High Signal Podcasts Evidence ledger
Method
Browse
How the evidence works

Public evidence record

Dario Amodei

CEO · Anthropic

Claims
70
Episodes
2
Shows
2
Named items
0

Claim ledger

What Dario said.

70 transcript-backed records

01 / prediction

On the basic hypothesis of, as you put it, within ten years we’ll get to what I call a “country of geniuses in a data center”, I’m at 90% on that. It’s hard to go much higher than 90% because the world is so unpredictable.

“On the basic hypothesis of, as you put it, within ten years we’ll get to what I call a “country of geniuses in a data center”, I’m at 90% on that. It’s hard to go much higher than 90% because the world is so unpredictable.”
Speaker
Dario Amodei
Publisher
Dwarkesh Podcast

03 / evaluation

I agree that that doesn’t solve the problem in the long run, particularly if the ability of AI models to make other AI models proliferates, then the whole thing can become harder to solve.

“I agree that that doesn’t solve the problem in the long run, particularly if the ability of AI models to make other AI models proliferates, then the whole thing can become harder to solve.”
Speaker
Dario Amodei
Publisher
Dwarkesh Podcast

08 / observation

The point is to get a reputation of telling the company the truth about what’s happening, to call things what they are, to acknowledge problems, to avoid the sort of corpo speak, the kind of defensive communication that often is necessary in public because the world is very large and full of people who are interpreting things in bad faith.

“The point is to get a reputation of telling the company the truth about what’s happening, to call things what they are, to acknowledge problems, to avoid the sort of corpo speak, the kind of defensive communication that often is necessary in public because the world is very large and full of people who are interpreting things in bad faith.”
Speaker
Dario Amodei
Publisher
Dwarkesh Podcast

10 / belief

I would bet it stays pretty fast even as the scale goes to the entire economy. So I think we should be thinking about this middle world where things are extremely fast, but not instant, where they take time because of economic diffusion, because of the need to close the loop.

“I would bet it stays pretty fast even as the scale goes to the entire economy. So I think we should be thinking about this middle world where things are extremely fast, but not instant, where they take time because of economic diffusion, because of the need to close the loop.”
Speaker
Dario Amodei
Publisher
Dwarkesh Podcast

11 / uncertainty

I don’t know what the critical moment is or if there’s a single critical moment. But I think there will be either a critical moment, a small number of critical moments, or some critical window where AI confers some large advantage from the perspective of national security, and one country or coalition has reached it before others.

“I don’t know what the critical moment is or if there’s a single critical moment. But I think there will be either a critical moment, a small number of critical moments, or some critical window where AI confers some large advantage from the perspective of national security, and one country or coalition has reached it before others.”
Speaker
Dario Amodei
Publisher
Dwarkesh Podcast

13 / observation

The problem is you have this hellish demand prediction problem when you’re buying the next year of compute and you might guess under and be very profitable but have no compute for research.

“The problem is you have this hellish demand prediction problem when you’re buying the next year of compute and you might guess under and be very profitable but have no compute for research.”
Speaker
Dario Amodei
Publisher
Dwarkesh Podcast

15 / belief

We’re under an incredible amount of commercial pressure and make it even harder for ourselves because we have all this safety stuff we do that I think we do more than other companies.

“We’re under an incredible amount of commercial pressure and make it even harder for ourselves because we have all this safety stuff we do that I think we do more than other companies.”
Speaker
Dario Amodei
Publisher
Dwarkesh Podcast

17 / belief

Given the serious dangers that I lay out in “Adolescence of Technology” around things like biological weapons and bioterrorism autonomy risk, and the timelines we’ve been talking about—10 years is an eternity—I think that’s a crazy thing to do.

“Given the serious dangers that I lay out in “Adolescence of Technology” around things like biological weapons and bioterrorism autonomy risk, and the timelines we’ve been talking about—10 years is an eternity—I think that’s a crazy thing to do.”
Speaker
Dario Amodei
Publisher
Dwarkesh Podcast

21 / prediction

If every year we predict exactly what the demand is going to be, we’ll be profitable every year. Because spending 50% of your compute on research, roughly, plus a gross margin that’s higher than 50% and correct demand prediction leads to profit.

“If every year we predict exactly what the demand is going to be, we’ll be profitable every year. Because spending 50% of your compute on research, roughly, plus a gross margin that’s higher than 50% and correct demand prediction leads to profit.”
Speaker
Dario Amodei
Publisher
Dwarkesh Podcast

24 / uncertainty

Someone’s going to come in, someone’s going to edit it. They’re going to be like, “Oh, I don’t know, Dario scratched his head and we could edit that out.

“Someone’s going to come in, someone’s going to edit it. They’re going to be like, “Oh, I don’t know, Dario scratched his head and we could edit that out.”
Speaker
Dario Amodei
Publisher
Dwarkesh Podcast

25 / recommendation

Then, if these risks emerge when we’re more certain of them—which I think we might be as soon as later this year—then I think we need to act very fast in the areas where we’ve actually seen the risk. I think the only way to do this is to be nimble.

“Then, if these risks emerge when we’re more certain of them—which I think we might be as soon as later this year—then I think we need to act very fast in the areas where we’ve actually seen the risk. I think the only way to do this is to be nimble.”
Speaker
Dario Amodei
Publisher
Dwarkesh Podcast

29 / prediction

One way I think about it is if the technology is advancing quickly, if it’s advancing exponentially, what that means is there’s always a surface area of new use cases that have been developed in the last three months.

“One way I think about it is if the technology is advancing quickly, if it’s advancing exponentially, what that means is there’s always a surface area of new use cases that have been developed in the last three months.”
Speaker
Dario Amodei
Publisher
Dwarkesh Podcast

30 / evaluation

I think there’s something going on where pre-training is not like the process of humans learning, but it’s somewhere between the process of humans learning and the process of human evolution.

“I think there’s something going on where pre-training is not like the process of humans learning, but it’s somewhere between the process of humans learning and the process of human evolution.”
Speaker
Dario Amodei
Publisher
Dwarkesh Podcast

31 / prediction

There’s a question of how much of that goes to the pharmaceutical company or the AI company, but there’s an enormous consumer surplus because —assuming we can get access for everyone, which I care about greatly—we cure all of these diseases.

“There’s a question of how much of that goes to the pharmaceutical company or the AI company, but there’s an enormous consumer surplus because —assuming we can get access for everyone, which I care about greatly—we cure all of these diseases.”
Speaker
Dario Amodei
Publisher
Dwarkesh Podcast

32 / prediction

So a few things, now we’re talking about hitting the limit before we get to the level of humans and the skill of humans. So I think one that’s popular today, and I think could be a limit that we run into, like most of the limits, I would bet against it, but it’s definitely possible, is we simply run out of data.

“So a few things, now we’re talking about hitting the limit before we get to the level of humans and the skill of humans. So I think one that’s popular today, and I think could be a limit that we run into, like most of the limits, I would bet against it, but it’s definitely possible, is we simply run out of data.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

33 / belief

I think generally the dynamic was I got this kind of inspiration from Ilya and from others, folks like Alec Radford, who did the original GPT-1 and then ran really hard with it, me and my collaborators, on GPT-2, GPT-3, RL from Human Feedback, which was an attempt to kind of deal with the early safety and durability, things like debate and amplification, heavy on interpretability.

“I think generally the dynamic was I got this kind of inspiration from Ilya and from others, folks like Alec Radford, who did the original GPT-1 and then ran really hard with it, me and my collaborators, on GPT-2, GPT-3, RL from Human Feedback, which was an attempt to kind of deal with the early safety and durability, things like debate and amplification, heavy on interpretability.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

34 / uncertainty

One, I realized it would be very long, and two, I’m very aware of and very much tried to avoid just being, I don’t know what the term for it is, but one of these people who’s overconfident and has an opinion on everything and says a bunch of stuff and isn’t an expert, I very much tried to avoid that.

“One, I realized it would be very long, and two, I’m very aware of and very much tried to avoid just being, I don’t know what the term for it is, but one of these people who’s overconfident and has an opinion on everything and says a bunch of stuff and isn’t an expert, I very much tried to avoid that.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

38 / belief

I think no matter how smart you are, people talk about, “Oh, we can make models of biological systems that’ll do everything the biological systems … ” Look, I think computational modeling can do a lot.

“I think no matter how smart you are, people talk about, “Oh, we can make models of biological systems that’ll do everything the biological systems … ” Look, I think computational modeling can do a lot.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

39 / belief

” Then I think at some point it’ll flip around where the AI systems will be the PIs, will be the leaders, and they’ll be ordering humans or other AI systems around.

“” Then I think at some point it’ll flip around where the AI systems will be the PIs, will be the leaders, and they’ll be ordering humans or other AI systems around.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

41 / belief

I mean, I think we’re still early in terms of our ability to see things, but I’ve been surprised at how much we’ve been able to look inside these systems and understand what we see.

“I mean, I think we’re still early in terms of our ability to see things, but I’ve been surprised at how much we’ve been able to look inside these systems and understand what we see.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

42 / belief

Currently, we’re not trying to make such IDEs ourself, rather we’re powering the companies like Cursor or Kognition or some of the other expo in the security space, others that I could mention as well that are building such things themselves on top of our API and our view has been let 1,000 flowers bloom.

“Currently, we’re not trying to make such IDEs ourself, rather we’re powering the companies like Cursor or Kognition or some of the other expo in the security space, others that I could mention as well that are building such things themselves on top of our API and our view has been let 1,000 flowers bloom.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

45 / belief

I also think it’s going to be five or 10 years more than it’s going to be five or 10 hours, because I’ve just seen how human systems work. And I think a lot of these people who write down these differential equations, who say AI is going to make more powerful AI, who can’t understand how it could possibly be the case that these things won’t change so fast.

“I also think it’s going to be five or 10 years more than it’s going to be five or 10 hours, because I’ve just seen how human systems work. And I think a lot of these people who write down these differential equations, who say AI is going to make more powerful AI, who can’t understand how it could possibly be the case that these things won’t change so fast.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

47 / belief

I think by the time, certainly within two to three years, whether we have these super powerful AIs or not, clusters are going to get to the size where you’ll be able to deploy millions of these.

“I think by the time, certainly within two to three years, whether we have these super powerful AIs or not, clusters are going to get to the size where you’ll be able to deploy millions of these.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

49 / belief

I think we look at some of the benchmarks where previous models were like, “Oh, could do it 6% of the time,” and now our model would do it 14 or 22% of the time.

“I think we look at some of the benchmarks where previous models were like, “Oh, could do it 6% of the time,” and now our model would do it 14 or 22% of the time.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

50 / uncertainty

If you say, “Well, I don’t know, we’re starting to get to PhD level, and last year we were at undergraduate level, and the year before we were at the level of a high school student,” again, you can quibble with what tasks and for what.

“If you say, “Well, I don’t know, we’re starting to get to PhD level, and last year we were at undergraduate level, and the year before we were at the level of a high school student,” again, you can quibble with what tasks and for what.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

51 / belief

I think that’s probably a bad idea during training because the model can be changing its policy, it can be changing what it’s doing and it’s having an effect in the real world.

“I think that’s probably a bad idea during training because the model can be changing its policy, it can be changing what it’s doing and it’s having an effect in the real world.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

53 / belief

There was Rich Sutton’s bitter lesson, Gwern wrote about the scaling hypothesis. But I think somewhere between 2014 and 2017 was when it really clicked for me, when I really got conviction that, “Hey, we’re going to be able to these incredibly wide cognitive tasks if we just scale up the models.

“There was Rich Sutton’s bitter lesson, Gwern wrote about the scaling hypothesis. But I think somewhere between 2014 and 2017 was when it really clicked for me, when I really got conviction that, “Hey, we’re going to be able to these incredibly wide cognitive tasks if we just scale up the models.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

55 / uncertainty

We’ll be at least 90%. So again, I would guess, I don’t know how long it’ll take, but I would guess again, 2026, 2027 Twitter people who crop out these numbers and get rid of the caveats, I don’t know.

“We’ll be at least 90%. So again, I would guess, I don’t know how long it’ll take, but I would guess again, 2026, 2027 Twitter people who crop out these numbers and get rid of the caveats, I don’t know.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

56 / commitment

I won’t give specific examples, but it’s been hard to get people to adopt even the technologies that we’ve developed, even ones where the case for their efficacy is very, very strong.

“I won’t give specific examples, but it’s been hard to get people to adopt even the technologies that we’ve developed, even ones where the case for their efficacy is very, very strong.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

58 / belief

I think all these questions of drama are profoundly uninteresting and the thing that matters is the ecosystem that we all operate in and how to make that ecosystem better because that constrains all the players.

“I think all these questions of drama are profoundly uninteresting and the thing that matters is the ecosystem that we all operate in and how to make that ecosystem better because that constrains all the players.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

59 / belief

Around 2016 or 2017, I first started to really believe in or at least confirm my belief in the scaling hypothesis when Ilya famously said to me, “The thing you need to understand about these models is they just want to learn.

“Around 2016 or 2017, I first started to really believe in or at least confirm my belief in the scaling hypothesis when Ilya famously said to me, “The thing you need to understand about these models is they just want to learn.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

61 / prediction

That’s my vision of the powerful AI system. But I think for much longer than we might expect, we will see that small parts of the job that humans still do will expand to fill their entire job in order for the overall productivity to go up.

“That’s my vision of the powerful AI system. But I think for much longer than we might expect, we will see that small parts of the job that humans still do will expand to fill their entire job in order for the overall productivity to go up.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

62 / evaluation

One of the properties of the RSP is that we don’t specify ASL-4 until we’ve hit ASL-3. And I think that’s proven to be a wise decision because even with ASL-3, again, it’s hard to know this stuff in detail, and we want to take as much time as we can possibly take to get these things right.

“One of the properties of the RSP is that we don’t specify ASL-4 until we’ve hit ASL-3. And I think that’s proven to be a wise decision because even with ASL-3, again, it’s hard to know this stuff in detail, and we want to take as much time as we can possibly take to get these things right.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

63 / uncertainty

” I’m going to have to write a whole other essay about that. But meaning is actually interesting because you think about the life that someone lives or something, or let’s say you were to put me in, I don’t know, like a simulated environment or something where I have a job and I’m trying to accomplish things and I don’t know, I do that for 60 years and then you’re like, “Oh, oops, this was actually all a game,” right?

“” I’m going to have to write a whole other essay about that. But meaning is actually interesting because you think about the life that someone lives or something, or let’s say you were to put me in, I don’t know, like a simulated environment or something where I have a job and I’m trying to accomplish things and I don’t know, I do that for 60 years and then you’re like, “Oh, oops, this was actually all a game,” right?”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

64 / recommendation

That’s where we should start. And can we increase the success rate of clinical trials by doing things in animal trials that we used to do in clinical trials and doing things in simulations that we used to do in animal trials?

“That’s where we should start. And can we increase the success rate of clinical trials by doing things in animal trials that we used to do in clinical trials and doing things in simulations that we used to do in animal trials?”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

65 / prediction

We came out with a new one just a few weeks ago and probably going forward, we might release new ones multiple times a year because it’s hard to get these policies right technically, organizationally from a research perspective.

“We came out with a new one just a few weeks ago and probably going forward, we might release new ones multiple times a year because it’s hard to get these policies right technically, organizationally from a research perspective.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

66 / observation

” I think we even had a “certainly” eval because again, at one point, the model had this problem where it had this annoying tick where it would respond to a wide range of questions by saying, “Certainly, I can help you with that.

“” I think we even had a “certainly” eval because again, at one point, the model had this problem where it had this annoying tick where it would respond to a wide range of questions by saying, “Certainly, I can help you with that.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

67 / evaluation

One way I think of it is the thing we’re all afraid of is the race to the bottom and the race to the bottom doesn’t matter who wins because we all lose.

“One way I think of it is the thing we’re all afraid of is the race to the bottom and the race to the bottom doesn’t matter who wins because we all lose.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

68 / prediction

We’ve seen similar things in graduate-level math, physics, and biology from models like OpenAi’s o1. So if we just continue to extrapolate this in terms of skill that we have, I think if we extrapolate the straight curve, within a few years, we will get to these models being above the highest professional level in terms of humans.

“We’ve seen similar things in graduate-level math, physics, and biology from models like OpenAi’s o1. So if we just continue to extrapolate this in terms of skill that we have, I think if we extrapolate the straight curve, within a few years, we will get to these models being above the highest professional level in terms of humans.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

69 / commitment

I like to think that Anthropic will, we do everything we can that we will, our RSP is checked by our long-term benefit trust, so we do everything we can to adhere to our own RSP.

“I like to think that Anthropic will, we do everything we can that we will, our RSP is checked by our long-term benefit trust, so we do everything we can to adhere to our own RSP.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast

70 / evaluation

I will say, while in theory there’s nothing you could do there that you couldn’t have done through just giving the model the API to drive the computer screen, this really lowers the barrier.

“I will say, while in theory there’s nothing you could do there that you couldn’t have done through just giving the model the API to drive the computer screen, this really lowers the barrier.”
Speaker
Dario Amodei
Publisher
Lex Fridman Podcast
Search evidence