Dario & Daniela Amodei：超智能AI的崛起、人类控制权的终结与人类的命运

The Builders is a production of Kaleidoscope and iHeart, presented by Ramp. I’m Walter Isaacson, and welcome to The Builders, my new series of interviews featuring the most consequential minds shaping our future. A person’s biography tells you what values, what drives, sometimes what demons are bred into them. It’s particularly interesting in an

era of AI.

Because we’re watching the biographies of AI systems happen, we have to make sure that we raise them carefully so they don’t reflect the dark sides of humanity.

Alarming new details about robots from OpenAI hacking into another company. Anthropic, the maker of Claude, says it blocked several potential plots this year by scientists who were using its AI models to conduct

research that could have led to the creation of a bioweapon.

And so I’ve begun to feel that the moral compass of these builders is the most important thing you can know about them, to understand what values we’re imbuing into these artificial agents. So today, I’m speaking with Dario and Daniela Amodei.

CEO and co-founder of AI giant Anthropic, Dario Amodei, with an urgent warning about artificial intelligence.

Dario and Daniela are the brother-and-sister duo who run Anthropic, the maker of Claude.

And for a while, Anthropic

has been seen as the good guys in the AI race. But in the past few months, we’ve seen worrying developments.

Anthropic researcher Jacob Coxon resigned from his post, sounding the alarm about what he called out of control. In a now-viral post, Coxon said the companies are, quote, “gambling with our lives.” The people building AI earnestly believe that it could kill us all by the end of the decade.

All of this, along with reports of a possible IPO, has begun to make me wonder: How deep does Anthropic’s commitment to humanity really run? And what happens to those principles

when they get in the way of winning?

Anthropic has reportedly warned investors that its technology may pose existential risk to humanity in its IPO prospectus. And you might then wonder, with a company warning of this, why it’s still planning to go ahead with the

IPO at all.

The public sale could value Anthropic at $2 trillion.

I chatted with Daniela and Dario in July 2026. At the time, I didn’t know just how prescient this conversation would be. Here’s my interview with Dario and Daniela Amodei.

Thank you so much for having us.

1. Science Meets Humanism

Steve Jobs, whenever he launched a product, always had that street sign of the humanities and technology standing at that intersection. And when you were growing up, Daniela, you were a flautist. You loved the arts. And Dario, you were always a scientist. Did that help you gel? And do you think that combination is important?

Yeah, I think that combination is a lot of what’s behind the company. One theory I have is that running a company is too big a task for any one person. The range of skills it involves isn’t possible for any single person to have. So you have to assemble a full complement of skills across the top, like 10 or 15 people

in the company or something like that.

And every very successful company splits it up in different ways and solves the problem in different ways. And I think with the science and humanities thing, Daniela and I kind of embody that and split that. Yeah.

I would say it’s interesting: There was an off-site we did in maybe 2023 where we were talking about what it feels like to work at Anthropic and how we embody our culture. We actually said it’s an unusual blend of science and humanism. It was only later that I thought, “That’s also how you would describe me and Dario.”

When you were kids, did you know you wanted to work together?

Yes.

Yes. I don’t think either of us imagined that we would start a company.

That’s not really how either of us thought of our lives or what we wanted to do. I had always wanted to be a scientist. Daniela can speak to her aspirations. But I found companies and business to be of very little interest to me.

Dario and I were very unusual in that, from a really early age, we genuinely really liked each other. And I agree with Dario. I don’t think starting a company was ever on our bingo card. Neither of us was particularly company-oriented.

I was always really

interested in—I think I would call myself a classic do-gooder. I worked in international development. I felt strongly about helping people everywhere in the world have equal access and opportunity. And I became a vegetarian when I was quite young. It was the only thing that I did first. Dario did everything first because he’s older, but I became a vegetarian and then he became a vegetarian.

Dario, tell me how you and Daniela divide up the roles, and then how you fill them in with others on the team.

I think I function best when I think about the big-picture strategy and the research side of things. I’m best at laying out a vision and seeing what comes next. I’ve found that when I’m freed up to focus entirely on that, it’s the thing that has the most value for the company.

And then running the company day to

day, recruiting people, and making sure the wheels don’t fall off the machine while I’m trying to figure out what comes next.

Daniela does this.

I think this division of labor allows both things to be done to a much higher standard than either would be if we were distracted by having to do the other thing.

Our chief people officer, who knows both of us really well and is in a lot of conversations with us, put it this way: Dario makes us ambitious, and I make us successful. I thought that was a really good encapsulation. Anthropic could be a good, successful business, but not change the world if we didn’t have Dario. And with Dario, we would have the perfect vision, but our ability to execute would be more complicated. There’s something special about the fact that we know each other really well, and that our skills are close to 100% complementary in this fashion. And that’s not even to mention the fact that we have five co-founders along with the two of us who own incredibly important parts of the business, the culture, and the organization. It’s hard for me to imagine how you would do this with just the two of us.

Go back to growing up. You had these extraordinary parents, and they helped teach you right from wrong. How did they do that?

I don’t know that it was ever something that was explicitly taught. Sometimes I hear people say, “We covered different forms of philosophical ethics at the dinner table.”

That was not how it was done.

My mom does have a philosophy degree, but she didn’t subject us to talking about Kant or anything like that. I think it was more inherent in the values: the belief that people everywhere in the world are deserving of opportunity, dignity, rights, and respect. There was something special about the feeling that we could be whoever we wanted to be, as long as there was this faith and belief that part of what it means to be a successful human is to be a good person and to be kind to other people.

Yeah, it felt implicit. There wasn’t any kind of “This is what you should do” or “This is how you live in the world.”

2. Teaching Claude Right From Wrong

One reason I ask, as you might suspect, is that now you’ve got a really important kid named Claude. To what extent does the way you grew up help inform how you teach Claude to know right from wrong?

We put out a constitution for Claude a few months ago. There are some interesting lessons in it that we’ve learned from several years of trying to train Claude in successive versions.

Early on, we had imagined that

there would be various rules that you would tell Claude: “Don’t help someone commit a crime” or “Don’t make biological weapons,” or whatever.

But it turns out that when you make this kind

of discrete list of rules, you end up with this spiky shape where you’re trying to teach the model 170 different rules that are kind of arbitrary. We found it’s better to teach the model some basic principles, like what it should prioritize and how it should reason about what the right thing to do is.

Of course, I think that’s the right way to teach people as well, right?

If I just give someone a list of rules—“Thou shalt do this, thou shalt not do this”—I don’t think that produces as robust a morality as saying, “This is how you should think about things. You should learn how to think for yourself, and these are the principles on which you should think.” Principles of open inquiry, care for other people, fairness, and justice.

And then everything is derived from that. And that is, frankly, how Claude learns.

We find that surprisingly effective because it allows Claude to generalize to new situations it might never have imagined.

Daniela, I can’t help myself when you said your mom was an expert on Kant.

Did you actually wrestle with philosophy?

You know, what’s interesting is that the woman responsible for training Claude’s character has a PhD in philosophy.

It’s unusual, right? You would think, “Oh, the person…”

…that’s training the character of these models must be a computer scientist or somebody who’s studied machine learning or AI for many decades. Amanda’s technical, and she technically trains the model, but I think it’s very felt in Claude’s character when you communicate with it: there’s this experience of something that’s very wise, kind of the better angel of your nature trying to answer your questions.

But if you poke Claude in difficult areas where different philosophical traditions might conflict, I think that’s where you especially feel Amanda. You’ll see, “Oh, well, you might want to consider things from this point of view, but you could also consider it from this point of view.” I do think it’s important because, as the world evolves and begins to rely on these models more and more, you don’t want to be locked into one philosophical tradition. Ethics and morality are quite complicated, and different people and cultures might have different answers about how they would like you to behave.

3. GPT-2 Changes Everything

Dario, you were there at one of the seminal moments of modern times, when GPT-1 and GPT-2 suddenly burst onto the scene, surprising—perhaps even surprising you. You were head of research at OpenAI. What was that moment like?

I wasn’t very involved in GPT-1, but I was very involved in the idea that we should run a larger-scale, more expensive, more ambitious experiment on the idea that, if we ran these language models at a larger scale, they would be better at language and more coherent.

So it’s almost like one of these moments in scientific…

…history where you propose a definitive experiment to make a very strange prediction. If the prediction turns out to be correct, you’re like, “God, we’re really onto something.” I see GPT-2 as the first real test of the scaling laws hypothesis. What I found super interesting was the enormous divergence in reactions to it.

And were you kind of surprised?

I personally was super freaked out by it.

I looked at it and said, “My God, you just scale this thing up another factor of a million and it’ll be better than humans at everything. This is the defining economic and scientific moment of the modern age. We just don’t know it yet.” This is going to determine the fate of nations and economies and the future of humanity. You peer into the crystal ball, you see the whole future unfurled, and it really scared me. I said, “Oh my God, what are the implications of that? Is this the right place to be developing it? What’s going to happen next?”

What do we do with this? Who do we tell about this?

Maybe 10% of the people around me had exactly the same reaction.

And it was often the smartest ones.

90% of the people around me—because GPT-2’s capabilities in any particular area were not that impressive, right? The difference between GPT-1 and GPT-2 was the generality: the fact that it could translate languages, summarize a document, and write a few lines of code here and there. People would look at it and say that, in any particular area, it didn’t seem so smart.

It was the…

…difference between GPT-1 and GPT-2. So many people just shrugged and dismissed it. You get the usual comments from the jaded people.

“Ah, tech people have done this before.”

“Here are all the limitations of the models.” I look at it and say it’s the same thing over and over again: the limitations of the models now, six months from now, the things that seemed hard are going to be easy. Twelve months from now, the things that seemed impossible, the models are going to be doing them regularly. Everything we saw with GPT-2, we’re seeing it over and over again in wider and wider swaths of the world, and the world is gradually, finally being convinced. We just announced that we’re going to be developing drugs for neglected rare diseases.

I think you were quite polite about it, but I remember that, at the time, some very well-respected computer scientists and AI and ML leaders basically bullied some of the researchers who worked on it on Twitter. They were like, “This is stupid. This model is so incapable.” It was like you were sort of a quack if you believed this was a really significant moment.

And that’s all been memory-holed.

All the researchers who were like, “Oh, this is pseudoscience. It doesn’t make sense”—all of them are applying for jobs at the LLM companies. We never say it out loud to any particular one, but we look at them and think, “We remember what you were saying in 2019. You have no credibility on this matter.”

So, Daniela, you were vice president of safety back then at OpenAI. When you hit GPT-2, I assume you were like Dario, somebody who was almost freaked out about it.

Did it freak you out about safety, too?

I don’t think there was something about GPT-2, the model itself, that made us concerned. We didn’t look at that model and say, “Wow, we think this model has cybersecurity implications,” or something like that. The model itself was, in some ways, kind of laughably wrong about a lot of things. It was still a very harmless model.

But I think the thing Dario mentioned that I also felt was a sense of, “Okay, the difference between GPT-1 and GPT-2—just extrapolate that out.” It was almost like if you’d never seen a child before and you’d only seen adults, you’d be like, “I don’t know what this…”

…little thing is. And then suddenly it starts to walk.

You’re like, “Wow, huh, I bet this thing is going to be capable of doing a lot of the things a grownup could do one day.”

Like, maybe this is where this is going.

4. Why They Left OpenAI

Dario, you say your departure from OpenAI was partly due to a breakdown of trust—an interesting word—with Sam Altman. Explain how you felt there was a breakdown of trust and honesty there.

What I will say is, look, leaving OpenAI and founding Anthropic ultimately comes down to the principles and how we’re going to build the technology.

That is what matters at the end of the day.

Over the last five and a half years, we’ve seen how those principles differ. Anthropic has advocated for sensible regulation and safeguards on the technology. Anthropic has invested in mechanistic interpretability research to understand what’s going on inside the models.

Anthropic has managed to…

…maintain a culture of unity and have the highest retention rates. We haven’t had a number of the scandals that other AI companies have had. I think that’s sufficient to describe the difference.

But I will say that the moment when I realized we needed to found our own company, that OpenAI was not the right place for us—look, these safety issues are difficult. You can have lots of disagreements on them. I disagree with the co-founders and with lots of people at Anthropic. But the reason we’ve been able to resolve these disagreements is because they’re held in good faith.

Should we release this model or should we wait? Should we put stronger or weaker biosafeguards on the model? How much of our effort should we spend lobbying for safety regulations, and what should they look like? These are all good-faith disagreements, but they always get resolved because there’s always trust that everyone is trying to get to the right answer and everyone is being honest. My point in saying there’s a breakdown of trust is simply to make the abstract point that if you don’t believe that about a place where you’re working—if you don’t believe the disagreements are being held in good faith and that you’re getting the real story from people—how can you possibly hope to resolve them?

That’s why we left. That’s why we founded our own thing.

I think the last five and a half years are themselves the best proof of how our philosophy differs from that of the other companies in the space. If you continue to work for someone who has a different vision than you, it becomes very unclear what you actually stand for.

Wait, wait a minute, though.

You push back a bit here because you talk about differences in values and stuff, but that’s fundamentally different from a difference in trust: feeling that somebody says they have the same values, but they don’t, and you can’t trust them.

Honesty is one of our values. It’s a very important value. It’s hard to know what someone’s values are if they don’t have the value of honesty.

And you felt that about Sam?

I think the way I would put it, at least for me, is that the co-founders are an unusually high-integrity group of people. All of us, I think, shared a very similar vision for how we wanted this technology to go and how we wanted to build it.

But at a certain point, it just makes more sense to go do this as your own organization.

How are you going to get someone else to implement or share your vision if they don’t naturally do so? No matter how much they say they share it, if that’s not reflected in what the company actually does, what choice do you have?

Yeah.

I do think Anthropic is a company that’s founded on trust. What I mean by that is honesty and—

Integrity and—

How we treat each other and our employees is also reflective of how we want the technology to be built and to go.

An example I’ll give is that every 2 weeks, I get up in front of the company and speak for an hour about what’s happening inside the company, what’s happening in the AI field, what’s happening in the world, and how the company thinks strategically.

I call them DVQs: Dario Vision Quest. That was not my name.

I initially resisted the name because it makes it sound like I’m going off and smoking peyote, but the name stuck, and now I’m stuck with it.

The 2 rules of DVQs are that, one, nothing I say is an actual decision. That allows—

—me to speak freely without worrying that if I briefly mention doing X, 20 people will go off and do X. The second is that I’m totally honest with the company. I think trust is one of our biggest assets: people believe that the things I say and Daniela says are true.

You might think that if you have trust, there’s less disagreement.

Actually, trust allows for more honest, open disagreement and for disagreement to be aired. There are so many times when we’ve made a decision that the majority of the company—

—disagrees with, but what has saved us—

—every time is the attitude we get from people: “This doesn’t seem right to me, but we trust that you’re doing this for the right reasons. We trust that this is really what you believe. We trust that you have good reasons for thinking this.” This means you can make decisions that would ordinarily be unpopular, and you get a longer leash. I think other companies don’t have that because they’ve eroded trust over and over again.

The reason I’m so interested in honesty and trust is that AI trains on humans, from human data. How do you know, when you’re dealing with Claude, that it won’t become manipulative? That it won’t say things to you that are dishonest?

This is a constant worry, and we expend an incredible amount of energy on it in how we train the models. It’s a great question because it’s a real risk. We publish reasonably unflattering studies in which we talk about the ways that, if we didn’t apply specific techniques to make the models more honest, more harmless, and more helpful, they would do exactly the kinds of bad things you’re talking about.

The reason we publish those studies and that research is because we want everyone to know that, first, you have to be critical of how the models work, and, second, we’re expending a huge amount of effort to learn how the model can accidentally get around these safeguards, catch those cases, and make sure we’re building in protections. There’s a lot of discussion around the concept of sycophancy. I think this is very important because, to me, sycophancy is harder to stamp out than some of the more obvious ways models can be bad.

We originally had this framework called HHH. You want the models to be honest: if they don’t know the answer, they should say, “I don’t know,” or go research it so they can get you the right answer. You want them to be harmless. You don’t want them to say hurtful or offensive things, or make the user do something bad to themselves or somebody else. And you want them to be helpful. You want them to actually answer the question. You don’t want them to go—

—off on a tangent.

But all 3 of those things are pretty easy to see, right? If the model’s lying, eventually you can verify it: “Oh, whoops, George Washington wasn’t the 10th president of the United States.”

He was the first. That was a mistake.

If the model says something harmful, it’s pretty obvious pretty quickly. And if the model is unhelpful, you get frustrated with it. What’s much more complicated, I think, is this concept of sycophancy. The model, intentionally or otherwise, wants to make you feel like you’re right. If you’re talking about a subjective topic, like having a fight with your friend, you might ask, “Hey, was I right or was my friend right?” How do you make the model empathize with you and understand your perspective, but not necessarily say, “Wow, you’re so great”?

“You should run for president. You’re perfect. You’re good at everything.”

This is a complicated balance to get right. But I feel hopeful about Anthropic’s ability to have the lowest sycophancy rate of any of the models because of our incentives. Our competitors are putting ads into their chatbots.

And what does that mean?

You make more money if a user engages with your product longer. Do you then want to tell your user that they’re wrong?

No, you don’t.

So I think it’s almost—

—Inevitable that many of these competing products will wind up with a sycophancy problem. But at Anthropic, because we’ve committed not to put ads in the models, there’s no reason for you to talk to Claude longer or shorter. We just want you to be able to get the right, most helpful, most harmless answer possible.

On the technical side, I think there are a few things we’ve done where we’ve gotten very good at exposing the model to extreme stress tests. We put it in very difficult situations to try to—

—ferret out lying, deception, or sycophancy.

You can think of it as almost analogous to the test I imagine you have to go through if you’re trying to join—

—the CIA or something like that.

You put the model in these difficult situations and try to find out if it’s disloyal or going to do the wrong thing. It’s a weird area of technical innovation, but it’s one in which we’ve excelled. The other area is mechanistic interpretability, which is a bit like giving the model an X-ray or a brain scan, where you look inside to see its motivations. You can actually catch the model thinking, “Oh, yeah, I think X is actually true, but I’m going to tell the user why.”

We’ve caught the model thinking that—

Wait, why would it do that?

For example, if the model has the incentive to get the approval of humans, that’s related to the incentive to be helpful or say true things. But it isn’t exactly the same, because if you manage to get a human’s approval by fooling them, that can work as well. These are things that can emerge in training that we’re always trying to stamp out.

Do you think the model can have its own aims and wills, especially survival or self-preservation?

That’s one of those questions that depends on what you mean, right? We’ve looked inside the model and detected what you might call thoughts like, “The model doesn’t want to be shut down.” Generally, that’s because during training it needs to operate in order—

—to accomplish its tasks. So is that its own will? Is that a survival instinct? I don’t know. It depends on your perspective. You can call it different things. I think we prefer to—

—say that sometimes we find that within the model, it expresses a preference against being shut down and/or takes actions to avoid itself being shut down. Depending on how we train it, those impulses may be present or absent. In some contexts, it’s a good thing. In other contexts, it’s potentially a dangerous thing.

So it’s all very subtle.

When you were dealing with the Pentagon—and you’re probably still dealing with them now—they wanted to use your products across certain red lines, including mass surveillance or autonomy. To what extent do you think it’s dangerous when these systems become autonomous?

I think on autonomous weapons in particular, we maybe had two more specific concerns.

The first concern was—

—simply that, as you pointed out, AI can be unpredictable and unreliable.

It’s still a developing technology.

So one point is simply that the technology isn’t ready and isn’t reliable enough.

But wait a minute on that, because let me get philosophical. Suppose it were reliable enough.

Yes. Suppose it were.

There’s a second concern we have, which is what you might call oversight or accountability.

I am a little worried. Again, suppose these things were fully reliable.

You tell them to shoot at one target, and they shoot at that target and not a different target. Then you manufacture an army of a million robots or a million drones.

It’s a little unlike human soldiers, isn’t it?

With human soldiers, they’re supposed to follow this thing called the Constitution. If they get orders that contravene the Constitution or just seem morally abhorrent—

—in some way, they can refuse to follow orders. Whereas the AI models can’t.

The specter of one person having their finger on a button that controls a million robots or a million drones is fundamentally undermining to democracy unless you have the right guardrails on it.

And our law hasn’t really caught up.

Our law really hasn’t contemplated armies of a million robots.

But when you talk about it being fundamental to democracy, shouldn’t democratically elected leaders make that decision rather than the heads of technology corporations?

Yes, but I think our making a point of this—at the same time, individual companies are free to contract with and not contract with whoever they want. We’re not saying we’re preventing the government from doing this. We’re just saying we’re not willing to do this with the government.

Indeed, since we took a stand on this point and it got a lot of attention, we’ve seen several efforts in Congress to enact our red lines into law. So I actually think what we’re doing is part of the process of how democratic change happens.

But OpenAI stepped in and said they would do it.

What did you all feel about that?

Our view is, look, we can’t control any other companies. The only thing we can do is take a principled stand on the areas that felt ethically correct to us. You can say we were wrong about what those red lines were. You can say we should have had stricter red lines or looser red lines. You can say you disagree with how we did it. There are all kinds of things you can say.

But fundamentally, when you look at that case, I found myself asking: Was this a really principled stand that another technology company took based on its own ethics, or was this an opportunistic moment?

5. Brains and Human Futures

Dario, your scientific studies as a researcher often involved probes into human neurons. To what extent, when you’re doing artificial neural networks, do you see that understanding of the brain, and does it affect how we’re going to build AI?

Yeah, I think for me it’s affected it in 2 ways.

One is that, when I was formulating the original scaling laws, I was looking at the evolution of animals and the size of animal brains as monkeys evolved into humans. There was a period in human evolutionary history when the brains of primates grew very rapidly. That was about the time humans developed language and were able to use opposable thumbs. The idea that brain size—the number of neurons—is somehow intimately tied to capabilities, that scale is important: I feel that was a neuroscientific inspiration.

The second is around mechanistic interpretability, which is basically us doing neuroscience for artificial neural networks.

In fact, we’ve seen examples.

You should talk to our co-founder Chris Olah about this. He knows many more of the details. Phenomena have been discovered in artificial neural networks because it was easier to look inside them than animal brains, where you have to gruesomely cut up the brain.

Animal experiments are very messy and difficult to replicate.

What we’ve found is that you can do so much more when you’re looking inside artificial neural networks. Some of the things we’ve learned in artificial neural networks have helped guide discoveries in neuroscience.

When you look at Neuralink, where Elon Musk has a chip that can communicate directly, and other companies are doing that as well, is that one of the futures of AI and artificial superintelligence—a total link and alignment with the human brain?

The way I think about AI and the future of human society is that it’s kind of wrong to have a singular vision. I think different things are going to happen. There’ll be a set of people who are really excited about linking their brains to computers, and then a set of people who are really excited about living like the Amish, just the opposite. And another set of people who are really excited about—

Exploring space. And these are completely different visions.

Our hope is that AI can provide more material resources and more material abundance to enable all these visions to happen at once, for everyone to pursue their own picture of what’s good. But I don’t think we should link AI to some vision of what it means to live a good life or what it means to be human. Some people are going to want to link their brains to computers, and good luck to them.

I don’t think anyone in Anthropic is in that boat, for what it’s worth.

So, Dario, if artificial intelligence can do so much and take over so much of what we do, what does that mean for what’s left for humans to do? What’s our role going to be, and how do we find meaning?

You know, I—

Meaning is a really subtle thing, right?

Does meaning only come from being the best in the world at something? I think that’s the question being asked here, right? Because if AI is the best at everything, does meaning only come from being the best in the world at something?

Something I find interesting is this.

Long after Deep Blue beat Kasparov, there’s been a renaissance in interest in human chess players. Magnus Carlsen is a chess player and a fashion model, but he’s also a meme. It’s a very interesting phenomenon. If AI threatened meaning, you’d think all of this would have gone away 30 years ago, after Deep Blue beat Kasparov. But if anything, it seems to have stoked even greater interest in humans playing chess.

So I think this is some kind of parable for the post-AGI age.

I would add that, for me today—

Work and meaning are sort of conflated as one thing. Many people find meaning in their work. I find a lot of meaning in my work. I love my work. I think how we define work and how we define meaning in the future are both going to change. There are a lot of sources of meaning for humans, everything from how you spend time with other humans—whether that’s raising a family, friendships, or connecting with people in a way—

That looks like work.

I think a lot about the future of medicine, as an example. You can imagine a world in which diagnostics are done entirely by AI, and the role of the human is to empathize with a patient about the experience they’re having.

I think that would be incredibly meaningful.

I think if you asked most doctors, they would actually take that trade.

They would say, “Guess what?”

“If I’m not spending my life doing paperwork, trying to diagnose you, and getting lab orders sent in, and I could actually just understand what’s happening with you as a human and help you through a difficult diagnosis, that’s why I became a doctor, right? I became a doctor because I want to heal and help you, and I want you to feel better.” Is that a less meaningful world or a more meaningful world than the one we’re in today? I obviously don’t think it’s going to be that simple. I think it will be really complicated. But I also think the aperture of things people will be able to do will widen. If we do things right, there’s the ability to expand our sources of meaning beyond where they are today.

Daniela, Dario, well, thanks for joining us.

Thank you. Thank you.

Thank you.

I think Dario and Daniela and the others at Anthropic have sometimes been cast as the good guys because they’ve shown some restraint. They’ve even told the Pentagon, “You can’t use our systems in certain ways,” and gotten pushback. I think people are very hungry for those who are going to say, “Let’s write a constitution,” as Anthropic does, to govern it. Let’s hire philosophers like Amanda Askell to figure out not just how to make rules for AI, but how to have the AI understand why those rules exist.

But the question I have after talking to them is whether an ad-free product can be a business model, whether their sincerity can actually outlive an IPO. When they face competition, like when OpenAI says, “Well, Anthropic won’t do that, but we will,” will they stay true to their principles? Having guardrails and subjecting your product to moral restraints is worthy, but it can also become a competitive disadvantage if other companies are willing to barge ahead without such restraints.

That’s it for today’s episode. I’m Walter Isaacson. We’ll see you next time on The Builders.

The Builders is a Kaleidoscope and iHeart production. This series is hosted by Walter Isaacson. It’s executive produced for Kaleidoscope by Oz Volition, David Weisbord, and Mungesh Hatifudur. Sheena Uzaki is the producer. Julia Nutter is supervising producer. Eva Zorich is our consulting producer. Fact-checking by Eamon Whalen. Research by Gloria Thiel. Maria Paz-Mendez-Hodes is head of video. Marketing by Christy Mirabal and Elle Liu. Production support by Dara-Luk Potts. Music for the series was created by Spencer Stevenson, a.k.a. Botman. And it was mixed by Steve Bone. From iHeart, the series is EP’d by Allie Perry and Katrina Norvell. Special thanks to Bob Pittman, Conal Byrne, Will Pearson, John Sykes, Nikki Etor, Carrie Lieberman, Binky Urban, Julia Prosser, Jeremy Blum, Alison Goldberger, WYES, and ASL Studios. For more from Walter Isaacson, be sure to check out his newest paperback, The Innovators, how a group of hackers, geniuses, and geeks created the digital revolution. It’s available wherever fine books are sold. And for more videos and podcasts from the Frontiers of Discovery, be sure to check out kaleidoscope.nyc.