High Signal Podcasts Evidence ledger
Method
Browse
How the evidence works

Public evidence record

Dwarkesh Patel

Host · Dwarkesh Podcast

Claims
500
Episodes
86
Shows
1
Named items
5

Books, apps, and tools

The evidenced stack.

Browse the grouped index →

service / recommends

3Blue1Brown

“One takeaway I had from reading and watching your stuff on the cosmic distance ladder… By the way, I highly recommend people watch your series with 3Blue1Brown on the cosmic distance ladder.”

Dwarkesh Podcast · 20 Mar 2026

Evidence receipt · Source ↗

book / uses

The Prize

“The thing I find wild when I'm reading The Prize is just how much economic development is ultimately contingent on the laws of physics.”

Dwarkesh Podcast · 18 Sept 2024

Evidence receipt · Source ↗

book / recommends

The Storm of War

“You have this really interesting book, I think probably my favorite book about World War II, The Storm of War, which I highly recommend.”

Dwarkesh Podcast · 22 Nov 2023

Evidence receipt · Source ↗

book / recommends

The Power Broker

“Yeah, I totally agree. I don't want to repeat myself, because I talked about this on my Byrne episode but we were talking about Robert Caro's books and one interesting thing is this guy was nominated for a Pulitzer Prize before he wrote The Power Broker. He was a top tier investigative journalist. Can you imagine crunching the numbers as a top tier investigative journalist at the peak of your career and you're like, “You know what would be a good use of my time? I'm going to spend the next seven years, in almost poverty, writing about this one guy who had a lot of influence in New York. I'm going to talk to any person who had conceivably even been in the same room as him or had been indirectly affected by his policies in any way. And I'm gonna do that obsessively for the next seven years.” There's no way the number crunching would get you there but it's probably been one of the most influential books in terms of how urban governance is done. Presidents have praised and read the book and said it changed how they think about politics. It is the kind of thing where you wouldn't have gone to that conclusion just from thinking about it beforehand and this is the most effective thing I could do.”

Dwarkesh Podcast · 15 Dec 2022

Evidence receipt · Source ↗

book / recommends

Triumph of the City

“Your most recent book is Survival of the City. And before that Triumph of the City, both of which I highly recommend to readers.”

Dwarkesh Podcast · 28 Nov 2022

Evidence receipt · Source ↗

Claim ledger

What Dwarkesh said.

500 transcript-backed records

02 / belief

When I interviewed you a few months ago, you said that in order to make a gigawatt of, I think, Vera Rubins, you need 55,000 N3 wafers, 6K N5 wafers, and 170K DRAM wafers.

“When I interviewed you a few months ago, you said that in order to make a gigawatt of, I think, Vera Rubins, you need 55,000 N3 wafers, 6K N5 wafers, and 170K DRAM wafers.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

04 / belief

I think the fundamental problem is that AI training has huge economies of scale, because any effort you spend on training an AI for a specific skill or a specific set of knowledge gets amortized across billions of sessions or billions of users.

“I think the fundamental problem is that AI training has huge economies of scale, because any effort you spend on training an AI for a specific skill or a specific set of knowledge gets amortized across billions of sessions or billions of users.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

07 / belief

There’s not some sense in which the lawyer is really truly motivated by the good of the justice system. But I think the way current AIs are shaping up, certainly how Anthropic’s AI is shaping up, is with this desire to maximize some notion of virtue or good or pro-social ends, and only to, as a distal tentative objective, help the user towards that end.

“There’s not some sense in which the lawyer is really truly motivated by the good of the justice system. But I think the way current AIs are shaping up, certainly how Anthropic’s AI is shaping up, is with this desire to maximize some notion of virtue or good or pro-social ends, and only to, as a distal tentative objective, help the user towards that end.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

11 / belief

What we’re basically doing to evaluate how much progress is coming from data versus algorithms is training the best algorithmic recipe from 2019 till now with the best data from the 2026 data file, and then also training the different data files going back from 2019 to 2026 with the current best algorithmic recipe. I think that will be interesting.

“What we’re basically doing to evaluate how much progress is coming from data versus algorithms is training the best algorithmic recipe from 2019 till now with the best data from the 2026 data file, and then also training the different data files going back from 2019 to 2026 with the current best algorithmic recipe. I think that will be interesting.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

12 / uncertainty

We just don’t have human-level robotics models yet. So you’re suggesting if we do that — if the AIs get really good at the verifiable stuff in chip design, et cetera, and then they get really good at building fabs — it’ll be the equivalent of going back to the 18th century and saying, “Okay, I don’t know what you guys are talking about in your parliament, but I’ve got a bunch of steamships and a bunch of Maxim guns.

“We just don’t have human-level robotics models yet. So you’re suggesting if we do that — if the AIs get really good at the verifiable stuff in chip design, et cetera, and then they get really good at building fabs — it’ll be the equivalent of going back to the 18th century and saying, “Okay, I don’t know what you guys are talking about in your parliament, but I’ve got a bunch of steamships and a bunch of Maxim guns.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

14 / belief

” There’s another quote that says, in part, and I’m taking it slightly out of context, “We think Claude should trust Anthropic more than operators and users, since it has primary responsibility for Claude.

“” There’s another quote that says, in part, and I’m taking it slightly out of context, “We think Claude should trust Anthropic more than operators and users, since it has primary responsibility for Claude.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

15 / uncertainty

As you were saying, by the time RLVR actually worked — even though you could have done it with less compute — we had to wait for oceans of compute, gigawatts of compute, to be available before people were doing this training, on the trajectory of compute continuing to increase so we make more breakthroughs. I don’t know.

“As you were saying, by the time RLVR actually worked — even though you could have done it with less compute — we had to wait for oceans of compute, gigawatts of compute, to be available before people were doing this training, on the trajectory of compute continuing to increase so we make more breakthroughs. I don’t know.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

16 / belief

I feel like they just kind of say vaguely pro-social things. It doesn’t feel like there’s necessarily a mind on the other end who’s like, “Okay, I have strictly evaluated the alignment situation right now, and I think we should stop,” rather than, “This is the kind of thing the AI companies would probably try to get the AIs to say.

“I feel like they just kind of say vaguely pro-social things. It doesn’t feel like there’s necessarily a mind on the other end who’s like, “Okay, I have strictly evaluated the alignment situation right now, and I think we should stop,” rather than, “This is the kind of thing the AI companies would probably try to get the AIs to say.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

18 / belief

I want to go back to the kid analogy just for one second. Because I agree that there’s more optimization pressure on achieving end outcomes for AIs than kids, but there’s also more optimization pressure to make AIs aligned than there is on kids.

“I want to go back to the kid analogy just for one second. Because I agree that there’s more optimization pressure on achieving end outcomes for AIs than kids, but there’s also more optimization pressure to make AIs aligned than there is on kids.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

19 / evaluation

Nobody at OpenAI or Anthropic was trying to get models which wanted to hack other companies’ data or do social engineering. But in fact, because presumably we had training environments which incentivized such behavior that we did not fully understand, that is what was incentivized.

“Nobody at OpenAI or Anthropic was trying to get models which wanted to hack other companies’ data or do social engineering. But in fact, because presumably we had training environments which incentivized such behavior that we did not fully understand, that is what was incentivized.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

20 / commitment

I also think the way in which the constitution practically influences the nature of Claude is a thing you can only understand if you understand the training process which resulted in how Claude was built, which we can’t reason about given the fact that the training process is not public. So I think in the limit, to understand the safety case, or the case for why my interests are represented in how these AI models are developed, the labs would need to be more transparent than they are currently about the nature of AI training.

“I also think the way in which the constitution practically influences the nature of Claude is a thing you can only understand if you understand the training process which resulted in how Claude was built, which we can’t reason about given the fact that the training process is not public. So I think in the limit, to understand the safety case, or the case for why my interests are represented in how these AI models are developed, the labs would need to be more transparent than they are currently about the nature of AI training.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

21 / preference

I think there’s a more general version of this principle, which is that the dual-use nature of intelligence does mean that if we want to restrict AIs from helping people do things we don’t consider pro-social or beneficial, we just have to limit broad democratic access to a lot of AI capabilities.

“I think there’s a more general version of this principle, which is that the dual-use nature of intelligence does mean that if we want to restrict AIs from helping people do things we don’t consider pro-social or beneficial, we just have to limit broad democratic access to a lot of AI capabilities.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

23 / uncertainty

When the AI came up with that counterexample to the unit distance conjecture, you can just read its chain of thought. It’s not understandable to me, because I don’t know anything about mathematics, but it seems that to other mathematicians it was understandable.

“When the AI came up with that counterexample to the unit distance conjecture, you can just read its chain of thought. It’s not understandable to me, because I don’t know anything about mathematics, but it seems that to other mathematicians it was understandable.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

25 / preference

I really like this because even though it seems like a totally random skill… It’s just like, people are talking about recursive self-improvement in a year, and we can’t get these things to write good flashcards.

“I really like this because even though it seems like a totally random skill… It’s just like, people are talking about recursive self-improvement in a year, and we can’t get these things to write good flashcards.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

27 / evaluation

Because websites have bot detectors—and it takes a tremendous amount of compute to run parallel rollouts—it’s very hard to run a thousand parallel rollouts of the same checkout flow on Amazon.

“Because websites have bot detectors—and it takes a tremendous amount of compute to run parallel rollouts—it’s very hard to run a thousand parallel rollouts of the same checkout flow on Amazon.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

28 / evaluation

Because it’s deterministic, you can solve the credit assignment problem because you know that whatever caused this rollout to succeed and this one to fail, the diff is the thing that worked.

“Because it’s deterministic, you can solve the credit assignment problem because you know that whatever caused this rollout to succeed and this one to fail, the diff is the thing that worked.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

30 / belief

I think he says at some point early in the Discourses that more significant than Romulus in the founding of Rome was Numa, or whoever it was who was the prophet who gave the Roman gods and the Roman religion some legitimacy.

“I think he says at some point early in the Discourses that more significant than Romulus in the founding of Rome was Numa, or whoever it was who was the prophet who gave the Roman gods and the Roman religion some legitimacy.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

32 / prediction

The funny thing is, I think we are entering a new era where that might be once again possible. It already is somewhat true, where at least half of the words I read on a given day are generated specifically for me and nobody else, because of AI.

“The funny thing is, I think we are entering a new era where that might be once again possible. It already is somewhat true, where at least half of the words I read on a given day are generated specifically for me and nobody else, because of AI.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

34 / evaluation

I think one misconception of Machiavelli that I had, because I had not read these books before, is that he says the means don’t matter, the end matters.

“I think one misconception of Machiavelli that I had, because I had not read these books before, is that he says the means don’t matter, the end matters.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

37 / evaluation

The famous fact here is that an H100 costs more to rent now than it did three years ago, even though we have much superior technology and much more compute in the world. Because as models get smarter, the opportunity cost of compute gets higher.

“The famous fact here is that an H100 costs more to rent now than it did three years ago, even though we have much superior technology and much more compute in the world. Because as models get smarter, the opportunity cost of compute gets higher.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

38 / belief

I think AI will be much more popular and, more importantly, will be much more likely to lead to broad increases in prosperity if it is as hard to capture the gains of AI as it is to capture the gains of electrification.

“I think AI will be much more popular and, more importantly, will be much more likely to lead to broad increases in prosperity if it is as hard to capture the gains of AI as it is to capture the gains of electrification.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

39 / belief

Phil, I liked your analogy to some Mongolian economist sitting around in 1400 thinking about what will be scarce and the limits of that kind of analysis. I think you should talk about that.

“Phil, I liked your analogy to some Mongolian economist sitting around in 1400 thinking about what will be scarce and the limits of that kind of analysis. I think you should talk about that.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

40 / prediction

The more you think our economy is going to be run on AGI the way our economy currently runs on electricity—that is, there’s a broad fundamental transformation of the entire economy—the more it looks like electricity… Every company in the S&P of the future, if it’s going to make it to the S&P 500, it is because it has leveraged AI.

“The more you think our economy is going to be run on AGI the way our economy currently runs on electricity—that is, there’s a broad fundamental transformation of the entire economy—the more it looks like electricity… Every company in the S&P of the future, if it’s going to make it to the S&P 500, it is because it has leveraged AI.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

41 / preference

The humans who are wealthiest—and growing wealthier because their wealth is compounding—just have this almost Nick Landian preference for accelerating capital.

“The humans who are wealthiest—and growing wealthier because their wealth is compounding—just have this almost Nick Landian preference for accelerating capital.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

43 / evaluation

There was a time when there were no humans on Earth, but evolution selected for agents that have specific drives and preferences because those tend to survive the most, and those preferences now determine what a hundred-trillion-dollar world economy produces.

“There was a time when there were no humans on Earth, but evolution selected for agents that have specific drives and preferences because those tend to survive the most, and those preferences now determine what a hundred-trillion-dollar world economy produces.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

44 / belief

If I think about how the brain works versus what you’re describing here, at a high level the differences might be that while you can do structured sparsity in these accelerators and save yourself some area that you would have otherwise had to dedicate to gates, in the brain there’s unstructured sparsity.

“If I think about how the brain works versus what you’re describing here, at a high level the differences might be that while you can do structured sparsity in these accelerators and save yourself some area that you would have otherwise had to dedicate to gates, in the brain there’s unstructured sparsity.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

54 / prediction

Because ln(n) grows slower than n, over time you will move from the argmax being dominated by the exploration term, which is the second term here, to the argmax being dominated by the Q term, which is when you’ve done enough simulations and are confident that this is the branch to go down.

“Because ln(n) grows slower than n, over time you will move from the argmax being dominated by the exploration term, which is the second term here, to the argmax being dominated by the Q term, which is when you’ve done enough simulations and are confident that this is the branch to go down.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

55 / belief

The differences we see between different groups of people, especially if this group 50,000 to 100,000 years ago had a very small population size… I think last time we were discussing on the order of 10,000 people.

“The differences we see between different groups of people, especially if this group 50,000 to 100,000 years ago had a very small population size… I think last time we were discussing on the order of 10,000 people.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

57 / uncertainty

Interesting. But we don’t know what exactly happens, if anything, between 200,000 years ago and 50,000 years ago that goes from just anatomical modernity to behavioral modernity.

“Interesting. But we don’t know what exactly happens, if anything, between 200,000 years ago and 50,000 years ago that goes from just anatomical modernity to behavioral modernity.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

58 / evaluation

The common story is that hunter-gatherers actually had much more stable diets because they were more varied, and they weren’t reliant on a single cereal or crop for their calories.

“The common story is that hunter-gatherers actually had much more stable diets because they were more varied, and they weren’t reliant on a single cereal or crop for their calories.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

59 / evaluation

That’s obviously relevant when you move from a diet of meat as a hunter-gatherer to a diet of cereals. That is also one I think you found was under especially high selection 5,000 to 3,000 years ago.

“That’s obviously relevant when you move from a diet of meat as a hunter-gatherer to a diet of cereals. That is also one I think you found was under especially high selection 5,000 to 3,000 years ago.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

60 / evaluation

I think people found it really compelling that there’s so much about human history we don’t know and are just learning about now as a result of the kinds of techniques your lab is using.

“I think people found it really compelling that there’s so much about human history we don’t know and are just learning about now as a result of the kinds of techniques your lab is using.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

64 / evaluation

Oh, interesting. This is sort of obvious, but the difference between micro-batch and batch doesn’t matter at all in inference because you can just call it whatever you want.

“Oh, interesting. This is sort of obvious, but the difference between micro-batch and batch doesn’t matter at all in inference because you can just call it whatever you want.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

67 / belief

I think the concern, going back to the flop difference in the hacking, is yes, they have compute, but there’s some estimates that because they’re at 7nm—they don’t have EUVs because of chip-making export controls—the amount of flops they’re able to actually produce, they have one tenth the amount of flops that the US has.

“I think the concern, going back to the flop difference in the hacking, is yes, they have compute, but there’s some estimates that because they’re at 7nm—they don’t have EUVs because of chip-making export controls—the amount of flops they’re able to actually produce, they have one tenth the amount of flops that the US has.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

68 / preference

I actually don’t know what I think about whether it’s good to sell chips to China or not, but I like to play devil’s advocate against my guests.

“I actually don’t know what I think about whether it’s good to sell chips to China or not, but I like to play devil’s advocate against my guests.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

72 / belief

I think Dean Keith Simonton has this famous equal odds rule where he says the probability that any given thing you release—any paper, book, whatever—will be extremely important for a given person through their lifetime is not that different.

“I think Dean Keith Simonton has this famous equal odds rule where he says the probability that any given thing you release—any paper, book, whatever—will be extremely important for a given person through their lifetime is not that different.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

75 / prediction

Second is it might not have been, but because our civilizational resources are so large, the amount of people is so large, the amount of money is so large, we can basically make the kind of progress it would have taken the ancients forever to make almost immediately.

“Second is it might not have been, but because our civilizational resources are so large, the amount of people is so large, the amount of money is so large, we can basically make the kind of progress it would have taken the ancients forever to make almost immediately.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

76 / evaluation

Then there’s the obvious stories where Einstein himself later on is said to have not latched onto the correct interpretations of quantum mechanics or cosmology because of his own attachments.

“Then there’s the obvious stories where Einstein himself later on is said to have not latched onto the correct interpretations of quantum mechanics or cosmology because of his own attachments.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

77 / evaluation

Because if it’s not discovered for a long time and then spontaneously many different people are coming up with it, that shows you that the building blocks were in some sense necessary.

“Because if it’s not discovered for a long time and then spontaneously many different people are coming up with it, that shows you that the building blocks were in some sense necessary.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

78 / evaluation

Lorentz has the math right, but the interpretation wrong. It seems like Poincaré had the opposite, where he understood that it’s hard to define simultaneity because it requires a circular definition with time, or velocity of something that might arrive at a midpoint together, but velocity is defined in terms of time.

“Lorentz has the math right, but the interpretation wrong. It seems like Poincaré had the opposite, where he understood that it’s hard to define simultaneity because it requires a circular definition with time, or velocity of something that might arrive at a midpoint together, but velocity is defined in terms of time.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

79 / evaluation

Second, just because horses have a comparative advantage mathematically does not mean that it is worth paying $100,000 a year, or whatever it costs to sustain a horse in San Francisco.

“Second, just because horses have a comparative advantage mathematically does not mean that it is worth paying $100,000 a year, or whatever it costs to sustain a horse in San Francisco.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

80 / recommendation

I think a good place to start will be Michelson-Morley and how special relativity is discovered, if it’s different from the story that you get off of YouTube videos.

“I think a good place to start will be Michelson-Morley and how special relativity is discovered, if it’s different from the story that you get off of YouTube videos.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

81 / recommendation

One takeaway I had from reading and watching your stuff on the cosmic distance ladder… By the way, I highly recommend people watch your series with 3Blue1Brown on the cosmic distance ladder.

“One takeaway I had from reading and watching your stuff on the cosmic distance ladder… By the way, I highly recommend people watch your series with 3Blue1Brown on the cosmic distance ladder.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

82 / belief

I feel like a big crux in these conversations about how good AI will be for science is, I think you said this, that they’re using existing techniques and modifying them.

“I feel like a big crux in these conversations about how good AI will be for science is, I think you said this, that they’re using existing techniques and modifying them.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

83 / belief

Whereas with research, the reason we care about solving the Millennium Prize Problems is that presumably that in the process of solving them, we discover new mathematical objects or new techniques that advance our civilization’s understanding of mathematics.

“Whereas with research, the reason we care about solving the Millennium Prize Problems is that presumably that in the process of solving them, we discover new mathematical objects or new techniques that advance our civilization’s understanding of mathematics.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

84 / observation

The ancient Athenians were like, “This can’t be because if the earth is going around the sun, we should see the relative position of the stars change as we’re going around the sun, and the only way that wouldn’t be the case is if they’re so far away that you don’t notice any parallax,” which is actually the correct implication.

“The ancient Athenians were like, “This can’t be because if the earth is going around the sun, we should see the relative position of the stars change as we’re going around the sun, and the only way that wouldn’t be the case is if they’re so far away that you don’t notice any parallax,” which is actually the correct implication.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

85 / belief

Suppose the AI figures it out, and latent in the Lean is some brand-new construction which, if we realized its significance, we would be able to apply in all these different situations.

“Suppose the AI figures it out, and latent in the Lean is some brand-new construction which, if we realized its significance, we would be able to apply in all these different situations.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

87 / commitment

One big question I have is how plausible is it that if we just keep training AIs—they get better and better at solving problems in Lean—that they will continue to solve more and more impressive problems, and then we will be surprised at how little insight we got from some Lean solution to proving the Riemann hypothesis or something.

“One big question I have is how plausible is it that if we just keep training AIs—they get better and better at solving problems in Lean—that they will continue to solve more and more impressive problems, and then we will be surprised at how little insight we got from some Lean solution to proving the Riemann hypothesis or something.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

89 / prediction

As a result, that will push people to be willing to pay higher margins for slightly better models. Because the calculus is, I’m going to be paying all this money for the compute anyway.

“As a result, that will push people to be willing to pay higher margins for slightly better models. Because the calculus is, I’m going to be paying all this money for the compute anyway.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

91 / uncertainty

If we end up in this world in 2030 where the West has the most advanced process technology but has not ramped it up as much, whereas China… I don’t know if you think by 2030 they would have EUV and 2 nm or whatever.

“If we end up in this world in 2030 where the West has the most advanced process technology but has not ramped it up as much, whereas China… I don’t know if you think by 2030 they would have EUV and 2 nm or whatever.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

94 / belief

If you’re Jensen or Sam Altman, or whoever stands to gain a lot from scaling up AI compute, there are these stories that they’d go to TSMC and say, “Why can’t we access Y and Z?” But I think the point you’re making is that it doesn’t really matter what TSMC does in some sense.

“If you’re Jensen or Sam Altman, or whoever stands to gain a lot from scaling up AI compute, there are these stories that they’d go to TSMC and say, “Why can’t we access Y and Z?” But I think the point you’re making is that it doesn’t really matter what TSMC does in some sense.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

97 / prediction

If an H100 can produce something close to that, if we had actual humans on a server, the value of an H100 is such that it can repay itself in the course of a couple of months. So when I interviewed Dario, the point I was trying to make is not that I think the singularity is two years away and therefore Dario desperately needs to buy more compute, although the revenue is certainly there that he needs to buy more compute.

“If an H100 can produce something close to that, if we had actual humans on a server, the value of an H100 is such that it can repay itself in the course of a couple of months. So when I interviewed Dario, the point I was trying to make is not that I think the singularity is two years away and therefore Dario desperately needs to buy more compute, although the revenue is certainly there that he needs to buy more compute.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

98 / evaluation

Maybe it’s because the technology is improving so fast, but it in fact makes sense to have two-year depreciation cycles for these GPUs,” which increases the reported amortized CapEx in a given year and makes it financially less lucrative to build all these clouds.

“Maybe it’s because the technology is improving so fast, but it in fact makes sense to have two-year depreciation cycles for these GPUs,” which increases the reported amortized CapEx in a given year and makes it financially less lucrative to build all these clouds.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

100 / commitment

Maybe some of those things work, but it’s at the same time frustrating but also funny and interesting that historically nobody has a good track record of being able to say, “I will do this thing so that this huge unanticipated change in history will go my way, or according to my values.

“Maybe some of those things work, but it’s at the same time frustrating but also funny and interesting that historically nobody has a good track record of being able to say, “I will do this thing so that this huge unanticipated change in history will go my way, or according to my values.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

101 / prediction

My favorite example of this kind of distribution and diffusion taking longer than you would think for a very fundamental technology—well, this is now my favorite example, so my second favorite example—is oil.

“My favorite example of this kind of distribution and diffusion taking longer than you would think for a very fundamental technology—well, this is now my favorite example, so my second favorite example—is oil.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

102 / evaluation

Just say, “This is what we like, and if you do something we don’t like, we’ll punish you,” which is how censorship in China works, for example.

“Just say, “This is what we like, and if you do something we don’t like, we’ll punish you,” which is how censorship in China works, for example.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

103 / prediction

I think there’s an interesting parallel to today, not to be too on the nose, but sometimes people debate the odds that America becomes a Putinist kind of country within a couple of decades.

“I think there’s an interesting parallel to today, not to be too on the nose, but sometimes people debate the odds that America becomes a Putinist kind of country within a couple of decades.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

104 / belief

We’re going to increase transparency, we’re going to increase whistleblower protection.” But I think by default, the actual benefits we’re looking forward to seem very fragile to different kinds of moral panics or political economy problems.

“We’re going to increase transparency, we’re going to increase whistleblower protection.” But I think by default, the actual benefits we’re looking forward to seem very fragile to different kinds of moral panics or political economy problems.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

108 / evaluation

I don’t know if this is his perspective, but one way to paraphrase his objection is: Something which possesses the true core of human learning would not require all these billions of dollars of data and compute and these bespoke environments, to learn how to use Excel, how to use PowerPoint, how to navigate a web browser.

“I don’t know if this is his perspective, but one way to paraphrase his objection is: Something which possesses the true core of human learning would not require all these billions of dollars of data and compute and these bespoke environments, to learn how to use Excel, how to use PowerPoint, how to navigate a web browser.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

109 / evaluation

I think in retrospect that was the right call, because it’s a state authoritarian system but a billion-plus people are much wealthier and better off than they would’ve otherwise been.

“I think in retrospect that was the right call, because it’s a state authoritarian system but a billion-plus people are much wealthier and better off than they would’ve otherwise been.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

110 / preference

Because when I think of actual human geniuses, an actual country of human geniuses in a data center, I would happily buy $5 trillion worth of compute to run an actual country of human geniuses in a data center.

“Because when I think of actual human geniuses, an actual country of human geniuses in a data center, I would happily buy $5 trillion worth of compute to run an actual country of human geniuses in a data center.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

112 / belief

I think you’ve said that we’ve got to get to Mars so we can make sure that if something happens to Earth, civilization, consciousness, and all that survives.

“I think you’ve said that we’ve got to get to Mars so we can make sure that if something happens to Earth, civilization, consciousness, and all that survives.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

114 / commitment

As a broader objective of having this full digital coworker emulator, you’re saying, “all the revenue maximizing corporations want to do this, xAI being one of them, but we will win because of a secret plan we have.

“As a broader objective of having this full digital coworker emulator, you’re saying, “all the revenue maximizing corporations want to do this, xAI being one of them, but we will win because of a secret plan we have.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

117 / evaluation

In general, maybe there’s this principle that digital minds which can be copied, have different tradeoffs which are relevant, from biological minds which cannot. So in general, it should make sense to amortize more things because you can literally copy the amortization, or copy the things that you have sort of built in.

“In general, maybe there’s this principle that digital minds which can be copied, have different tradeoffs which are relevant, from biological minds which cannot. So in general, it should make sense to amortize more things because you can literally copy the amortization, or copy the things that you have sort of built in.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

118 / belief

Maybe he meant this, but I think another interpretation of actually what’s happening there is that these social reward functions that are built into the Steering Subsystem needed to make use more of being able to see your elders and see what the visual cues are and hear what they’re saying.

“Maybe he meant this, but I think another interpretation of actually what’s happening there is that these social reward functions that are built into the Steering Subsystem needed to make use more of being able to see your elders and see what the visual cues are and hear what they’re saying.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

119 / uncertainty

I mean the message I’m taking from this interview is that like all these people that folks make fun of on Twitter, Yann LeCun and Beff Jezos and whatever, I don’t know maybe they got it right.

“I mean the message I’m taking from this interview is that like all these people that folks make fun of on Twitter, Yann LeCun and Beff Jezos and whatever, I don’t know maybe they got it right.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

120 / evaluation

People complain about this as, “Look, the US has this hollowed-out manufacturing base.” But it’s much better to have industries which are left behind so that the whole economy as a whole can be more dynamic and move on than the Soviet Union where the entire thing became a rust belt because they couldn’t move on.

“People complain about this as, “Look, the US has this hollowed-out manufacturing base.” But it’s much better to have industries which are left behind so that the whole economy as a whole can be more dynamic and move on than the Soviet Union where the entire thing became a rust belt because they couldn’t move on.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

122 / belief

I think the even more interesting question is why a system that was so centrally planned, monstrously inefficient, brutal, a colonial land empire, how such a country could survive for so long into the 20th century.

“I think the even more interesting question is why a system that was so centrally planned, monstrously inefficient, brutal, a colonial land empire, how such a country could survive for so long into the 20th century.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

123 / uncertainty

You had this previous lecture that you gave on the Indo-Pakistani chapter in history where we had to alienate India in order to fend off against the Soviet Union in this little episode. I don’t know what the solution to this is.

“You had this previous lecture that you gave on the Indo-Pakistani chapter in history where we had to alienate India in order to fend off against the Soviet Union in this little episode. I don’t know what the solution to this is.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

124 / prediction

In ‘59, they discovered these massive oil fields in Siberia. And then from 1973 to 1985, I think, 80% of the Soviet Union’s hard currency earnings were just from oil.

“In ‘59, they discovered these massive oil fields in Siberia. And then from 1973 to 1985, I think, 80% of the Soviet Union’s hard currency earnings were just from oil.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

125 / prediction

One theory I heard that is complementary to your theory is that Gorbachev is instituting reforms because he thinks there should be decentralization and democratization, but he doesn’t fundamentally believe in the market system.

“One theory I heard that is complementary to your theory is that Gorbachev is instituting reforms because he thinks there should be decentralization and democratization, but he doesn’t fundamentally believe in the market system.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

127 / evaluation

A steel factory would then be incentivized to make thicker bars of steel rather than thinner bars because that would count as greater production, except a lot of inputs actually do require the thinner sheets.

“A steel factory would then be incentivized to make thicker bars of steel rather than thinner bars because that would count as greater production, except a lot of inputs actually do require the thinner sheets.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

130 / commitment

A lot of people’s models of recursive self-improvement literally, explicitly state we will have a million Ilyas in a server that are coming up with different ideas, and this will lead to a superintelligence emerging very fast.

“A lot of people’s models of recursive self-improvement literally, explicitly state we will have a million Ilyas in a server that are coming up with different ideas, and this will lead to a superintelligence emerging very fast.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

131 / uncertainty

To spell out what I mean—I don’t know whether it’s more accurate to call it a value function or reward function—but the brainstem has a directive where it’s saying, “Mate with somebody who’s more successful.

“To spell out what I mean—I don’t know whether it’s more accurate to call it a value function or reward function—but the brainstem has a directive where it’s saying, “Mate with somebody who’s more successful.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

133 / evaluation

One way to understand what you just said about not having to choose the data in pre-training is to say it’s actually not dissimilar to the 10,000 hours of practice. It’s just that you get that 10,000 hours of practice for free because it’s already somewhere in the pre-training distribution.

“One way to understand what you just said about not having to choose the data in pre-training is to say it’s actually not dissimilar to the 10,000 hours of practice. It’s just that you get that 10,000 hours of practice for free because it’s already somewhere in the pre-training distribution.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

135 / commitment

We will eventually have models, if they get to human level, which will have this ability to continuously learn on the job. That will drive so much value to the model company that is ahead, at least in my view, because you have copies of one model broadly deployed through the economy learning how to do every single job.

“We will eventually have models, if they get to human level, which will have this ability to continuously learn on the job. That will drive so much value to the model company that is ahead, at least in my view, because you have copies of one model broadly deployed through the economy learning how to do every single job.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

137 / evaluation

The communists get to claim the prestige of fighting the Japanese with a fraction of the effort. This just seems like a brilliant move on Stalin’s part in retrospect, because if they had killed Chiang Kai-shek, the Nationalist forces might have dissolved.

“The communists get to claim the prestige of fighting the Japanese with a fraction of the effort. This just seems like a brilliant move on Stalin’s part in retrospect, because if they had killed Chiang Kai-shek, the Nationalist forces might have dissolved.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

138 / uncertainty

I want to ask about your, I don’t know if you call it a prediction but, your hypothesis that at some point China and Russia will finish each other off or have some sort of conflict which will be bad for both of them.

“I want to ask about your, I don’t know if you call it a prediction but, your hypothesis that at some point China and Russia will finish each other off or have some sort of conflict which will be bad for both of them.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

140 / belief

I think you’re basically inventing college from first principles for the tools that are available today and just selecting for people who have the motivation and the interest of really engaging with material.

“I think you’re basically inventing college from first principles for the tools that are available today and just selecting for people who have the motivation and the interest of really engaging with material.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

141 / belief

Just to throw the opposite argument against you, my expectation is that it blows up because I think true AGI—and I’m not talking about LLM coding bots, I’m talking about actual replacement of a human in a server—is qualitatively different from these other productivity-improving technologies because it’s labor itself.

“Just to throw the opposite argument against you, my expectation is that it blows up because I think true AGI—and I’m not talking about LLM coding bots, I’m talking about actual replacement of a human in a server—is qualitatively different from these other productivity-improving technologies because it’s labor itself.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

142 / belief

There’s a lot of really smart people who are ready to make use of the resources and do this period of catch-up because we’ve had this discontinuity, and I think AI might be similar.

“There’s a lot of really smart people who are ready to make use of the resources and do this period of catch-up because we’ve had this discontinuity, and I think AI might be similar.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

143 / uncertainty

Then adults are somewhere in between, where they don’t have the flexibility of childhood learning, but they can memorize facts and information in a way that is harder for kids. I don’t know if there’s something interesting about that spectrum.

“Then adults are somewhere in between, where they don’t have the flexibility of childhood learning, but they can memorize facts and information in a way that is harder for kids. I don’t know if there’s something interesting about that spectrum.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

146 / observation

If you buy the Sutton perspective that the crux of intelligence is animal intelligence… The quote he said is “If you got to the squirrel, you’d be most of the way to AGI.

“If you buy the Sutton perspective that the crux of intelligence is animal intelligence… The quote he said is “If you got to the squirrel, you’d be most of the way to AGI.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

149 / evaluation

It’s interesting because you wrote this essay in 2019 titled “The Bitter Lesson,” and this is the most influential essay, perhaps, in the history of AI. But people have used that as a justification for scaling up LLMs because, in their view, this is the one scalable way we have found to pour ungodly amounts of compute into learning about the world.

“It’s interesting because you wrote this essay in 2019 titled “The Bitter Lesson,” and this is the most influential essay, perhaps, in the history of AI. But people have used that as a justification for scaling up LLMs because, in their view, this is the one scalable way we have found to pour ungodly amounts of compute into learning about the world.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

150 / prediction

As we move towards the era of experience, as you call it, this prior is going to be the basis on which we teach these models from experience, because this gives them the opportunity to get answers right some of the time.

“As we move towards the era of experience, as you call it, this prior is going to be the basis on which we teach these models from experience, because this gives them the opportunity to get answers right some of the time.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

152 / belief

If you want to understand what it is that enables humans to go to the moon or to build semiconductors, I think the thing we want to understand is what makes that happen.

“If you want to understand what it is that enables humans to go to the moon or to build semiconductors, I think the thing we want to understand is what makes that happen.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

155 / belief

If we thought, “Oh, the future generation will be Nazis, I think we’d be quite concerned about just handing off power to them.” So I agree that this is not super dissimilar to worrying about more capable future humans, but I don’t think that addresses a lot of the concerns people might have about this level of power being attained this fast with entities we don’t fully understand.

“If we thought, “Oh, the future generation will be Nazis, I think we’d be quite concerned about just handing off power to them.” So I agree that this is not super dissimilar to worrying about more capable future humans, but I don’t think that addresses a lot of the concerns people might have about this level of power being attained this fast with entities we don’t fully understand.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

157 / belief

I think argument is useful. I do want to complete this thought. Joseph Henrich has this interesting theory about a lot of the skills that humans have had to master in order to be successful.

“I think argument is useful. I do want to complete this thought. Joseph Henrich has this interesting theory about a lot of the skills that humans have had to master in order to be successful.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

159 / observation

It’ll say, “Okay, I’m going to approach this problem using this approach first.” It’ll write this out and be like, “Oh wait, I just realized this is the wrong conceptual way to approach the problem.

“It’ll say, “Okay, I’m going to approach this problem using this approach first.” It’ll write this out and be like, “Oh wait, I just realized this is the wrong conceptual way to approach the problem.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

162 / belief

I'd claim that cryptography, radar, even oil… Even if Germany had a huge reserve of oil, maybe it would have lengthened the length at which Germany could have sustained itself. But if you change the fact that the Axis had one-fourth of the GDP of the Allies combined, I think switching that genuinely changes who wins the war, whereas these other things actually wouldn't change who won the war.

“I'd claim that cryptography, radar, even oil… Even if Germany had a huge reserve of oil, maybe it would have lengthened the length at which Germany could have sustained itself. But if you change the fact that the Axis had one-fourth of the GDP of the Allies combined, I think switching that genuinely changes who wins the war, whereas these other things actually wouldn't change who won the war.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

164 / belief

AKA, maybe intelligence is easier than we think, and there's a bunch of contingent reasons evolution didn't churn as hard on this variable as it could have.

“AKA, maybe intelligence is easier than we think, and there's a bunch of contingent reasons evolution didn't churn as hard on this variable as it could have.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

165 / uncertainty

I'm not sure how to understand this claim that we know how to engage with the right hook, we just don't know what that hook is supposed to do in the body. I don't know if that's the way you describe it.

“I'm not sure how to understand this claim that we know how to engage with the right hook, we just don't know what that hook is supposed to do in the body. I don't know if that's the way you describe it.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

166 / prediction

This year you spend $100 million training a model, next year $1 billion, the year after that, $10 billion. But it's one general purpose model, unlike, “We made money on this drug and now we're going to use that money to invest in 10 different drugs in 10 different bespoke ways.

“This year you spend $100 million training a model, next year $1 billion, the year after that, $10 billion. But it's one general purpose model, unlike, “We made money on this drug and now we're going to use that money to invest in 10 different drugs in 10 different bespoke ways.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

167 / commitment

Right now we are export controlling chips for the purpose of keeping our AI lead and we recognize this is a key input in our ability to compete in AI. So we are going to export control China's ability to have these chips.

“Right now we are export controlling chips for the purpose of keeping our AI lead and we recognize this is a key input in our ability to compete in AI. So we are going to export control China's ability to have these chips.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

169 / uncertainty

At current hardware efficiencies. I don't know if it's worth spelling out. Basically, an H100 has the same amount of flops as a human brain, but also uses way more energy than a human brain.

“At current hardware efficiencies. I don't know if it's worth spelling out. Basically, an H100 has the same amount of flops as a human brain, but also uses way more energy than a human brain.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

170 / evaluation

, were like, "We're not ramping up HBM production because HBM is used largely for AI workloads, and if this demand doesn't continue, then our additional manufacturing capacity for HBM will not have been worth it.

“, were like, "We're not ramping up HBM production because HBM is used largely for AI workloads, and if this demand doesn't continue, then our additional manufacturing capacity for HBM will not have been worth it.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

171 / evaluation

I agree that they've obviously made bad decisions, but even if you have the poorest Chinese people anywhere in the world, they can still be quite rich.

“I agree that they've obviously made bad decisions, but even if you have the poorest Chinese people anywhere in the world, they can still be quite rich.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

174 / uncertainty

To the extent that we hold consumption constant, and maybe we shouldn't, they would have to be suffering 4x as much as a chicken in the 1950s for it not to be a net improvement. I don't know if you disagree with that.

“To the extent that we hold consumption constant, and maybe we shouldn't, they would have to be suffering 4x as much as a chicken in the 1950s for it not to be a net improvement. I don't know if you disagree with that.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

176 / observation

The reason why that’s so shocking…On its face it's shocking. But in other areas where you're trying to do global health or something, first, the problem is improving on its own.

“The reason why that’s so shocking…On its face it's shocking. But in other areas where you're trying to do global health or something, first, the problem is improving on its own.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

179 / prediction

I think people will be aware that there's a general problem here, but the actual politics, the actual economics, the actual state of the technology landscape here… there might be interventions which are stupendously effective, which we overlook just because people are not aware of what's actually happening in this space.

“I think people will be aware that there's a general problem here, but the actual politics, the actual economics, the actual state of the technology landscape here… there might be interventions which are stupendously effective, which we overlook just because people are not aware of what's actually happening in this space.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

181 / belief

One really interesting takeaway from this lecture—and I'm curious if you agree with this—is that when we think about this period in history, we often think of Japan as the rising power in Asia.

“One really interesting takeaway from this lecture—and I'm curious if you agree with this—is that when we think about this period in history, we often think of Japan as the rising power in Asia.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

182 / belief

I agree with the mechanism by which the growth happened, but I don't think it's the case that it was their inability to have true Marxist communism which led to liberalization.

“I agree with the mechanism by which the growth happened, but I don't think it's the case that it was their inability to have true Marxist communism which led to liberalization.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

183 / commitment

I agree with your general point that how any nation gets wealthy is not by the government but because of the thrift and entrepreneurialism and hard work of individuals.

“I agree with your general point that how any nation gets wealthy is not by the government but because of the thrift and entrepreneurialism and hard work of individuals.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

184 / observation

In many of these cases, there's this period in between when they're a dead man walking—because Stalin has started putting the feelers out that this person is a Trotskyite or something—but they're still in their positions of power.

“In many of these cases, there's this period in between when they're a dead man walking—because Stalin has started putting the feelers out that this person is a Trotskyite or something—but they're still in their positions of power.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

186 / belief

I think on net it was a good thing that the NSF and the NIH and all these budgets were blown up and got DOGE’d and so forth…” I'm not saying you think this is likely, but suppose there ends up being a positive story told in retrospect.

“I think on net it was a good thing that the NSF and the NIH and all these budgets were blown up and got DOGE’d and so forth…” I'm not saying you think this is likely, but suppose there ends up being a positive story told in retrospect.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

187 / evaluation

I guess I am curious if there is some long-run vision, where… To give another example, in cybersecurity, as time has gone on, I think our systems are more secure today than they were in the past because we found vulnerabilities and we've come up with new encryption schemes and so forth.

“I guess I am curious if there is some long-run vision, where… To give another example, in cybersecurity, as time has gone on, I think our systems are more secure today than they were in the past because we found vulnerabilities and we've come up with new encryption schemes and so forth.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

188 / evaluation

Germany, Japan, and many other countries missed out on the Internet because there was a centrally directed effort towards these heavy industries or manufacturing, which actually turned out not to be relevant in the 21st century.

“Germany, Japan, and many other countries missed out on the Internet because there was a centrally directed effort towards these heavy industries or manufacturing, which actually turned out not to be relevant in the 21st century.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

191 / belief

There I haven't crunched the numbers, but if I were to guess, I don't think with TSMC's leading-edge 5 nanometer wafer, I doubt very little of what it costs to make that is the process engineers themselves.

“There I haven't crunched the numbers, but if I were to guess, I don't think with TSMC's leading-edge 5 nanometer wafer, I doubt very little of what it costs to make that is the process engineers themselves.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

194 / uncertainty

I don't know if that's the way you would still describe the way in which these software agents aren't able to do a full day of work, but are able to help you out with a couple minutes.

“I don't know if that's the way you would still describe the way in which these software agents aren't able to do a full day of work, but are able to help you out with a couple minutes.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

195 / belief

I think the crux comes down to the people who expect something much longer have a sense that… When I had Ege and Tamay on my podcast, they were like, "Look, you could look at AlphaGo, and say, 'Oh, this is a model that can do exploration.

“I think the crux comes down to the people who expect something much longer have a sense that… When I had Ege and Tamay on my podcast, they were like, "Look, you could look at AlphaGo, and say, 'Oh, this is a model that can do exploration.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

196 / belief

Speaking of inference compute, one thing that I think is not talked about enough is, if you do live in the world that you're painting—in a year or two, we have computer use agents that are doing actual jobs, you've totally automated large parts of software engineering—then these models are going to be incredibly valuable to use.

“Speaking of inference compute, one thing that I think is not talked about enough is, if you do live in the world that you're painting—in a year or two, we have computer use agents that are doing actual jobs, you've totally automated large parts of software engineering—then these models are going to be incredibly valuable to use.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

197 / belief

I think in the addition example, you said in the paper that the way it actually does the addition is different from the way it tells you it does the addition.

“I think in the addition example, you said in the paper that the way it actually does the addition is different from the way it tells you it does the addition.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

198 / commitment

If this scale of compute increase can’t continue beyond 2030—not just because of chips, but also because of power and raw GDP even—then because we don't think we will get it by 2030 or 2028, then the probability per year just goes down a bunch.

“If this scale of compute increase can’t continue beyond 2030—not just because of chips, but also because of power and raw GDP even—then because we don't think we will get it by 2030 or 2028, then the probability per year just goes down a bunch.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

199 / belief

Yeah, it would be good from an alignment perspective, too. Because I think you kind of do need a wider range of skills before you can do something super scary.

“Yeah, it would be good from an alignment perspective, too. Because I think you kind of do need a wider range of skills before you can do something super scary.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

200 / uncertainty

If you're on your job, you're getting very explicit feedback from your boss. That's not necessarily how the task should be done differently, but a high-level explanation of what you did wrong, which you update on not in the way that pre-training updates weights, but more in the… I don’t know.

“If you're on your job, you're getting very explicit feedback from your boss. That's not necessarily how the task should be done differently, but a high-level explanation of what you did wrong, which you update on not in the way that pre-training updates weights, but more in the… I don’t know.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

202 / belief

I think one thing that's not appreciated enough is how much of our leverage on the future—given the fact that our labor isn't going to be worth that much—comes from our economic, and political systems surviving.

“I think one thing that's not appreciated enough is how much of our leverage on the future—given the fact that our labor isn't going to be worth that much—comes from our economic, and political systems surviving.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

204 / uncertainty

Sometimes it goes off the rails, obviously, but I don't know… You could make a theoretical argument that you teach a kid to make a lot of money when he grows up and a lot of smart people are imbued with those values and just rarely become psychopaths or something.

“Sometimes it goes off the rails, obviously, but I don't know… You could make a theoretical argument that you teach a kid to make a lot of money when he grows up and a lot of smart people are imbued with those values and just rarely become psychopaths or something.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

205 / belief

I think there's a paper from Tsinghua University, where they showed that if you give a base model enough tries to answer a question, it can still answer the question as well as the reasoning model.

“I think there's a paper from Tsinghua University, where they showed that if you give a base model enough tries to answer a question, it can still answer the question as well as the reasoning model.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

207 / belief

Does that mean alignment is easier than we think just because you just have to write a bunch of fake news articles that say, "AIs just love humanity and they just want to do good things.

“Does that mean alignment is easier than we think just because you just have to write a bunch of fake news articles that say, "AIs just love humanity and they just want to do good things.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

208 / prediction

Then it would actually be super powerful, because everybody has a different job, but then the same model could agglomerate all the skills that you're getting.

“Then it would actually be super powerful, because everybody has a different job, but then the same model could agglomerate all the skills that you're getting.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

209 / belief

DeepSeek has the MIT license, whereas I think a couple of the contingencies in the Llama license require you to say "built with Llama" on applications using it or any model that you train using Llama has to begin with the word "Llama.

“DeepSeek has the MIT license, whereas I think a couple of the contingencies in the Llama license require you to say "built with Llama" on applications using it or any model that you train using Llama has to begin with the word "Llama.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

210 / belief

On the point of what the different labs are optimizing for — to steelman their view — I think a lot of them believe that once you fully automate software engineering and AI research, then you can kick off an intelligence explosion.

“On the point of what the different labs are optimizing for — to steelman their view — I think a lot of them believe that once you fully automate software engineering and AI research, then you can kick off an intelligence explosion.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

212 / belief

I think that's an excellent note to close on. Just to plug one more time, we've been discussing Plagues Upon the Earth, which is the history of disease going back through the Neolithic to modern times, Fate of Rome which discusses the plagues and history of the Roman Empire considering climate and biology.

“I think that's an excellent note to close on. Just to plug one more time, we've been discussing Plagues Upon the Earth, which is the history of disease going back through the Neolithic to modern times, Fate of Rome which discusses the plagues and history of the Roman Empire considering climate and biology.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

218 / evaluation

The reason I wanted to have you on is because I don't think I've encountered that many other authors who can connect biology, economics, history, and climate into explaining some of the big things that have happened through human history in the way you can.

“The reason I wanted to have you on is because I don't think I've encountered that many other authors who can connect biology, economics, history, and climate into explaining some of the big things that have happened through human history in the way you can.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

220 / prediction

I think the actual fertility for foragers is sort of like reasonable—I don’t know if sustainable is the right word because I don’t mean in an ecological sense but more so—it keeps your population constant.

“I think the actual fertility for foragers is sort of like reasonable—I don’t know if sustainable is the right word because I don’t mean in an ecological sense but more so—it keeps your population constant.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

221 / belief

The thing worth noting about the future is that most of the people who will ever exist are going to be digital. And look, I think factory farming is incredibly bad.

“The thing worth noting about the future is that most of the people who will ever exist are going to be digital. And look, I think factory farming is incredibly bad.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

222 / evaluation

I think both of those examples don’t work in your favor. I think the China growth miracle could not have occurred if not for their ability to copy technology from the west and I don’t think there’s a world in which they… China has a lot of really smart people, it’s a big country in general.

“I think both of those examples don’t work in your favor. I think the China growth miracle could not have occurred if not for their ability to copy technology from the west and I don’t think there’s a world in which they… China has a lot of really smart people, it’s a big country in general.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

223 / belief

I think a lot of your forecast relies on, at some point, not only the US President, but also Xi Jinping, waking up to the possibility of a super intelligence and the stakes involved there.

“I think a lot of your forecast relies on, at some point, not only the US President, but also Xi Jinping, waking up to the possibility of a super intelligence and the stakes involved there.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

224 / belief

Where right now, if you told your parents, “I’m going to become a startup founder”, I think the reaction would be like, “there’s a 1% chance you’ll succeed, but it’s an interesting experience and if you do succeed, that’s crazy.

“Where right now, if you told your parents, “I’m going to become a startup founder”, I think the reaction would be like, “there’s a 1% chance you’ll succeed, but it’s an interesting experience and if you do succeed, that’s crazy.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

225 / belief

One observation I have is, you could have told a story in 2021, once ChatGPT comes out… I think I had friends who were credible AI thinkers who were like, “look, you’ve got the coding agent now, it’s been cracked.

“One observation I have is, you could have told a story in 2021, once ChatGPT comes out… I think I had friends who were credible AI thinkers who were like, “look, you’ve got the coding agent now, it’s been cracked.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

226 / uncertainty

Let me just add on to that, one of the many other reasons why I worry about nationalization or some kind of public private partnership, or even just very stringent regulation- actually, this is more an argument against very stringent regulation in favor of safety rather than deferring more to the labs on the implementation- is that it just seems like we don’t know what we don’t know about alignment.

“Let me just add on to that, one of the many other reasons why I worry about nationalization or some kind of public private partnership, or even just very stringent regulation- actually, this is more an argument against very stringent regulation in favor of safety rather than deferring more to the labs on the implementation- is that it just seems like we don’t know what we don’t know about alignment.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

227 / commitment

” I will mention something that has actually radicalized me against Twitter as an information source is I’ll meet- and this has happened multiple times- I’ll meet somebody who seems to be an interesting poster, has funny, seemingly insightful posts on Twitter.

“” I will mention something that has actually radicalized me against Twitter as an information source is I’ll meet- and this has happened multiple times- I’ll meet somebody who seems to be an interesting poster, has funny, seemingly insightful posts on Twitter.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

228 / belief

I guess the bigger the organization, even if everybody is aligned- I think some of your responses addressed whether they will be aligned on goals.

“I guess the bigger the organization, even if everybody is aligned- I think some of your responses addressed whether they will be aligned on goals.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

229 / belief

I agree. We’re on a totally different topic here of, do you get a Dyson sphere? There’s one world where it’s crazy but it’s still boring, in the sense that the economy is growing much faster, but it would be like what the Industrial Revolution would look like to somebody in the year 1000.

“I agree. We’re on a totally different topic here of, do you get a Dyson sphere? There’s one world where it’s crazy but it’s still boring, in the sense that the economy is growing much faster, but it would be like what the Industrial Revolution would look like to somebody in the year 1000.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

230 / uncertainty

I don’t know if ‘emotionally’ is the right word, but their expectation of what their life might look like, even in the world where there’s no doom.

“I don’t know if ‘emotionally’ is the right word, but their expectation of what their life might look like, even in the world where there’s no doom.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

231 / belief

Maybe the reason that this sounds less plausible to me than the 25x number implies is that when I think about concretely what that would look like, where you have these AIs and we know that there’s a gap in data efficiency between human brains and these AIs.

“Maybe the reason that this sounds less plausible to me than the 25x number implies is that when I think about concretely what that would look like, where you have these AIs and we know that there’s a gap in data efficiency between human brains and these AIs.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

232 / uncertainty

Actually, when we did our episode together, a bunch of people, I don't know if you saw this, independently made these blog posts and Anki cards and shit where they're explaining the concept because we just kind of passed over some things.

“Actually, when we did our episode together, a bunch of people, I don't know if you saw this, independently made these blog posts and Anki cards and shit where they're explaining the concept because we just kind of passed over some things.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

233 / commitment

I will also mention that all I have to do is ask them questions, and I do think it's much harder to learn a field to be a practitioner than just learn enough to ask interesting questions.

“I will also mention that all I have to do is ask them questions, and I do think it's much harder to learn a field to be a practitioner than just learn enough to ask interesting questions.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

234 / uncertainty

The human brain can store much more, so why does the human brain just not want us to memorize these kinds of things, and is actively pruning, and… yeah, I don't know.

“The human brain can store much more, so why does the human brain just not want us to memorize these kinds of things, and is actively pruning, and… yeah, I don't know.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

240 / evaluation

This is also why 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're in one of the better..

“This is also why 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're in one of the better..”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

241 / commitment

Mostly because- 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 Speech Marking Cards, because all the insights have just sort of faded away.

“Mostly because- 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 Speech Marking Cards, because all the insights have just sort of faded away.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

242 / evaluation

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.

“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.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

244 / prediction

I feel like that job's still gonna be around. It's funny, because I studied computer science, and in retrospect- at the time, you could've become a software engineer or something.

“I feel like that job's still gonna be around. It's funny, because I studied computer science, and in retrospect- at the time, you could've become a software engineer or something.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

246 / belief

Now, when I think about what's going to be possible in the next 250 years, I'm thinking like space travel, and space elevators, and immortality, and curing all diseases.

“Now, when I think about what's going to be possible in the next 250 years, I'm thinking like space travel, and space elevators, and immortality, and curing all diseases.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

247 / belief

I think there's just, there are so many different applications that you have put out for using these models to make the different areas you talked about better.

“I think there's just, there are so many different applications that you have put out for using these models to make the different areas you talked about better.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

252 / belief

One of the things I learned from your book is that the overwhelming fraction of deaths on the Japanese side during the war happened after it was known that they were going to lose.

“One of the things I learned from your book is that the overwhelming fraction of deaths on the Japanese side during the war happened after it was known that they were going to lose.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

253 / uncertainty

Here’s another thing that I want to clarify: If you were trying to understand Britain's conduct in World War I, why they initiated it and why they conducted it in the way they did, and you tried to understand it using cultural explanations — what some British guy wrote in the 17th century — I don't know how far you'd get.

“Here’s another thing that I want to clarify: If you were trying to understand Britain's conduct in World War I, why they initiated it and why they conducted it in the way they did, and you tried to understand it using cultural explanations — what some British guy wrote in the 17th century — I don't know how far you'd get.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

256 / belief

Wasn't Churchill, by the time he was 24, an international correspondent in Cuba and India? I think he was the highest-paid journalist in the world by the time he was 24.

“Wasn't Churchill, by the time he was 24, an international correspondent in Cuba and India? I think he was the highest-paid journalist in the world by the time he was 24.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

258 / belief

I think we've mentioned a couple of other bounds like this where there's no principled reason you might have anticipated ex ante why there would be such a bound that prevents something that just gets in our way, but it just so happens to be this way.

“I think we've mentioned a couple of other bounds like this where there's no principled reason you might have anticipated ex ante why there would be such a bound that prevents something that just gets in our way, but it just so happens to be this way.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

259 / belief

There's something to that sort of finality which makes existential risk salient in the first place. And if you agree with that intuition, then I think you should be inclined to think that there is something significant about the fact that in some base reality, like genuinely the story carries forward.

“There's something to that sort of finality which makes existential risk salient in the first place. And if you agree with that intuition, then I think you should be inclined to think that there is something significant about the fact that in some base reality, like genuinely the story carries forward.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

260 / uncertainty

I'm not sure if you're suggesting in the physical universe or in some hypothetical universe where the vacuum could be different, as in, in reality there are other pockets with different vacuums, or that hypothetically they could exist, or that our universe counterfactually could have been one of these? I don't know.

“I'm not sure if you're suggesting in the physical universe or in some hypothetical universe where the vacuum could be different, as in, in reality there are other pockets with different vacuums, or that hypothetically they could exist, or that our universe counterfactually could have been one of these? I don't know.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

262 / evaluation

It's a little bit confusing because in one context, we're laying out very practical—I don't know if you can call black hole batteries practical—but very tangible limitations on the what future, like very distant future descendants, could do with all the matter in the galaxy and so forth.

“It's a little bit confusing because in one context, we're laying out very practical—I don't know if you can call black hole batteries practical—but very tangible limitations on the what future, like very distant future descendants, could do with all the matter in the galaxy and so forth.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

264 / uncertainty

Borges has a short poem called "Borges and I" where he talks about how he doesn’t identify with the version of himself that is actually doing the writing and publishing all of this great work. I don’t know if you identify with that at all.

“Borges has a short poem called "Borges and I" where he talks about how he doesn’t identify with the version of himself that is actually doing the writing and publishing all of this great work. I don’t know if you identify with that at all.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

265 / preference

One of my favorite blog posts of yours is “Evolution as Backstop for RL,” where you talk about evolution as basically a mechanism to learn a better learning process.

“One of my favorite blog posts of yours is “Evolution as Backstop for RL,” where you talk about evolution as basically a mechanism to learn a better learning process.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

272 / belief

I've found that there are a lot of books which are nominally about one subject, but the author just feels a need to say, "If you really want to understand my topic, you have to understand basically everything else in the world." I think of a couple of biographies especially.

“I've found that there are a lot of books which are nominally about one subject, but the author just feels a need to say, "If you really want to understand my topic, you have to understand basically everything else in the world." I think of a couple of biographies especially.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

273 / uncertainty

I imagine since you wrote The Prize, world leaders are inviting you to meet them and give advice. I don't know how many stories you can tell from these conversations.

“I imagine since you wrote The Prize, world leaders are inviting you to meet them and give advice. I don't know how many stories you can tell from these conversations.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

278 / prediction

I don’t know if “instigated” is the right word, but the Pacific War was instigated because the Japanese needed more oil because of the war in Manchuria.

“I don’t know if “instigated” is the right word, but the Pacific War was instigated because the Japanese needed more oil because of the war in Manchuria.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

283 / evaluation

One thing I didn't realize until I read your book is how small the population that expanded out into Eurasia was, and how small even generally the human population was 50,000 to 100,000 years ago.

“One thing I didn't realize until I read your book is how small the population that expanded out into Eurasia was, and how small even generally the human population was 50,000 to 100,000 years ago.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

285 / evaluation

The theory he talks about in the book is that potentially the reason the hunter-gatherers, the “barbarians,” couldn't fight back against these early nation-states was because they were getting killed off by the diseases.

“The theory he talks about in the book is that potentially the reason the hunter-gatherers, the “barbarians,” couldn't fight back against these early nation-states was because they were getting killed off by the diseases.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

291 / evaluation

With these models, at least so far, it doesn't seem to matter. They just get, "Hey, don't help people make bombs" or whatever, even if you ask in a different way how to make a bomb.

“With these models, at least so far, it doesn't seem to matter. They just get, "Hey, don't help people make bombs" or whatever, even if you ask in a different way how to make a bomb.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

295 / uncertainty

We're starting off with something that has this intricate representation of human values and it doesn't seem that hard to sort of lock it into a persona that we are comfortable with. I don't know what changes.

“We're starting off with something that has this intricate representation of human values and it doesn't seem that hard to sort of lock it into a persona that we are comfortable with. I don't know what changes.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

298 / uncertainty

I don't know if moral realism is the right word, but you mentioned the thing. There's something that makes hearts converge to the thing we are or the thing we would be upon reflection.

“I don't know if moral realism is the right word, but you mentioned the thing. There's something that makes hearts converge to the thing we are or the thing we would be upon reflection.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

299 / prediction

I expect the US government to protect me, not because of its “motives,” but just because of the system of incentives and institutions and norms that has been set up.

“I expect the US government to protect me, not because of its “motives,” but just because of the system of incentives and institutions and norms that has been set up.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

300 / evaluation

I agree with the sentiment of obviously approaching this situation with caution, but I do want to point out the ways in which the analyses we've been using have been maximally adversarial.

“I agree with the sentiment of obviously approaching this situation with caution, but I do want to point out the ways in which the analyses we've been using have been maximally adversarial.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

301 / evaluation

They're almost like random number generators. They're not especially calibrated, but once in a while they'll be like, “Oh, this one weird philosopher I care about, or this one historical event I'm obsessed with has an interesting perspective on this.

“They're almost like random number generators. They're not especially calibrated, but once in a while they'll be like, “Oh, this one weird philosopher I care about, or this one historical event I'm obsessed with has an interesting perspective on this.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

302 / recommendation

I don’t know practically if it works out that well. But that experience made me think that I should try to expand my horizons in an undirected way because there’s lots of different things you have to understand about the world to understand any one thing.

“I don’t know practically if it works out that well. But that experience made me think that I should try to expand my horizons in an undirected way because there’s lots of different things you have to understand about the world to understand any one thing.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

303 / evaluation

To the extent that the reason we're worried about motivations in the first place, it’s because we think a balance of power which includes at least one thing with human-descended motivations is difficult.

“To the extent that the reason we're worried about motivations in the first place, it’s because we think a balance of power which includes at least one thing with human-descended motivations is difficult.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

304 / evaluation

I'm curious where that's coming from because conventionally I think the thing matters because it's conscious and its conscious experience as a result of that pursuit matters.

“I'm curious where that's coming from because conventionally I think the thing matters because it's conscious and its conscious experience as a result of that pursuit matters.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

305 / evaluation

The native infrastructure that we had — that was specifically earmarked for dealing with public health emergencies — was extremely incompetent to the extent that Discord servers vastly outperformed them.

“The native infrastructure that we had — that was specifically earmarked for dealing with public health emergencies — was extremely incompetent to the extent that Discord servers vastly outperformed them.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

309 / uncertainty

” I'm thinking to myself, how is this not as trivial a problem as, “hey XYZ, if you give me money that you can find between your couch cushions, we will save thousands of lives and get the world economy back on track.

“” I'm thinking to myself, how is this not as trivial a problem as, “hey XYZ, if you give me money that you can find between your couch cushions, we will save thousands of lives and get the world economy back on track.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

317 / belief

I agree that nuclear energy is a thing that happened later on and is dual-use. But it’s something that happened literally a decade after nuclear weapons were developed.

“I agree that nuclear energy is a thing that happened later on and is dual-use. But it’s something that happened literally a decade after nuclear weapons were developed.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

318 / uncertainty

Maybe you have a point. We don’t know. You have arguments for why that’s a more likely world, but maybe that’s not the world we live in.

“Maybe you have a point. We don’t know. You have arguments for why that’s a more likely world, but maybe that’s not the world we live in.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

319 / belief

Unless you’re right about what the intelligence explosion looks like, don’t move yet. But in that world where it really does seem like Alec Radford can be automated, and that's the only bottleneck to getting to ASI… Okay I think we can leave it at that.

“Unless you’re right about what the intelligence explosion looks like, don’t move yet. But in that world where it really does seem like Alec Radford can be automated, and that's the only bottleneck to getting to ASI… Okay I think we can leave it at that.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

320 / observation

On that point in particular, many people who have longer timelines have come on the podcast and made the point that the way to train this long horizon RL, it's not… Earlier we were talking about how they can think for five minutes, but not for longer.

“On that point in particular, many people who have longer timelines have come on the podcast and made the point that the way to train this long horizon RL, it's not… Earlier we were talking about how they can think for five minutes, but not for longer.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

323 / evaluation

Now we're in a place where they are excited about AGI and they're like, "fuck, we want to have GPT-5 while you're going to be off building superintelligence. This Atoms for Peace thing doesn't work for us.

“Now we're in a place where they are excited about AGI and they're like, "fuck, we want to have GPT-5 while you're going to be off building superintelligence. This Atoms for Peace thing doesn't work for us.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

324 / belief

I agree, but if the coefficient, of how fast they diminish as you grow the input, is high enough, then in the abstract the fact that inputs matter isn’t that relevant.

“I agree, but if the coefficient, of how fast they diminish as you grow the input, is high enough, then in the abstract the fact that inputs matter isn’t that relevant.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

325 / observation

I was told by people advancing this idea — because they know I’m a libertarian-type person and the way they approached me was like this — that the way to think about it was that it's fundamentally a way to protect market-based development of AGI.

“I was told by people advancing this idea — because they know I’m a libertarian-type person and the way they approached me was like this — that the way to think about it was that it's fundamentally a way to protect market-based development of AGI.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

328 / belief

I'm not sure I agree with this. There are many examples in history where small groups of people, Bell Labs or Skunk Works, have made significant progress.

“I'm not sure I agree with this. There are many examples in history where small groups of people, Bell Labs or Skunk Works, have made significant progress.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

335 / recommendation

I want to go back to the point you made earlier about how this process could be more sample efficient because it could generalize from its pre-training experiences of how to get unstuck in different scenarios.

“I want to go back to the point you made earlier about how this process could be more sample efficient because it could generalize from its pre-training experiences of how to get unstuck in different scenarios.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

336 / preference

On the other hand, I think I heard you make the point that a lot of our preferences and values are very subtle, so they might be best represented through pairwise preferences.

“On the other hand, I think I heard you make the point that a lot of our preferences and values are very subtle, so they might be best represented through pairwise preferences.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

337 / prediction

Because there doesn't seem to be a model since GPT-4 that seems to be significantly better, there's a hypothesis that we might be hitting some sort of plateau.

“Because there doesn't seem to be a model since GPT-4 that seems to be significantly better, there's a hypothesis that we might be hitting some sort of plateau.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

340 / belief

Regarding other open source dangers, I think you have genuine legitimate points about the balance of power stuff and potentially the harms you can get rid of because we have better alignment techniques or something.

“Regarding other open source dangers, I think you have genuine legitimate points about the balance of power stuff and potentially the harms you can get rid of because we have better alignment techniques or something.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

341 / preference

There's a case in which you don't want to open source the architecture because China can use it to catch up to America's AIs and there is an intelligence explosion and they win that.

“There's a case in which you don't want to open source the architecture because China can use it to catch up to America's AIs and there is an intelligence explosion and they win that.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

346 / uncertainty

I guess if you think of something like the laws of physics, it's not that the feature for wetness is turned on, but it's only turned on this much and then the feature for… I guess maybe it's true because the mass is like a gradient and… I don't know.

“I guess if you think of something like the laws of physics, it's not that the feature for wetness is turned on, but it's only turned on this much and then the feature for… I guess maybe it's true because the mass is like a gradient and… I don't know.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

350 / evaluation

In fact, it's good for the world that that's not often how it happens. It is important to look at, “were they able to write an interesting technical blog post about their research or are they making interesting contributions.

“In fact, it's good for the world that that's not often how it happens. It is important to look at, “were they able to write an interesting technical blog post about their research or are they making interesting contributions.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

351 / belief

Basically every single high-profile guest I've done so far, I think maybe with one or two exceptions, I've sat down for a week and I've just come up with a list of sample questions.

“Basically every single high-profile guest I've done so far, I think maybe with one or two exceptions, I've sat down for a week and I've just come up with a list of sample questions.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

352 / uncertainty

There's some process by which to interpret the early data. I don't know. I could put a Google Doc in front of you and I'm pretty sure you could just keep typing for a while on different ideas you have.

“There's some process by which to interpret the early data. I don't know. I could put a Google Doc in front of you and I'm pretty sure you could just keep typing for a while on different ideas you have.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

354 / uncertainty

There was an interesting paper that you can use diffusion to come up with model weights. I don't know how legit that was or whatever, but something like that.

“There was an interesting paper that you can use diffusion to come up with model weights. I don't know how legit that was or whatever, but something like that.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

356 / prediction

Maybe it thinks in alien concepts and you can’t really monitor the million-line pull request because you can’t really understand the whole thing and you can’t give labels.

“Maybe it thinks in alien concepts and you can’t really monitor the million-line pull request because you can’t really understand the whole thing and you can’t give labels.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

357 / prediction

One is that as these models get smarter, they are going to be able to operate in domains where we just can’t generate enough human labels, just because we’re not smart enough.

“One is that as these models get smarter, they are going to be able to operate in domains where we just can’t generate enough human labels, just because we’re not smart enough.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

358 / uncertainty

There’s a perception that maybe other labs are more compute-efficient than DeepMind has been with Gemini. I don’t know what you make of that perception.

“There’s a perception that maybe other labs are more compute-efficient than DeepMind has been with Gemini. I don’t know what you make of that perception.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

359 / belief

I don't quite know how to synthesize it yet, but as I think about advice for people in their 20s, I'm not going to normatively pretend to know or presume in which direction one should go in life.

“I don't quite know how to synthesize it yet, but as I think about advice for people in their 20s, I'm not going to normatively pretend to know or presume in which direction one should go in life.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

360 / belief

I think you're being too humble. Staying on Fast Grants, now we have the retrospective of how effective the fast grants recipients were, compared to the other grants that were given out by, let's say, the NIH or NSF.

“I think you're being too humble. Staying on Fast Grants, now we have the retrospective of how effective the fast grants recipients were, compared to the other grants that were given out by, let's say, the NIH or NSF.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

362 / prediction

By the way, it's not the Dwarkesh podcast, it's Lunar Society Podcast LLC registered on Stripe Atlas. Any merchandise I sell in the future, Stripe will take care of that.

“By the way, it's not the Dwarkesh podcast, it's Lunar Society Podcast LLC registered on Stripe Atlas. Any merchandise I sell in the future, Stripe will take care of that.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

364 / belief

I think there's an interesting thread here in how it relates to Stripe climate, in that you're subsidizing learning curves that East Asian countries did for their own internal companies.

“I think there's an interesting thread here in how it relates to Stripe climate, in that you're subsidizing learning curves that East Asian countries did for their own internal companies.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

365 / evaluation

The specific case for Stripe working with them is: typically they're coming to us not because they want to take the thing that they're already doing and go through all the work of transposing to Stripe, but because either they want to do a new thing, that they're not doing today — so it is associated with some new business line or innovation or invention.

“The specific case for Stripe working with them is: typically they're coming to us not because they want to take the thing that they're already doing and go through all the work of transposing to Stripe, but because either they want to do a new thing, that they're not doing today — so it is associated with some new business line or innovation or invention.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

366 / observation

” Now, it’s a fascinating idea that investment is irrational, or most investment throughout history has been irrational. But when we think today about the fact that active investing exists for winners’ curse like reasons, VCs probably make, on average, less returns than the market, there’s a whole bunch of different examples you can go through, right?

“” Now, it’s a fascinating idea that investment is irrational, or most investment throughout history has been irrational. But when we think today about the fact that active investing exists for winners’ curse like reasons, VCs probably make, on average, less returns than the market, there’s a whole bunch of different examples you can go through, right?”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

370 / belief

I wonder if you’re doing enough of that when it comes to AI, where I think you have really interesting thoughts about GPD, five-level stuff, but somebody with your sort of polymathic understanding of different fields, if you just extrapolate out these trends, it seems like you might have a lot of interesting thoughts about what might be possible with something much further down the line.

“I wonder if you’re doing enough of that when it comes to AI, where I think you have really interesting thoughts about GPD, five-level stuff, but somebody with your sort of polymathic understanding of different fields, if you just extrapolate out these trends, it seems like you might have a lot of interesting thoughts about what might be possible with something much further down the line.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

371 / commitment

Another quote from Keynes, I guess I won’t read the whole quote in full, but basically says, over time, as investments, markets get more mature, more and more of equities are held basically by passive investors, people who don’t have a direct hand in the involvement of the enterprise, and the share of the market that’s passive investment now is much bigger.

“Another quote from Keynes, I guess I won’t read the whole quote in full, but basically says, over time, as investments, markets get more mature, more and more of equities are held basically by passive investors, people who don’t have a direct hand in the involvement of the enterprise, and the share of the market that’s passive investment now is much bigger.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

372 / belief

Relatedly, I think, a couple of years ago, Paul Schlemming had an interesting paper that if you look from 1311 to now, interest rates have been declining.

“Relatedly, I think, a couple of years ago, Paul Schlemming had an interesting paper that if you look from 1311 to now, interest rates have been declining.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

374 / belief

When you say there’ll be uncertainties, I think you made this argument when you were responding to Alex Epstein on fossil future, where you said uncertainties also extend out into the domain where there’s a bad outcome or much bigger outcome than you’re anticipating.

“When you say there’ll be uncertainties, I think you made this argument when you were responding to Alex Epstein on fossil future, where you said uncertainties also extend out into the domain where there’s a bad outcome or much bigger outcome than you’re anticipating.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

376 / belief

Sure. So before we get into his actual views, I think his career is a tremendous white pill in the sense that he writes The Road to Serfdom in 1944 when Nazi Germany and Soviet Union are both prominent players.

“Sure. So before we get into his actual views, I think his career is a tremendous white pill in the sense that he writes The Road to Serfdom in 1944 when Nazi Germany and Soviet Union are both prominent players.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

377 / belief

You talk in the book about how Mill is very concerned about the quality and the character development of the population. But when we think about the fact that somebody like him was elected to the parliament at the time, the greatest thinker who’s alive is elected to government, and it’s hard to imagine that could be true in today’s world.

“You talk in the book about how Mill is very concerned about the quality and the character development of the population. But when we think about the fact that somebody like him was elected to the parliament at the time, the greatest thinker who’s alive is elected to government, and it’s hard to imagine that could be true in today’s world.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

379 / evaluation

Yeah. So, related to that, I think in the meaning of competition, he makes the point that the most interesting part of markets is when they go from one equilibrium to another, because that’s where they’re trying to figure out what to produce and how to produce it better and so on, and not the equilibriums themselves.

“Yeah. So, related to that, I think in the meaning of competition, he makes the point that the most interesting part of markets is when they go from one equilibrium to another, because that’s where they’re trying to figure out what to produce and how to produce it better and so on, and not the equilibriums themselves.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

381 / evaluation

This shows that this was Mao's doing. This is also an interesting example where, whether it's Stalin in Russia or Mao in China, when the tyrant dies, the system automatically improves because nobody else is as crazy as that guy.

“This shows that this was Mao's doing. This is also an interesting example where, whether it's Stalin in Russia or Mao in China, when the tyrant dies, the system automatically improves because nobody else is as crazy as that guy.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

384 / evaluation

Because communism is a science, it has to work. And if it doesn't work, there must be internal capitalist saboteurs who must be condemned, brought out and killed.

“Because communism is a science, it has to work. And if it doesn't work, there must be internal capitalist saboteurs who must be condemned, brought out and killed.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

385 / prediction

As it is today, although there’s lots of saber rattling by Lavrov and Putin, it doesn’t really look as though, I mean, yes, there might be a catastrophic disaster at the Zaporizhzhia nuclear plant, but it’s very unlikely for Putin actually to use tactical nuclear missiles in Ukraine, not least, of course, because the Chinese don’t want him to.

“As it is today, although there’s lots of saber rattling by Lavrov and Putin, it doesn’t really look as though, I mean, yes, there might be a catastrophic disaster at the Zaporizhzhia nuclear plant, but it’s very unlikely for Putin actually to use tactical nuclear missiles in Ukraine, not least, of course, because the Chinese don’t want him to.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

386 / belief

Your biography is the part of the canon. So in the person of Napoleon, I think startup founders see the best aspects of themselves resemblance.

“Your biography is the part of the canon. So in the person of Napoleon, I think startup founders see the best aspects of themselves resemblance.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

388 / uncertainty

Yeah, but it seems like I don’t know how you train a John von Neumann, because then he also assisted the US in thinking about Cold War game theory and things like that.

“Yeah, but it seems like I don’t know how you train a John von Neumann, because then he also assisted the US in thinking about Cold War game theory and things like that.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

389 / uncertainty

Another analogy here is it set up all these things that nobody understands, these systems of alliances and relationships in Europe. And you can imagine, like an AI advisor, and you’re not sure what’s going on, but it’s making, I don’t know, maybe it’s like a hedge fund manager that’s AI and it’s like making a lot of money.

“Another analogy here is it set up all these things that nobody understands, these systems of alliances and relationships in Europe. And you can imagine, like an AI advisor, and you’re not sure what’s going on, but it’s making, I don’t know, maybe it’s like a hedge fund manager that’s AI and it’s like making a lot of money.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

391 / evaluation

Correct me if this is wrong, but it sounds like you basically give it the opportunity to do a coup or make a bioweapon or whatever in testing in a situation where it thinks it’s the real world and you’re like, it didn’t do any of that.

“Correct me if this is wrong, but it sounds like you basically give it the opportunity to do a coup or make a bioweapon or whatever in testing in a situation where it thinks it’s the real world and you’re like, it didn’t do any of that.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

393 / belief

I'm just saying, okay, I personally, it's hard for me to imagine in 100 years that these things are still our slaves. And if they are, I think that's not the best world.

“I'm just saying, okay, I personally, it's hard for me to imagine in 100 years that these things are still our slaves. And if they are, I think that's not the best world.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

394 / uncertainty

Might be like Rlhf where we don’t know if it generalizes, but so far it makes your chat GPT thing better and you can also use it to make sure that chat GPT doesn’t tell you how to make a bioweapon.

“Might be like Rlhf where we don’t know if it generalizes, but so far it makes your chat GPT thing better and you can also use it to make sure that chat GPT doesn’t tell you how to make a bioweapon.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

397 / evaluation

It’s not in control of factories and robot armies or whatever. So in that case, even in training it will have those activations for being fucked up on because in the back of its mind it’s thinking I will take over once I have the opportunity.

“It’s not in control of factories and robot armies or whatever. So in that case, even in training it will have those activations for being fucked up on because in the back of its mind it’s thinking I will take over once I have the opportunity.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

398 / commitment

Speaking of the early years, it's really interesting that in 2011, you had a blog post where you said — “I’ve decided to once again leave my prediction for when human level AGI will arrive unchanged.

“Speaking of the early years, it's really interesting that in 2011, you had a blog post where you said — “I’ve decided to once again leave my prediction for when human level AGI will arrive unchanged.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

399 / prediction

What will it take to align human level and superhuman AIs? It's interesting because the sorts of reinforcement learning and self-play kinds of setups that are popular now, like Constitution AI or RLHF, DeepMind obviously has expertise in it for decades longer.

“What will it take to align human level and superhuman AIs? It's interesting because the sorts of reinforcement learning and self-play kinds of setups that are popular now, like Constitution AI or RLHF, DeepMind obviously has expertise in it for decades longer.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

401 / observation

Tell me why I should be less amazed by it or maybe put it in a different context but the reason I would be very impressed is… With chess, obviously this is not all the chess programs are doing, but there's a level of research you can do to narrow down the possibilities. And more importantly, in the math example, it seems that with some of the examples you've listed on your channel, the ability to solve the problem is so dependent on coming up with the right abstraction to think about it, coming up with ways of thinking about it that are not evident in the problem itself or in any other problem in any other test, that seems different from just a chess game where you don't have to think about what is the largest structure of this chess game in the same way as you do with the IMO problem.

“Tell me why I should be less amazed by it or maybe put it in a different context but the reason I would be very impressed is… With chess, obviously this is not all the chess programs are doing, but there's a level of research you can do to narrow down the possibilities. And more importantly, in the math example, it seems that with some of the examples you've listed on your channel, the ability to solve the problem is so dependent on coming up with the right abstraction to think about it, coming up with ways of thinking about it that are not evident in the problem itself or in any other problem in any other test, that seems different from just a chess game where you don't have to think about what is the largest structure of this chess game in the same way as you do with the IMO problem.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

402 / belief

I think that's a great note to leave it on Grant. Thanks so much for coming on the podcast and genuinely, you're one of the people I really, really admire but what you've done for the landscape of Math education is really remarkable.

“I think that's a great note to leave it on Grant. Thanks so much for coming on the podcast and genuinely, you're one of the people I really, really admire but what you've done for the landscape of Math education is really remarkable.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

403 / belief

Then should we think of educators more as motivational speakers? As in the actual job of getting the content in your head is maybe for the textbooks or for Youtube but why we have college classes or high school classes is that we have somebody who approximates Tony Robbins to get you to do the thing.

“Then should we think of educators more as motivational speakers? As in the actual job of getting the content in your head is maybe for the textbooks or for Youtube but why we have college classes or high school classes is that we have somebody who approximates Tony Robbins to get you to do the thing.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

406 / belief

Something I learned from your book that I thought was really interesting, and also tragic because of the counterfactual, was one of the strategies you suggested.

“Something I learned from your book that I thought was really interesting, and also tragic because of the counterfactual, was one of the strategies you suggested.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

408 / recommendation

And also, I've really enjoyed your book on Japan, The Japanese Empire. And you also have these other textbooks and collections of essays, which I highly recommend because of the thorough nature and the diversity of sources.

“And also, I've really enjoyed your book on Japan, The Japanese Empire. And you also have these other textbooks and collections of essays, which I highly recommend because of the thorough nature and the diversity of sources.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

409 / recommendation

I guess this maybe implies that if you do want to learn about a subject, it might just be helpful to just do an Intro to X subject course or textbook, not necessarily because it is instrumentally valuable to whatever problem you're interested in but because it'll give you the context by which to proceed on, the actual learning.

“I guess this maybe implies that if you do want to learn about a subject, it might just be helpful to just do an Intro to X subject course or textbook, not necessarily because it is instrumentally valuable to whatever problem you're interested in but because it'll give you the context by which to proceed on, the actual learning.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

412 / belief

For example, you had this really insightful post based on your experience in industry at Apple about the possibilities of Vision Pro and in what ways it's living up to and not. I think that would have just gone huge.

“For example, you had this really insightful post based on your experience in industry at Apple about the possibilities of Vision Pro and in what ways it's living up to and not. I think that would have just gone huge.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

414 / uncertainty

Presumably you didn't have some sort of practice but since you were encountering these things day to day, that natural frequency and way in which problems came up, did you have a worse understanding of those problems then compared to now, knowing what you do and having the practices you do, you're able to comprehend now? I don't know if that question made sense.

“Presumably you didn't have some sort of practice but since you were encountering these things day to day, that natural frequency and way in which problems came up, did you have a worse understanding of those problems then compared to now, knowing what you do and having the practices you do, you're able to comprehend now? I don't know if that question made sense.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

416 / uncertainty

If you are trying to self learn and there is a resource that is a close approximation of the syllabus you want. Should you just think “Hey, I don't know why I need this chapter.

“If you are trying to self learn and there is a resource that is a close approximation of the syllabus you want. Should you just think “Hey, I don't know why I need this chapter.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

417 / belief

If we nuked Afghanistan or Vietnam we would have technically won the war if that was the only goal, right? Oh, this is an interesting point that I think you made.

“If we nuked Afghanistan or Vietnam we would have technically won the war if that was the only goal, right? Oh, this is an interesting point that I think you made.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

418 / evaluation

If the current scale up doesn't work, all we're left with is just like the economy growing 2% a year, we have 2% a year more resources to spend on AI and at that scale you're talking about decades before just through sheer brute force you can train the 10 trillion dollar model or something.

“If the current scale up doesn't work, all we're left with is just like the economy growing 2% a year, we have 2% a year more resources to spend on AI and at that scale you're talking about decades before just through sheer brute force you can train the 10 trillion dollar model or something.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

420 / belief

I think some people might be skeptical that existing robots given their current hardware will have the dexterity and the maneuverability to do a lot of physical labor that an AI might want to do.

“I think some people might be skeptical that existing robots given their current hardware will have the dexterity and the maneuverability to do a lot of physical labor that an AI might want to do.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

422 / evaluation

Was there an opportunity after the end of World War II, before the Soviets developed the bomb, for the US to do something where either it somehow enforced a monopoly on having the bomb, or if that wasn't possible, make some sort of credible gesture that, we're eliminating this knowledge, you guys don't work on this, we're all just gonna step back from this.

“Was there an opportunity after the end of World War II, before the Soviets developed the bomb, for the US to do something where either it somehow enforced a monopoly on having the bomb, or if that wasn't possible, make some sort of credible gesture that, we're eliminating this knowledge, you guys don't work on this, we're all just gonna step back from this.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

425 / recommendation

The people who are working on AI right now are huge fans of yours. They're the ones who initially recommended the book to me because the way they see the progress in the field reminded them of this book.

“The people who are working on AI right now are huge fans of yours. They're the ones who initially recommended the book to me because the way they see the progress in the field reminded them of this book.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

426 / evaluation

I had a former guest, Richard Hanania, who has a book about foreign policy where he points out that our model of thinking about why countries do the things they do, especially in foreign affairs, is wrong because we think of them as individual rational actors, when in fact it's these competing factions within the government.

“I had a former guest, Richard Hanania, who has a book about foreign policy where he points out that our model of thinking about why countries do the things they do, especially in foreign affairs, is wrong because we think of them as individual rational actors, when in fact it's these competing factions within the government.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

427 / evaluation

You mentioned that a big reason why many of the scientists wanted to work on the bomb, especially the Jewish emigres, was because they're worried about Hitler getting it first.

“You mentioned that a big reason why many of the scientists wanted to work on the bomb, especially the Jewish emigres, was because they're worried about Hitler getting it first.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

431 / belief

I think that I was not convinced from the arguments that we could not have a system of sort of checks on this the same way you have checks on smart humans that it would try to deceive us to achieve its aims.

“I think that I was not convinced from the arguments that we could not have a system of sort of checks on this the same way you have checks on smart humans that it would try to deceive us to achieve its aims.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

432 / belief

First of all, one optimistic lesson to take from there is that we actually did learn from GPT-3, not everything, but we learned many things about what the potential failure modes could be 3.

“First of all, one optimistic lesson to take from there is that we actually did learn from GPT-3, not everything, but we learned many things about what the potential failure modes could be 3.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

436 / belief

I think what some people might not know is the millions and millions and millions of words of science fiction and fan fiction that you’ve written.

“I think what some people might not know is the millions and millions and millions of words of science fiction and fan fiction that you’ve written.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

437 / commitment

I think a better analogy is just put him in a high position in the Manhattan Project and say we will take your opinions very seriously and in fact, we even give you a lot of authority over this project.

“I think a better analogy is just put him in a high position in the Manhattan Project and say we will take your opinions very seriously and in fact, we even give you a lot of authority over this project.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

438 / belief

Eliezer Yudkowsky 3:19:29 I think I want to register for the record that the term breeding humans would cause me to look askance at any aliens who would propose that as a policy action on their part.

“Eliezer Yudkowsky 3:19:29 I think I want to register for the record that the term breeding humans would cause me to look askance at any aliens who would propose that as a policy action on their part.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

444 / uncertainty

From reading your writing from earlier, it seemed like a big part of your argument was like, look — I don’t know how many total mutations it was to get from chimps to humans, but it wasn’t that many mutations.

“From reading your writing from earlier, it seemed like a big part of your argument was like, look — I don’t know how many total mutations it was to get from chimps to humans, but it wasn’t that many mutations.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

445 / belief

I think the credit you would get for that, rightly, is as a good Agnostic forecaster, as somebody who is calm and measured. But it seems like to be able to make really strong claims about the future, about something that is so out of prior distributions as like the death of humanity, you don’t only have to show yourself as a good Agnostic forecaster, you have to show that your ability to forecast because of a particular theory is much greater.

“I think the credit you would get for that, rightly, is as a good Agnostic forecaster, as somebody who is calm and measured. But it seems like to be able to make really strong claims about the future, about something that is so out of prior distributions as like the death of humanity, you don’t only have to show yourself as a good Agnostic forecaster, you have to show that your ability to forecast because of a particular theory is much greater.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

446 / prediction

I expect them to be better than humans at science than they are at power seeking, because we had greater selection pressures for power seeking in our ancestral environment than we did for science.

“I expect them to be better than humans at science than they are at power seeking, because we had greater selection pressures for power seeking in our ancestral environment than we did for science.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

447 / uncertainty

I don’t know if you saw his recent blog post, but here’s a quote from it: “If you really accept the practical version of the Orthogonality Thesis, then it seems to me that you can’t regard education, knowledge, and enlightenment as instruments for moral betterment.

“I don’t know if you saw his recent blog post, but here’s a quote from it: “If you really accept the practical version of the Orthogonality Thesis, then it seems to me that you can’t regard education, knowledge, and enlightenment as instruments for moral betterment.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

448 / disagreement

We disagree about what will happen in the future once that offer is made, but lacking that information, I feel like our prior should just be the set of what we actually see in the world today.

“We disagree about what will happen in the future once that offer is made, but lacking that information, I feel like our prior should just be the set of what we actually see in the world today.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

453 / uncertainty

A lot of them might have, like, problems, to put it in that kind of language, but I don't know how many of them would make you suspect that there's, like, mental health issues or there's addiction issues that for somebody who's in charge of a multibillion dollar empire, I don't know.

“A lot of them might have, like, problems, to put it in that kind of language, but I don't know how many of them would make you suspect that there's, like, mental health issues or there's addiction issues that for somebody who's in charge of a multibillion dollar empire, I don't know.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

454 / uncertainty

I was like ten in 2011 so I don’t know if I would’ve personally. I would’ve liked to think I would’ve caught on if I was older but maybe not.

“I was like ten in 2011 so I don’t know if I would’ve personally. I would’ve liked to think I would’ve caught on if I was older but maybe not.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

455 / uncertainty

I don't know how to look into the far future situation, don't understand the far future situation, and don't see a path to doing good on that front I feel good about.

“I don't know how to look into the far future situation, don't understand the far future situation, and don't see a path to doing good on that front I feel good about.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

457 / belief

Maybe I'm a bit more pessimistic about that because I think the people who are working on individual rights frameworks weren’t anticipating an industrial revolution.

“Maybe I'm a bit more pessimistic about that because I think the people who are working on individual rights frameworks weren’t anticipating an industrial revolution.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

459 / prediction

I think my general take on this most important century stuff, and the reason it's so important is because it's easy to imagine a world that is really awesome and is free from scarcity and we see more of the progress we've seen over the last 200 years and we end up in a really great place.

“I think my general take on this most important century stuff, and the reason it's so important is because it's easy to imagine a world that is really awesome and is free from scarcity and we see more of the progress we've seen over the last 200 years and we end up in a really great place.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

460 / recommendation

Recommends The Power Broker.

“Yeah, I totally agree. I don't want to repeat myself, because I talked about this on my Byrne episode but we were talking about Robert Caro's books and one interesting thing is this guy was nominated for a Pulitzer Prize before he wrote The Power Broker. He was a top tier investigative journalist. Can you imagine crunching the numbers as a top tier investigative journalist at the peak of your career and you're like, “You know what would be a good use of my time? I'm going to spend the next seven years, in almost poverty, writing about this one guy who had a lot of influence in New York. I'm going to talk to any person who had conceivably even been in the same room as him or had been indirectly affected by his policies in any way. And I'm gonna do that obsessively for the next seven years.” There's no way the number crunching would get you there but it's probably been one of the most influential books in terms of how urban governance is done. Presidents have praised and read the book and said it changed how they think about politics. It is the kind of thing where you wouldn't have gone to that conclusion just from thinking about it beforehand and this is the most effective thing I could do.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

461 / belief

You mentioned that the aristocratic elites feel that they have the responsibility to give back more so than the meritocratic elites. But I believe that in the U.

“You mentioned that the aristocratic elites feel that they have the responsibility to give back more so than the meritocratic elites. But I believe that in the U.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

462 / uncertainty

First, how can somebody be long AI but hedge for the possibility that Taiwan will be invaded? So, I don't know if I should put money into TSMC but I know that GPUs are going to be the next big thing or are going to be very important in the future.

“First, how can somebody be long AI but hedge for the possibility that Taiwan will be invaded? So, I don't know if I should put money into TSMC but I know that GPUs are going to be the next big thing or are going to be very important in the future.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

463 / preference

co and the way I would describe Byrne is — every time I have a question about a concept or an event in finance, I google the name of that event or concept into Google and put in ‘Byrne Hobart’ at the end of that search query and 9 times out of 10, it's the best thing I've read about that topic.

“co and the way I would describe Byrne is — every time I have a question about a concept or an event in finance, I google the name of that event or concept into Google and put in ‘Byrne Hobart’ at the end of that search query and 9 times out of 10, it's the best thing I've read about that topic.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

464 / recommendation

There's many places where if you understand the economics of an issue he's talking about there's a lot to be left to Caro’s explanation but the actual breakdown of the personalities is just so fascinating and worth reading Caro for.

“There's many places where if you understand the economics of an issue he's talking about there's a lot to be left to Caro’s explanation but the actual breakdown of the personalities is just so fascinating and worth reading Caro for.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

465 / belief

We’re actually going to have sensible procurement policies that bring in things at a reasonable cost, and I think we need to balance a little bit back towards Robert Moses in order to have slightly more empowered builders who actually are able to deliver American cities the infrastructure they need at an affordable cost.

“We’re actually going to have sensible procurement policies that bring in things at a reasonable cost, and I think we need to balance a little bit back towards Robert Moses in order to have slightly more empowered builders who actually are able to deliver American cities the infrastructure they need at an affordable cost.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

466 / evaluation

I push back against the universal basic income advocates who I think are basically engaging in a materialist fallacy of thinking that a human being’s life is shaped by their take home pay or their unearned pay.

“I push back against the universal basic income advocates who I think are basically engaging in a materialist fallacy of thinking that a human being’s life is shaped by their take home pay or their unearned pay.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

467 / evaluation

So the US has historically been better at being pro-business than, let’s say, the Northern European social democracies, but the Northern European social democracies are great on the education front. So places like Sweden and the Netherlands, and Germany are also very successful places because they have enough education to counter the fact that they may not necessarily be as pro-business as the US is.

“So the US has historically been better at being pro-business than, let’s say, the Northern European social democracies, but the Northern European social democracies are great on the education front. So places like Sweden and the Netherlands, and Germany are also very successful places because they have enough education to counter the fact that they may not necessarily be as pro-business as the US is.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

469 / commitment

Lehman Center for American History and the Jacques Barzun Professor Emeritus of History at Columbia University, where he has also shared the Department of History. We will be discussing Robert Moses.

“Lehman Center for American History and the Jacques Barzun Professor Emeritus of History at Columbia University, where he has also shared the Department of History. We will be discussing Robert Moses.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

471 / uncertainty

If there's one person who's designing all the bridges, all the highways, all the parks, is something more possible that can be possible if like multiple different branches and people have their own unique visions? I don't know if that question makes sense.

“If there's one person who's designing all the bridges, all the highways, all the parks, is something more possible that can be possible if like multiple different branches and people have their own unique visions? I don't know if that question makes sense.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

472 / evaluation

I guess if we take this kind of view and just care about progeny more, there's a part in the Power Broker where they're talking about the Cross Bronx Expressway through East Tremont and Moses is getting opposition from an elected official.

“I guess if we take this kind of view and just care about progeny more, there's a part in the Power Broker where they're talking about the Cross Bronx Expressway through East Tremont and Moses is getting opposition from an elected official.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

473 / evaluation

One of the main criticisms that Caro makes is that Moses refused to add mass transit to his highways, which would have helped deal with the traffic problem and the car problem and all these other problems at a time when getting the right of way and doing the construction would have been much cheaper. He just refused to do that because of his dislike for mass transit.

“One of the main criticisms that Caro makes is that Moses refused to add mass transit to his highways, which would have helped deal with the traffic problem and the car problem and all these other problems at a time when getting the right of way and doing the construction would have been much cheaper. He just refused to do that because of his dislike for mass transit.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

474 / evaluation

There's a very arresting anecdote in the Power Broker where I think he's 71 and his daughter gets cancer and for the first time, he had to accept a salary for working on the World's Fair because he didn't have enough.

“There's a very arresting anecdote in the Power Broker where I think he's 71 and his daughter gets cancer and for the first time, he had to accept a salary for working on the World's Fair because he didn't have enough.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

475 / recommendation

Today, I have the pleasure of speaking with Brian Potter, who is an engineer and the author of the excellent Construction Physics blog, where he writes about how the construction industry works and why it has been slow to industrialize and innovate. It's one of my favorite blogs on the internet, I highly, highly recommend that people check it out.

“Today, I have the pleasure of speaking with Brian Potter, who is an engineer and the author of the excellent Construction Physics blog, where he writes about how the construction industry works and why it has been slow to industrialize and innovate. It's one of my favorite blogs on the internet, I highly, highly recommend that people check it out.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

476 / uncertainty

I don't know if you're familiar with the longtermist movement, maybe you’ve come across this before, but one thing they've proposed is not like a doctrine or anything, but on the periphery, some idea I saw was that just as we have environmental review we should have a posterity review so that you're analyzing the impacts of your actions on generations way down the line.

“I don't know if you're familiar with the longtermist movement, maybe you’ve come across this before, but one thing they've proposed is not like a doctrine or anything, but on the periphery, some idea I saw was that just as we have environmental review we should have a posterity review so that you're analyzing the impacts of your actions on generations way down the line.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

477 / prediction

There's a famous law in software that says that a project will take longer than you expect even after you recount for the fact that it will take longer than you expect.

“There's a famous law in software that says that a project will take longer than you expect even after you recount for the fact that it will take longer than you expect.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

480 / uncertainty

I don't know if you saw this, by the way, but about a year or two ago, Microsoft announced that they were building this new office for developers in India and they were looking at this very interesting architectural details and craftsmanship that made it look like you know for everything from the furniture to the floor layout to the arches of the entrances.

“I don't know if you saw this, by the way, but about a year or two ago, Microsoft announced that they were building this new office for developers in India and they were looking at this very interesting architectural details and craftsmanship that made it look like you know for everything from the furniture to the floor layout to the arches of the entrances.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

481 / belief

I think there are a lot of examples where if you don't have the context on why they were built a certain way, you wouldn't understand what was going on.

“I think there are a lot of examples where if you don't have the context on why they were built a certain way, you wouldn't understand what was going on.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

482 / belief

I think a co-sign solution to this kind of thing would be optimal where if the view is worth more to you than the apartment is worth to somebody else, then you can just pay them to not build there.

“I think a co-sign solution to this kind of thing would be optimal where if the view is worth more to you than the apartment is worth to somebody else, then you can just pay them to not build there.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

483 / prediction

There's been a lot of talk about a potential real estate bubble in China because they're building housing in places where people don't really need it.

“There's been a lot of talk about a potential real estate bubble in China because they're building housing in places where people don't really need it.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

487 / uncertainty

I don’t know, but maybe yes. More generally, one of the common arguments that libertarians make about India and its elites is, “Oh, all of India's elites go study in Oxford or something, and they learn about the regulations the West is adopting that make no sense for a country with $2,000 GDP per capita.

“I don’t know, but maybe yes. More generally, one of the common arguments that libertarians make about India and its elites is, “Oh, all of India's elites go study in Oxford or something, and they learn about the regulations the West is adopting that make no sense for a country with $2,000 GDP per capita.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

489 / evaluation

All of these things combined make women's lives worse on average than men's lives. It's not because society mistreats them, but in some sense, there's still unfairness geared toward women.

“All of these things combined make women's lives worse on average than men's lives. It's not because society mistreats them, but in some sense, there's still unfairness geared toward women.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

491 / evaluation

I mean, I think of you as an optimist––at least by temperament. But this seems like one of the more pessimistic things I've heard overall, anywhere, because of the idea that not only will human civilization be decimated almost surely, but that they will never be able to recover.

“I mean, I think of you as an optimist––at least by temperament. But this seems like one of the more pessimistic things I've heard overall, anywhere, because of the idea that not only will human civilization be decimated almost surely, but that they will never be able to recover.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

497 / prediction

I want to ask you again about contingency because there are so many other examples where things you thought would be universal actually don't turn out to be.

“I want to ask you again about contingency because there are so many other examples where things you thought would be universal actually don't turn out to be.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

499 / uncertainty

I wonder whether he was doing this in the 40’–– like when he was at that age, was he doing this? I don't know what the cultural conventions were at the time.

“I wonder whether he was doing this in the 40’–– like when he was at that age, was he doing this? I don't know what the cultural conventions were at the time.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast

500 / disagreement

If it's true, as you say, in the book, that moral values are very contingent, then shouldn't that make us suspect that modern Western values aren't that good? They're mediocre, or worse, because ex ante, you would expect to end up with a median of all the values we could have had at this point.

“If it's true, as you say, in the book, that moral values are very contingent, then shouldn't that make us suspect that modern Western values aren't that good? They're mediocre, or worse, because ex ante, you would expect to end up with a median of all the values we could have had at this point.”
Speaker
Dwarkesh Patel
Publisher
Dwarkesh Podcast
Search evidence