh Patel31:10

   2. TBTrenton Bricken31:10

      How do you groom your beard? It's majestic.

   3. DPDwarkesh Patel31:12

      (laughs) Um, I, I don't know what to say, just (laughs) genetics. (laughs)

   4. SDSholto Douglas31:19

      (laughs)

   5. TBTrenton Bricken31:19

      (laughs)

   6. DPDwarkesh Patel31:19

      I do trim it, but-

   7. TBTrenton Bricken31:21

      No beard oil?

   8. DPDwarkesh Patel31:22

      I, sometimes I do beard oil.

   9. TBTrenton Bricken31:23

      Oh, okay.

   10. DPDwarkesh Patel31:23

       So, yeah.

   11. SDSholto Douglas31:23

       How often?

   12. DPDwarkesh Patel31:24

       Uh, o- once every couple of days.

   13. SDSholto Douglas31:26

       Okay.

   14. TBTrenton Bricken31:26

       That's not sometimes. (laughs)

   15. DPDwarkesh Patel31:28

       (laughs)

   16. SDSholto Douglas31:28

       (laughs)

   17. TBTrenton Bricken31:28

       That's pretty often.

   18. DPDwarkesh Patel31:29

       And, uh, do you have sh- different shampoo for your head and your beard? No.

   19. SDSholto Douglas31:32

       What kind of shampoo do you use?

   20. DPDwarkesh Patel31:33

       (laughs) Anti-dandruff. (laughs)

   21. SDSholto Douglas31:36

       (laughs)

   22. TBTrenton Bricken31:36

       (laughs)

   23. DPDwarkesh Patel31:36

       Do you condition it?

   24. SDSholto Douglas31:38

       Um, yeah.

   25. DPDwarkesh Patel31:40

       How often do you shampoo it?

   26. TBTrenton Bricken31:42

       (laughs)

   27. SDSholto Douglas31:42

       (laughs)

   28. DPDwarkesh Patel31:43

       Who pushed you up to this?

   29. SDSholto Douglas31:44

       We're, we're, we're, we're-

   30. TBTrenton Bricken31:44

       Big shampoo. (laughs)
8. 33:02 – 35:16

   Who is best suited for running an AI lab?

   1. SDSholto Douglas33:02

   2. TBTrenton Bricken33:03

      Which historical figure would be best suited to run a frontier AI lab?

   3. DPDwarkesh Patel33:06

      This is definitely a question for you guys.

   4. SDSholto Douglas33:08

      Oh. No, I mean, I'm curious what your take is first.

   5. DPDwarkesh Patel33:11

      Yeah.

   6. SDSholto Douglas33:11

      You've spoken to more of the heads of AI labs than, than I have.

   7. DPDwarkesh Patel33:14

      I, I was gonna say, LBJ? Sorry, is the question who would be best at running an AI lab or would be best for the world, or?

   8. TBTrenton Bricken33:22

      Let's say-

   9. SDSholto Douglas33:23

      Yeah, what's, what outcome do you want? (laughs)

   10. DPDwarkesh Patel33:24

       (laughs)

   11. TBTrenton Bricken33:24

       Yeah, what outcome do you want?

   12. DPDwarkesh Patel33:26

       'Cause I imagine, it seems like what the best lab, AI lab CEOs succeed at is raising money, building up hype, um, setting a coherent vision. I don't know how much it matters for the CEO themselves to have good research taste or something. But it seems like it, their role is more as a sort of emissary to the rest of the world. And I feel like LBJ would be pretty good at this. Like, just getting the right concessions, making projects move along, um, coordinating among different groups to... Maybe r- oh, Robin Moses.

   13. SDSholto Douglas34:00

       Yeah.

   14. DPDwarkesh Patel34:00

       Again, not necessarily best for the world, but just in terms of, like, making shit happen.

   15. TBTrenton Bricken34:04

       Yeah. I mean, I think best for the world is a pretty important precondition.

   16. DPDwarkesh Patel34:07

       Oh, right.

   17. TBTrenton Bricken34:07

       Um...

   18. DPDwarkesh Patel34:08

       Uh, there's a Lord Ackwood quote of, "Great people are very rarely good people."

   19. TBTrenton Bricken34:13

       Hmm.

   20. DPDwarkesh Patel34:14

       So, it's hard to think of a great person in history who are like, I feel like they'd really move the ball forward and also I trust their moral judgment.

   21. TBTrenton Bricken34:20

       We're lucky in many senses with, like, the set today, right?

   22. DPDwarkesh Patel34:23

       That's right.

   23. TBTrenton Bricken34:23

       Like, the set of people today are both, like, they try and care a lot about the moral side as well as, uh-

   24. DPDwarkesh Patel34:29

       Yeah.

   25. TBTrenton Bricken34:30

       ... just sort of drive the labs forward.

   26. DPDwarkesh Patel34:31

       This is also why-

   27. TBTrenton Bricken34:32

       Uh-

   28. DPDwarkesh Patel34:32

       ... I'm skeptical of big, grand schemes, like nationalization or some public-private partnership, or just generally shaking up the landscape too much, because I do think we've, we're in, like, one of the better... I mean, the sort of, like, the difficulty of whether it's alignment or whether it's some kind of deployment, um, safety risks, that is just, like, the nature of the universe is gonna make that some level of difficulty. But the human factors, in a lot of the counterfactual universes, I feel like we don't end up with people... Like, we could even be in a universe where they don't even pay lip service.

   29. TBTrenton Bricken35:05

       Yeah.

   30. DPDwarkesh Patel35:05

       This is like, not an idea that anybody had.
9. 35:16 – 40:50

   Preparing for fast AGI timelines

   1. DPDwarkesh Patel35:16

   2. TBTrenton Bricken35:16

      How are you preparing for fast timelines?

   3. DPDwarkesh Patel35:19

      If there's fast timelines, then there will be this six-month period in which the most important decisions in human history are being made. And I feel like having an AI podcast during that time might be useful. Um, that's basically the plan.

   4. SDSholto Douglas35:35

      Have you made any shorter term decisions, um, with regards to, like, spending or health or anything else?

   5. DPDwarkesh Patel35:43

      After I interviewed Zuckerberg, my business bank balance was negative 23 cents. Uh, when the ad hi- money hit, I immediately reinvested it into NVIDIA. So, that is the... (laughs)

   6. SDSholto Douglas35:56

      (laughs)

   7. DPDwarkesh Patel35:56

      Um, sorry, but you were asking from a sort of altruistic perspective?

   8. SDSholto Douglas36:00

      No, no, just in general, like, have you, have you changed the way you live at all because of your AGI timelines?

   9. DPDwarkesh Patel36:06

      I never looked into getting a Roth IRA. (laughs)

   10. TBTrenton Bricken36:10

       He brought us Fiji water before. (laughs)

   11. SDSholto Douglas36:13

       (laughs)

   12. DPDwarkesh Patel36:13

       (laughs)

   13. SDSholto Douglas36:13

       Which was in plastic bottles, so...

   14. TBTrenton Bricken36:14

       Dwarkesh has changed. (laughs)

   15. SDSholto Douglas36:16

       (laughs)

   16. DPDwarkesh Patel36:16

       (laughs) Well, have you guys changed your lifestyle as a result?

   17. TBTrenton Bricken36:21

       Not really, no. I, I just, like, work all the time.

   18. DPDwarkesh Patel36:24

       (laughs)

   19. SDSholto Douglas36:25

       (laughs)

   20. DPDwarkesh Patel36:25

       But you were doing that anyways, or would you not?

   21. TBTrenton Bricken36:26

       Yeah. (laughs)

   22. DPDwarkesh Patel36:26

       Ah...

   23. TBTrenton Bricken36:30

       I would probably be going very intensely at whatever, like, thing I'd picked to devote myself to.

   24. DPDwarkesh Patel36:34

       Yeah.

   25. TBTrenton Bricken36:34

       Yeah.

   26. DPDwarkesh Patel36:35

       How about you?

   27. SDSholto Douglas36:35

       I canceled my 401K contributions earlier.

   28. DPDwarkesh Patel36:37

       (laughs) Oh, really?

   29. SDSholto Douglas36:38

       Yeah. Yeah, that, that felt like a, a more serious one. Um, it's, it's hard for me to imagine a world in which I'm, like, s- have all this money that's just sitting in this account and waiting till I'm 60 and things look so different then.

   30. DPDwarkesh Patel36:52

       But that, you could be like a trillionaire with your, uh, marginal 401K contribution.
10. 40:50 – 49:33

    Growing the podcast

    1. TBTrenton Bricken40:50

    2. DPDwarkesh Patel40:50

       The, um, the thing I have takes on, which I rarely get asked about, is distribution.

    3. SDSholto Douglas40:54

       Distribution of AI?

    4. DPDwarkesh Patel40:55

       No, no, sorry. (laughs) Like MrBeast-style d- distribution.

    5. SDSholto Douglas40:58

       Oh, yeah.

    6. DPDwarkesh Patel41:00

       Um.

    7. SDSholto Douglas41:00

       Yeah, yeah, yeah.

    8. DPDwarkesh Patel41:00

       Um, where people, I think, rightly focus on the content. Um, and if that's not up to snuff, I think you ju- you won't succeed. But to the extent that somebody is trying to do similar things, the thing they consistently underrate is putting the ... yeah, the, putting the time into getting distribution right. Um, I just take, like, random takes about, for example, the most successful thing for my podcast in terms of growth has been YouTube Shorts.

    9. TBTrenton Bricken41:30

       Mm-hmm.

    10. DPDwarkesh Patel41:30

        It's a thing you would never have predicted beforehand. And, you know, they're, like, responsible for, like, basically m- at least half the growth of the podcast or something. Uh-

    11. TBTrenton Bricken41:41

        I mean, I'd buy that.

    12. DPDwarkesh Patel41:42

        Yeah.

    13. TBTrenton Bricken41:42

        Why wouldn't you have predicted it? Like, I mean, like, I mean, I guess there's the contrast of, like, the long form deep content-

    14. DPDwarkesh Patel41:48

        Mm-hmm.

    15. TBTrenton Bricken41:48

        ... and, like, YouTube Shorts and stuff. But I do definitely think they're, they're good hooks.

    16. DPDwarkesh Patel41:52

        That's right.

    17. TBTrenton Bricken41:52

        Good content.

    18. DPDwarkesh Patel41:52

        Yeah, yeah.

    19. TBTrenton Bricken41:53

        Yeah.

    20. DPDwarkesh Patel41:53

        And I have, like, takes on how to write tweets and stuff. The main intuition being-

    21. TBTrenton Bricken41:56

        Yeah.

    22. DPDwarkesh Patel41:56

        ... like, write like you're writing to a group chat.

    23. TBTrenton Bricken41:58

        Yeah.

    24. DPDwarkesh Patel41:58

        Um-

    25. TBTrenton Bricken41:59

        Yeah.

    26. DPDwarkesh Patel41:59

        ... to a group chat of your friends rather than this, like, formal whatever. I don't know, I, I, I could ... but just, like, these sort of, like ...

    27. SDSholto Douglas42:06

        Yeah, no, I mean, what el- what else comes to mind here?

    28. TBTrenton Bricken42:08

        Well, maybe it's interesting, the difference between, like, TikTok and YouTube Shorts.

    29. DPDwarkesh Patel42:11

        Oh yeah, we've never cracked TikTok.

    30. TBTrenton Bricken42:12

        Yeah.

Episode duration: 49:33