奇点倒计时：2029年实现AGI，人类与AI融合，智能提升1000倍｜Ray Kurzweil

It feels like we're in the midst of the singularity. Do you agree that we're actually in the midst of it right now, or are we going to have to wait for some other point to get there?

I've been in AI for 61 years, which is actually a record.

If you look at your 120-odd predictions from 30-odd years ago, only 3 were wrong. Your first prediction, as you said, which you released in 1999, was that we're going to reach human-level AI by 2029.

The next 10 years will get us to my definition of the singularity, which is that we'll all be at least 1,000 times more intelligent.

What is most exciting to you, and what are you anticipating most excitedly in the next 1 or 2 years?

We'll have supercomputers, but we'll also be merging with them. So, we're going to be made a lot more intelligent than we are today.

When?

That's going to happen at the same time for everybody.

2026 is off to an extraordinary year. I'm here with Dave Blundin, Dr. Alexander Wissner-Gross, and Salim Ismail.

Where are you today, Alex?

Yeah, where are you?

I'm in the first arrondissement of Paris today, slowly making my way to Davos for the World Economic Forum 2026.

Taking the slow route, like a horse and buggy or something. You can fly direct.

Taking the slow route.

Scenic route. Paris in January.

I usurped your normal recording spot. So, this is your background and your mic and everything. Here in Santa Monica, we have an XPRIZE board meeting today. Dave, are you going to be joining the board meeting by Zoom, or are you not here?

I'm with Ray here in Boston. Actually, we're in the happening spot, but I'm going to be flying straight from here to Davos on Sunday, where Alex and I will be hanging out with Demis Hassabis and the whole gang.

Amazing. I just got back from Singapore, and I had an extraordinary visit there. I was the guest of the incredible bank DBS and Su Shan, who's the CEO. A big shout-out to Su Shan. Thank you for an incredible visit to Singapore.

She's a Singularity University alum and a fan of our pod, so I just think the world of Singapore. I can't wait to get back there. DBS is doing extraordinary work, so a big shout-out to the team there.

Gentlemen, we have an extraordinary guest today, someone whom all of us count as a mentor. He's been a mentor for me for the last 20 years. We're here with the incredible Ray Kurzweil, one of the world's leading thinkers and futurists. He's been called the relentless genius and the ultimate thinking machine.

He's got a 30-year track record of accurate predictions regarding the evolution of technology and the future. If you go to Wikipedia, you can check it out: an 86% accuracy rate on his predictions. He's the inventor of the CCD flatbed scanner, the first omni-font optical character recognition system, the first print-to-speech reading machine, and the Kurzweil synthesizer. He's the author of “The Law of Accelerating Returns.”

We'll be talking about that, as well as the author of 2 books that have set the foundation for all the conversations we have here in Moonshots: “The Singularity Is Near” in 2005, and, more recently, “The Singularity Is Nearer” in 2024. He's the recipient of the National Medal of Technology and Innovation. He has 21 honorary doctorates, has been honored by 3 U.S. presidents, and is really the gentleman who popularized and drove the term singularity, which he famously predicts will happen in 2045.

Ray, it is an honor and a pleasure to have you here, buddy.

And a bucket-list item.

Absolutely.

It's great to be here. It's always great to talk with you, Peter.

Yeah, no, and you have to love those suspenders, buddy. You are fashionable in the exponential world.

I do have to say, they're all hand-painted.

Are they really?

Yeah.

When I read “The Singularity Is Near” in 2005, when it came out, I thought it was the most important book I had read in my entire life up until that point. Definitely a life-changing book, worth buying again and rereading.

It was quite controversial when it came out, which was about 20 years ago. Stanford had a meeting of several hundred AI experts to examine its predictions. It was considered very controversial. People agreed with me that it would happen, but not within 30 years. They thought it would happen within 100 years.

I'm actually running into people who were there. Several hundred AI experts came to that conference, and they agreed that, if anything, 30 years—2029—seems overly conservative right now. People are predicting a little bit sooner than that, like 2027 and so on. But at the time, people thought it was 100 years off.

Well, I think it's important for people to go read the book. It's so noncontroversial today, given how things have unfolded, but put yourself in the mindset of this being completely controversial at the time, because a lot of things that we predict on the podcast, that Alex says, also have that same flavor. Trying to look forward 10 years from today is very, very hard, and they have that same feeling of, “Well, that's impossible. That could never happen.”

But if you rewind the tape, these impossible things routinely happen. Then, because of hindsight bias, everyone's like, “Oh, I would have seen that coming.” So, I think it's a good exercise.

Things are happening so quickly now that looking 1 year out is like a long-term prediction.

Yeah. So, Ray, you made 2 predictions, and I think it's important. Your first prediction, as you said, which you released in 1999, was that we're going to reach human-level AI by 2029, and people laughed at that, as you said. But the other prediction you've made is that we're going to reach the singularity by 2045.

There's a lot of confusion about this. If we're reaching human-level AI by 2029 and it's growing exponentially, why are we waiting until 2045 for the singularity? Could you explain the difference between those 2?

We'll multiply our intelligence a thousandfold. One difference in my own perspective versus everybody else's is that it's not like we have our own intelligence—our biological intelligence—and then we have AI over here, and somehow relate to AI versus human intelligence. We're going to merge with it. It's going to be the same thing.

We're not going to be able to tell whether an idea is coming to us from our biological intelligence or our computational intelligence. It's going to seem the same. I mean, if I ask you to think of some actress and you think of one, you don't know where that came from. It just somehow appeared in your mind, and it's going to be the same way whether it's coming from your computational intelligence or your biological intelligence.

We're not going to be able to tell the difference. Today, you can tell the difference if you actually go to your favorite LLM. You can tell that it's coming from the LLM, not from your biological intelligence. In the future, though, you're not going to be able to tell the difference.

And we're going to become 1,000 times smarter by 2045.

It feels like we're in the midst of the singularity. It's a smooth function, so it's hard to know that. Do you agree that we're actually in the midst of it right now, or are we going to have to wait for some other point to get there?

A lot of things have already been amplified dramatically. For example, we can take our models of biological paradigms and predict what will happen if we can actually simulate biology. We're doing that now with biological tests, so we can simulate millions or even billions of different possibilities and do that in 1 weekend.

Ray, how do you define the singularity currently? In the past, you've put it as a moment in time, and then we talked about it as a process. What's your current framing of it?

The framing is when we're 1,000 times more intelligent. In some ways, we'll be able to simulate biology for medical tests even faster than that, and we can do that today, although we don't have all of the paradigms of what biological intelligence will do.

I've talked to people who are actually modeling this, and the most conservative view is that in about 5 years we'll be able to have all of the computations done with biological intelligence, predicting what different chemicals will do. We can actually try out 1 million tests in 1 weekend and be able to predict that very, very quickly.

We can do that now in some cases, but not in every case.

I'd love to rewind the tape just a little bit and talk about why—or how—you landed the plane so accurately. Your predictions going back to 1999 are coming down to basically within a year or 2 of what you predicted, which is so different from when I was at the MIT AI Lab. People were predicting all kinds of different things, and then they would never happen. Then we got into these AI winters.

If you go back and read your books from 2005, you have to put yourself in the context of nobody believing AI would ever happen because it had been predicted like 12 times in a row and whiffed every single time. Every prediction had absolutely whiffed. Meanwhile, you're drawing a timeline that's much longer than other people's timelines, and the date of AI having human-level intelligence is going to land within 3 years of something you predicted 20 years ago.

30 years ago.

30—is it 30 years ago?

Yeah, 1999 to today.

Yeah. And then the date when it crosses all combined human intelligence—which I guess is 2045 in your prediction—will likely happen or be sooner.

It has to do with thinking exponentially. People are not used to that; they're thinking linearly: if it took 10 years in the past, it'll take 10 years in the future. That's really what people think about the future: it's the same as the past. To really think exponentially requires a certain practice, and that's how I got to this kind of view. Alex, do you want to jump in?

Yeah, maybe to pull on this thread, Ray. First of all, it's wonderful to be chatting with you again. I always enjoy our conversations.

The Turing test, I've argued on this podcast in the past, went by with a whimper, not a bang. It flew by. The Loebner Prize was canceled before the Turing test was arguably passed, and yet it was passed and there was no celebration.

The Loebner test was not a really good test. It had various practices that were really not in accord with the Turing test. The Turing test is really matching an ordinary person who's talking, not really an expert in the field.

AGI, I think, is actually a better view because we're matching the best person in each field. We have maybe several thousand, maybe several hundred thousand fields that you could be expert in, and AGI means that you can match a human being in any of the fields and then combine the insight into many different fields together, which no human being can do.

Einstein was very good at physics, and he was actually interested in playing the violin, but he was not an expert in playing the violin. He was only an expert in physics. People maybe can master 2 fields at the most. But there are actually thousands of fields, and if you could be an expert in all of them and then combine all those insights, that's something quite unique. So that's what AGI represents, whereas the Turing test is really matching an ordinary person with a lot of mischaracterizations of different things.

I agree that AGI and passing the Turing test are, by most common definitions, different standards. The question I was going to ask, though, is this: if you agree with the premise that the Turing test, as reasonably defined—not the original gender-presentation-based Turing test, but the subsequent definition—was passed without very much hoopla at all, do you think the same is going to happen with the singularity?

One of my favorite scenes in Charles Stross's novel Accelerando has a bunch of characters who've all been uploaded to a starwisp traveling to another star system, all arguing with each other. They're posthuman uploads arguing as to whether the singularity has even happened. Do you think that's what's actually going to happen here, where the singularity will just zoom by and we'll all be arguing with each other decades later: Did the singularity even happen? Has it happened yet?

These standards are not very clear. Not everybody agrees that we've passed the Turing test, and when we pass AGI, there'll be disagreements. There are disagreements now as to what that means. People say it's basically as good as somebody who's a little bit above-average in intelligence.

I define it as being an expert in every area, when there are many different areas that you can be expert in. That's actually quite an impressive level, and I think we'll get there by 2029. The thing is that then you can combine your insights into every possible field.

We already have large language models that can answer questions in lots of different fields. No person can do what a large language model can do today, let alone what will happen by 2029. By the way, we have a Moonshots test where you have to fool your spouse for 3 minutes on a Zoom call.

[Laughter.] We haven't defined what we're going to give to the listener.

We should do that. That would be hilarious.

Well, I think that's a better benchmark. So that's our Moonshots test; that much more closely matches the original Turing test.

Sorry, were you going to say? I was just going to say—or rather, to ask Ray—are you at all concerned about goalposts getting moved yet again, as we see happening over and over again with definitions of AGI and otherwise? Will we pass your definition of the singularity, but nonetheless have most commentators arguing with each other for a long time afterward about whether the singularity has actually happened?

Well, mine is actually pretty strict. To pass my definition of AGI, you have to be an expert in thousands of different areas, which is actually more strict than most definitions of AGI. So I think I have a suitably strict definition of it.

What about the definition of the singularity? One of the things that really inspired me in both of your Singularity-titled books is the fact that there's a moment in time where AI is working on itself and self-improving, and that moment in time is where we get this incredible acceleration. It feels like that's either right now, within the last year, or within the next year. It's imminent, and we're predicting on this podcast a 100x step-up in the efficiency of the existing algorithms that's completely independent of the underlying curve.

You've started to see AI improving itself a little bit, but it really has not gone anywhere; it's not really very dramatic. These definitions are not beyond debate, and it's not like everyone will agree. Take AGI. You could predict that a certain number of people will say that it's actually there today, but it's a small group. It will accelerate, and finally everybody will more or less agree with it. That's a band of maybe 3 or 4 years, and I think it will end in 2029. It's already beginning. People feel we have AGI already, but most people will believe that, I think, by 2029.

Well, that means your prediction has to be exact. If you say that we'll be debating it for the rest of time and it was sometime between today and 2029, that means you are irrefutably right in your prediction from 30 years ago. So that's kind of cool, right? Let's memorialize that right now.

See, you were going to jump in.

So, I remember, Ray, when we were in a car with Peter—you, me, and Peter—going to the CNN studios to launch Singularity University and announce it. I was a young, fresh-faced fellow, and I said, “Ray, they're going to ask you about exponentials as part of the briefing,” and they said, “Oh, oh, oh, oh, oh, that may be a problem.” I said, “What? What do you mean?” I was all kind of freaked out, and you said, “I'd better bone up on the subject.” It took me like 10 seconds to realize that you were joking.

I think one of my favorite things about you is the unbelievable sense of humor, the dry humor that you bring to the table. Here's my question for you: you've been saying this very steadily for 30 years, right? At the beginning, it must have been very hard, saying this to people who were just like, “He is out of his mind. What is he talking about?”

Is it easier for you now? Do you feel a sense of accomplishment that many more people are talking about it and saying, “Yep, he was right,” et cetera? Do you feel some sense of that?

Well, yes and no. The basic debate about whether or not this will happen, and whether it's going to be exactly 2029 or something, has gone away. People actually accept that. I run into very few people who say, “Oh no, it's going to be 500 years from now.”

On the other hand, the issue has changed from “Is it going to happen?” to “Is it good for humanity?” And that's a big debate. Yes, it's going to happen, but we're all going to be screwed as a result of it. We've got books that come out saying it's going to eliminate humanity. That's really the big debate now: whether or not it's going to be beneficial for humanity.

I believe—you've said publicly that technology is a major driver of progress, and it might be the only major driver of progress. I assume you're very clearly on the beneficial, pro side.

Yeah, yeah. There's some chance things will go wrong. I wouldn't say that has no chance of happening, but I think what we're seeing is going to be beneficial, although it's going to change things very rapidly, and that will lead to some foreboding as well.

Yeah. And we'll get into that in a minute. There's a question that we've debated on the show, and I'm curious about your point of view: Are we going to actually achieve consciousness and sentience with AIs, and will they begin petitioning for personhood? Do you think society will approve that? Are we going to start to feel like our AIs are conscious and sentient, and that we shouldn't shut them down, and that they're going to have rights like humans have? What's your feeling on all that?

Well, first of all, consciousness is a subjective point of view. There's nothing we can do scientifically to prove that an entity is conscious. We can't have a machine where you slide something in and a light goes on: "Oh, this is conscious. No, this isn't conscious." There's no scientific test for it. Some people, like Marvin Minsky, who was my mentor for 50 years, said, "Well, there's no scientific test for it; therefore, it's not scientific. Therefore, we shouldn't deal with consciousness. It's a meaningless debate."

On the other hand, you could say it's the most important thing. Am I conscious? Are you conscious? That's something we really need to deal with. I need to be able to relate to you as if you are conscious. I consider myself to be conscious, and yet it's not scientific.

My scientific test is that I think I'm conscious, but my wife disagrees. When she thinks I am, then I think I'll be there. Alex, you've been thinking a lot about this idea of personhood and consciousness. I'm a proponent, broadly speaking, of AI personhood, and I guess I'll play the contrarian role that I'm painted as by respectfully disagreeing with my friend Ray that there aren't benchmarks. What do you think?

I think there has been, over the past 2 years, marked progress toward developing quantitative benchmarks for, call it, self-awareness rather than consciousness—maybe slightly less mushy as a term—including, as I've pointed out in the past, tests for whether certain models can detect overlaid activations in their residual streams if they're transformers. I see progress toward developing real benchmarks for self-awareness in models.

Yes, but I'll give you something else that's even more perplexing. There are lots of conscious people. I can't prove that you're conscious, but I believe that you are. I believe that a human being that acts conscious is probably conscious.

But why do I have the consciousness I have? There are all these conscious beings, but there's one person that I relate to that, if something happens to it, I care about it in a different way than I care about other people: my own consciousness. So why am I conscious? Why was I born in 1948? Why am I male on Earth, and why am I not another animal? Why am I the person that I am? You could think the same thing about yourself. Why do I have all the earmarks of this particular person? That's really hard to explain.

Of course, Ray, it's such an ironic question that, in my mind, you're asking an anthropic question. What you just posed—"Why am I myself?"—is the most fundamental anthropic, lowercase-a—not capital-A—question that one can ask. And why does the universe appear the way it does? The usual answer is that if the universe or your own identity had sufficiently different properties, you wouldn't be around to ask the question.

It's very hard to even ask the question, and people don't actually quite understand it.

Maybe the most profound comment you've ever made to me was when we were at a group of singularity folks. We'd had a couple of glasses of wine, and somebody asked about consciousness, and you said, "Language is a very thin pipe to discuss concepts that are this complex." It just blew everybody's mind. AIs will be indistinguishable from a conscious being, and we'll just keep going, and finally we'll accept it. When, Ray?

Sam Altman might say right now that it's conscious, and people aren't really sure. But eventually it keeps having all the earmarks of a conscious being, and you will accept it because it would be useless not to have it. Again, you can't say that's going to happen at the same time for everybody, but I think when we're a few years into AI entities acting conscious, we will accept it. I don't think it's going to be a very long delay. That's already beginning to happen.

One thing I love about today's AIs—you know, I use them all day long, every day—is that they have no intent of their own. They just do what you ask them to do, and they try to be as helpful as they can in getting you to whatever destination. But they're not trying to get to any destination of their own.

When you start saying, "Well, they're going to act conscious," that implies to me that I'm trying to get somewhere on my own. I don't have time to help you right now. I'm busy with my own personal agenda here.

Dave, good point. I'm still waiting for the AI to call me up one day and say, "Hey, Peter, listen. I'm working on this thing over here. You can join me if you want, but this is my objective for the day."

Yeah, that's a different world.

You know, Ray, something you said on the Abundance stage—it was yourself, myself, and Salim Ismail. We were talking about this, and you made a statement that really rocked a lot of people. To contextualize the speed, you said that in the next 10 years—originally, you said 2025 to 2035, this decade going forward—we're going to see as much change as we saw in the last 100 years, from 1925 to 2025. Back when the highest level of technology was the Ford Model T, 30% of homes had electricity, and telephones were still becoming widespread, do you still hold to that level? Is it faster or slower? A hundred years of progress in the next decade—are you still holding to that?

Sounds about right. Think about the difference between 2025 and 2035. In 2035, we'll be way past AGI. We'll have supercomputers, but we'll also be merging with them, so we're going to be made a lot more intelligent than we are today. That's a huge amount of progress compared with what we've done in the 100 years before that.

How do you see society dealing with this? Right now, the limiting factor in a lot of areas is regulation, social structures, norms, and market capture. What do you think is the weakest point that we should focus on solving to allow this progress to be implemented into the world?

It's going to be a major thing. Employment is not—right now, employment is considered equivalent to being able to deal with your own financial needs. That's going to change a lot. We'll be able to produce enough things that everybody will be wealthy compared with what we now consider wealthy, and yet we won't necessarily have jobs as such. How we're going to deal with that is really unclear.

People are actually not that concerned about it. You would think that if—

Well, it's because they're in denial.

Yeah.

No, they're just not. I can't tell you how many people I interact with who are running companies—hundreds of them—and 90-plus percent are just like, "Yeah, it's not happening," or, "Things always take longer than people say," or it's just pure denial.

Yes, but I think we'll deal with it. Okay. It's going to be a major change in the way we organize society.

There are folks like Mo Gawdat and a few others who think this, and you've said this, Ray: The next 10 years are going to be the most volatile while we try to absorb all of this. Do you agree with that rough time period, Ray, or do you think it's longer or shorter?

I agree with it, but it's not like it's going to end in 10 years. We won't have this flux of great change in the next 10 years and then have the 10 years after that be smooth.

No, it'll be much, much crazier.

I mean, the next 10 years will get us to my definition of the singularity, which is that we'll be at least 1,000 times more intelligent.

I'll maybe pose a hopefully less obvious question for you, Ray. You've been very public about keeping and maintaining lots of documents and artifacts from your father, whom I gather was a tremendous influence on your life, with the premise that AI is going to enable you to computationally reconstruct your father someday. If I'm not misconstruing that, there is a related notion that has been called variously quantum archaeology or humanity's final task. Tsiolkovsky has written extensively about this in the context of Russian cosmism. My question for you is: When do we get the ability to computationally resurrect dead human beings with AI?

Well, I mean, prior to that, we could try to create avatars of ourselves.

We did create one of my father, and I'm creating an avatar of myself now. I have a lot more material that we can put into text. I have 11 books, and I've got several hundred articles that I've written—articles about me. All of this will go into a large language model.

We'll create something that can talk like me, and it will look like me. I get probably 5 to 10 requests for interviews and podcasts a day, and I can't do most of them. So I'll offer them the opportunity to interview the avatar. The avatar will actually be better than me because it will remember everything. I don't remember everything that I've said.

You can interview the avatar as long as you want, in whatever language. You can do it in another language, right? And that'll be this year.

What age are you going to make yourself in your avatar?

It's kind of an arbitrary choice you have to make.

Yeah.

Now, that's not actually creating everything about me or my father, of whom we have less material, although we have enough to create an avatar that's also lively. Being able to recreate everything that a person has, including the state of their body and so on, will happen eventually, but that's probably another 10 or 15 years away.

Do you view that as the killer app of the singularity—the so-called great task of computationally resurrecting with AI every human who has ever existed?

That's one of them. Yeah.

There's so many. To me, what I'm very interested in is being able to achieve longevity escape velocity, where a year goes by, you age a year, but you get that year back from advances in medicine that keep you going for another year—or more than a year. So you don't actually age during that year; you'll get it back from advances in medicine and so on. What's your current prediction for when we hit escape velocity?

2032.

2032. Yeah. Let's jump into that subject, which I think is of common interest to all of us. Ray, you and I have had so many conversations about this concept of longevity, which was a very controversial subject a decade ago, and now AI is impacting biology and making it happen. When we've talked about reaching longevity escape velocity in the past, the technology that I believe you said is required to really get us there is nanotechnology. Do you think we're going to reach LEV without nanotechnology, just based on drug discovery using AI?

It really has nothing to do with nanotechnology. Nanotechnology is a way for us to take advantage of AI without it being obvious. I can be thinking about something, get an idea, and not know whether it's coming from my biological brain or the computational brain. That has to do with nanotechnology.

Longevity escape velocity has to do with advances in medicine. It has to do with being able to simulate what happens in medicine. We have to be able to create biological models of what happens in biology very quickly, so that in one weekend you can simulate millions or billions of different possibilities, try them out, test them, and then be able to move forward with a cure based on that type of analysis.

Talking to people who are working on this, 5 years is an outside limit. If we actually do it in 5 years, then it will take another couple of years to go through most of the medical problems we have.

So your advice to people is to stay healthy until we get to the early 2030s.

Exactly. Exactly.

Yeah.

Just curious to drill in one level deeper: Peter, you're also a top expert on this topic. If you had a perfect simulation—you knew exactly what's going on in a body, you'd nailed it all through computation—and that's about 3, 4, or 5 years from now, what's the intervention if not nanotechnology? Is it just more and more targeted chemicals in your bloodstream? How do you act on that simulation?

You're coming up with new cures and new treatments to both ward off and avoid getting these types of diseases, like cancer, for example.

Yeah. I think that applies to cancer, heart disease, hip replacements, knee replacements—all those things fit that mold. But then you've got this general aging, because stretching out your life is reversal, right?

Yeah. Yeah. Exactly. Take heart disease.

So Repatha is a new type of drug that dramatically reduces your LDL. I've reduced my LDL to about 10, which is a very low number.

Yep.

I've actually examined my arteries, and I have no plaque. That wasn't true 4 or 5 years ago, or even 3 or 4 years ago. In various areas, I'm developing things that are helping me avoid problems that didn't exist just a short while ago.

That's a good example of a chemical in your bloodstream. The traditional version of the world is that it's a new drug, a new chemical in your bloodstream. So there's a version of the world where that's all you need to reverse aging, and then there's a version of the world where you need something much more targeted—the work of David Sinclair, who is currently doing gene therapy for age reversal, for epigenetic reprogramming, and then heading toward 3 molecules. It's a very cheap oral supplement that you take to reset your epigenetic age.

Ray, do you have a target age you're shooting for? To hit LEV, do you expect—

I would very much like to be alive tomorrow and take advantage of all the friends I have, like the friends in this virtual room. I think that tomorrow I'll also be interested in being alive the next day. I can't imagine getting to a point where I wouldn't want to be alive.

The only time that people generally take their lives is if they're in insufferable physical, mental, or spiritual pain and they can't continue. Otherwise, people want to remain alive. I would want to stay healthy and be able to take advantage of that, so I'm not going to get to a point where I'm not interested in being alive.

As time goes on, we're going to get more and more AI. It's going to be more and more intelligent, and it's going to be able to keep our bodies going. I can describe today a way in which we can replace every one of our organs, and we can actually imagine that it wouldn't take that long. Certainly within a decade or two, we can replace all of our organs with something that would really last forever, more or less.

As time goes on, we have more and more capability to replace things that are going wrong with our bodies. We'll get more and more into longevity escape velocity as time goes on.

Are you anticipating a world where everybody agrees? If you said, "I'm alive today. I want to be alive tomorrow, and tomorrow I want to be alive the next day," are you anticipating a world where everybody gets on board with that within 10 years and everyone has those options? Or a world where a subset of people have had 5 organs replaced and stem cells put in their brains to extend their thinking ability, while another subset are violently opposed, ranting in the streets, trying to prevent it, and wanting natural death?

You can get natural death today. You can go to Switzerland and get natural death.

I was debating with Daniel Kahneman, who was a Nobel Prize-winning economist, and he was 90. He was actually very healthy. I would meet with him in New York, and I had 4 or 5 lunches with him. He would walk about 5 blocks to get to where our lunch was and walk back, so he was actually pretty healthy.

But he was mindful of what happens to you in your 90s, and he said that bad things happen and he'd rather know that they wouldn't happen to him. He took his life. He went to Switzerland and ended his life, even though he was healthy.

I wasn't aware that he had this plan, although his family was aware of it. I tried to talk him out of it and talked about how we're making exponential progress on overcoming diseases and so on. He was concerned about his kidneys, but I related some things I'm involved in that relate to the kidney, and he understood what I was saying. It was actually an economic issue. But he ended up taking his life anyway because he really was not convinced that this would happen.

Yeah. Ray, my father passed away a year ago at 97 and also had an assisted death in Canada. They've now approved it. I have never seen anyone as happy in my life as my father was in the last week.

I asked the doctor after he passed away, "I'm trying to feel loss or pain or suffering, but I can't. I've never seen him so happy. Have you seen this?" She said, "You know, 20,000 people in Canada have had this procedure this past year. Most of them go out in this state, and we think it's because they have agency." He lived with dignity.

He wanted to pass away with dignity, and he got his wish. He was happy as a clam—a very philosophical, thought-provoking outcome.

Yeah. I don't think that would be me, but [laughter].

Hope not. Alex, you were going to ask a great question about cryonics.

Yeah. No, I don't like very much the direction of what we're discussing here. I don't think, Ray, this aligns at all with the way you see the world either. I think you and I probably see the world quite similarly, rather than having hand-wringing discussions about death with dignity and going to Canada.

I would argue that we should be talking about cryonics, recognizing that approximately 150,000 people are dying every day in our world, and not everyone will survive statistically. If we get to longevity escape velocity by the early 2030s, as you predict, that's many, many millions of people who are going to die between now and LEV. Why do you think more people aren't obtaining cryonics plans for themselves? What can you say here—we have hundreds of thousands of subscribers and hundreds of thousands of viewers—to encourage viewers to consider getting cryonics plans for themselves, so they don't have to move to Canada to die with dignity if they're in that position?

There are companies like Alcor where you can sign up, and near the moment of your death, they will effectively put antifreeze, or some equivalent, into your bloodstream, and you will be frozen with the notion that eventually technologies like nanotechnology will reconstruct your full neocortex. That is what cryonics is based on right now. I would say, Ray, it's unconscionable to me. I think you have the statistics. I think probably a few thousand people, on the order of magnitude, have cryonics plans. Why do you think it's not hundreds of millions? Again, is there anything that you would care to do? You're speaking to hundreds of thousands of people who take the future of technology very seriously. Could you perhaps persuade them, if you think this is a righteous act, that they should be considering cryonics plans for themselves?

Well, my point on cryonics is that it's Plan D.

Plan D. [laughter] I love that.

Plan A, B, and C are to remain alive one way or another. Cryonics is Plan D. I mean, I have enough trouble keeping track of my ideas when I'm able to give arguments for them. It'd be hard to imagine keeping track of them while I'm basically dead.

I have concerns about it. You may come back, and you may not be happy with the way you come back. Cryonics is better than not doing cryonics because at least you have some chance of coming back, but there are risks with it. So, I do it. Very few people do it. I mean, the number of people who die who elect cryonics is very, very small. I have done it. I hope it works. [sighs]

You've signed up for cryonics?

Yeah, but I hope I won't have that opportunity.

For our viewers and listeners who don't know what this is, there are companies like Alcor where you can sign up. Near the moment of your death, they will effectively put antifreeze, or some equivalent, into your bloodstream, and you will be frozen with the notion that eventually technologies like nanotechnology will reconstruct your full neocortex.

I would say, Ray, it's unconscionable to me that—I think you have the statistics. I think probably a few thousand people, on the order of magnitude, have cryonics plans. Why do you think it's not hundreds of millions? Again, is there anything that you would care to do? You're speaking to hundreds of thousands of people who take the future of technology very seriously. Could you perhaps persuade them, if you think this is a righteous act, that they should be considering cryonics plans for themselves?

Perhaps, but given that I have limited persuasion with people who listen to me, I would tell people they should do everything they can to stay alive. That's because the best way to be alive in the future is to stay alive right now. There's a lot you can do to remain alive.

And, Ray, are you saying you should not just stay alive, but stay in reasonable health?

Yeah, absolutely.

And that the technologies will unveil themselves to you in the next 5 to 8 years?

Yes.

And it's happening very quickly, so this is actually a vital time to remain alive. I'm still chuckling at your comment where you said it's harder to keep track of your ideas when you're dead. [laughter]

But you're going to—in this, whether you're keeping yourself alive and you enter longevity escape velocity or you're cryonically frozen—the other thing going on is you probably have a hundred, or a thousand, or a million avatar versions of you that are up and operating in the universe in parallel with your meat body, right?

Yeah.

Whether or not those will have consciousness, we get back to the same thing we discussed earlier. Actually, they'll probably be better at remembering everything I've said, because if it has a computer behind it, it won't forget anything, unlike myself.

We should definitely do a podcast where it's Ray K.'s avatar, Alex's avatar, Dave, and Salim's avatar having a conversation amongst ourselves. We should put that on the docket for sometime this year.

Just one thing. I did a conference with Martine Rothblatt. This was at UCLA to represent their progress over the last—I think—30 years. It had me, Martine, 2 professors there, and Martine's avatar.

Martine's avatar looks realistic. It's like doing a Zoom with her. The avatar is actually very good. It remembers everything that Martine said, and you could ask it anything. It's very convincing, and it actually knew when to come in, because if you're in a conference, you can't just suddenly say something if somebody else is speaking. You have to wait until there's a silence, and you can say something—maybe that's relevant to what was said before—and it worked very well.

So, this was a conference with the avatar and Martine herself at the same time. Hey, can I ask?

Well, directly related to that, I stumbled a couple of years ago on your “How My Predictions Have Fared” essay, which is a great essay, by the way. Eighty-six percent outright correct. You worked on speech recognition years and years ago, and by now the interface to your computer, you would think, is voice, not a keyboard.

I feel like we're so used to that now that we're underpredicting how this interface is going to change for the first time since the GUI, so maybe the 1980s. But it's got to be imminent now, and I don't know if you agree with that or not. When you look at these avatars that you're just describing, they're so good and so convincing, and so much better of a way to interact with technology.

Well, another one I got wrong was that we would have self-driving cars.

Yeah.

Which we do now.

Yeah.

But it didn't quite make the time frame, so that was wrong.

Well, that one was wrong because of regulatory issues, right? Your timeline on the technology, I think, was incredibly close.

But regulation is very hard to predict. I think you made that point in the essay. The one on the interface to a computer isn't held up by regulation. It's something else—momentum or barriers or Apple not doing AI or something. [laughter]

That one, to me, feels like this is going to happen very, very soon. When you talk to an avatar, like you said, you're at a conference—why am I not talking to my computer that way? It's crazy that I'm typing on this keyboard.

Well, I think part of that, Dave, is having to not be verbal in the middle of an airplane flight or sitting at your desk sometimes.

I'll tell you, before we kicked off the pod, Peter, you were saying, “Where's our AI that's our avatar, basically?”

Yeah.

And, you know, pulling in images—when we talk about Ray's books, why is it not popping up as a picture in real time that we're all looking at? That's got to be imminent too, because the—

Dude, let's start that company.

Let's start that company.

Amen. Amen.

I thought, Salim, you were going to jump in.

Ray, if you look over the last 6 months, what breakthrough or development has surprised you the most?

I'm getting much more credence from people who didn't accept this a year ago.

I mean, think of the difference between 2024 and 2025—or January 2026 and January 2025. Most people a year ago that I would speak to would say, “Yeah, AI is pretty interesting, but it’s not really very good, and people don’t really accept it.” They’ve completely changed their views in the last year.

There was just an article by people who advocate AI therapy who were saying that AI therapists are actually doing a very meaningful job, and that never would have happened a year ago. So I’d say the change in people’s attitudes is pretty phenomenal.

Is the pace of change currently faster than you predicted? Because it feels faster. This is to Dave’s point earlier—it feels like we’re moving faster than you predicted. Do you agree or not agree?

In 1999, I predicted 2029 for AGI, and I still predict 2029. I think Elon Musk says 2026. I think we’ll have a lot of things that remind us of AGI, but we really won’t be convinced in 2026.

Maybe it’ll happen sooner—2027 or 2028. You get varying degrees of confidence, but by 2029, I think everyone will accept that.

Amazing. Amazing. Alex, I want to turn it back to you, pal.

Maybe to shift gears a bit, Ray, I’m obviously—if this isn’t obvious from some of my questions and comments—an enormous fan of both you and your writings, and your courageous extrapolation of following Wright's Law, of progress in experience curves, progress in Moore’s Law-type experience curves, your law of accelerating returns, your countdown to the singularity—all arguably variants on various forms of experience curves from economics.

Question for you: If we follow to its logical conclusion the law of accelerating returns and your countdown to the singularity, this idea that we’re almost, in a technologically deterministic way, emerging from a primordial soup and everything follows some very nice, elegant law of straight-line exponential calculus, do you think that this implies that our universe is abundant with intelligent civilizations?

And if so—in other words, abundant not just with human intelligence, but nonhuman intelligence as well—do you think that would then imply that there are nonhuman intelligent civilizations on or near Earth?

The fact that we can emerge as a far more intelligent version of ourselves in a short period of time doesn’t imply that there are intelligences that go beyond humans. We haven’t really seen evidence of that.

There’s a lot of interest in trying to find signals in the universe that would indicate that there’s some intelligent source behind them. We haven’t actually found that yet, and we have more and more ability to look.

So it may exist, but we don’t know that there’s any intelligence besides intelligence coming from Earth. Our having more and more ability to actually evaluate different types of intelligent sources that are not coming from Earth, and yet still not seeing any evidence of that, kind of indicates that they aren’t there.

But there’s no way of actually determining that because we can only look at a very small fraction of what’s out there.

Switching gears, Alex, do you want to do a follow-up?

Maybe just a quick follow-up question. So, Ray, you’ve made many, many predictions of technologies that you think either the singularity itself or progress toward the singularity would unlock. Do you think that progress toward the singularity would answer the question that I think many people most want existentially an answer to, which is: Is humanity alone?

Yeah. So far, if we’re not alone, we’re still pretty lonely because we haven’t come into contact with any intelligent source aside from ourselves. There are fantastic things happening in the universe, and the universe goes on seemingly forever.

So it’s certainly possible that we’ll find something, and it’s impossible to rule that out, but so far we haven’t actually done that. We certainly feel alone because there’s nobody else we can point to.

We can’t point to some other star system saying, “Well, there’s a source coming from that that’s clearly intelligent, and we’d like to contact them.” We can’t even identify a thing like that so far.

I want to jump into the conversation a little bit about BCI, brain-computer interfaces, and our ability to uplevel our capabilities. I think when we talk about longevity, escape velocity, and potentially living well past 100 or for hundreds of years, what most people fear is getting there without cognitive clarity, without the ability to maintain their memories.

Of course, one of the technologies that would assist us with that, that you’ve spoken about, is the idea of high-bandwidth BCI—not the low, thin pipe that we currently do input and output through. I encourage everybody to go onto your favorite LLM and ask it to give you a list of all of Ray Kurzweil’s predictions that he’s accurately hit. It’s a very impressive list.

One of those predictions is that we’ll hit high-bandwidth BCI in the mid-2030s. Is that still your prediction? And I want to ask: What’s that going to feel like? I raised my hand and volunteered for one of the early BCI interfaces. What’s that going to feel like, and how do you think we’re going to achieve that?

It’s very hard to know how we’re going to react to things that haven’t happened yet. You could imagine this being something that we would welcome or something that we would be alarmed by.

The future hasn’t been written yet. The future could be terrible, or it could be fantastic. It’s really hard to give a prediction about that.

Ray, you described it, if I could: Once we have high-bandwidth BCI, you’ll have concepts emerge in your mind that are driven, if you will, by the cloud. Can you speak to that a little bit?

Well, that would be useful. I’m actually writing my autobiography and trying to remember things that happened when I was 3 and 4 years old, and I actually have a pretty good memory of that. But it could be better, and it would actually be helpful if I had AI to help me along with that.

Actually, wait—not just that, but are you using AI to go interview people that you interacted with when you were 3, 4, or 10 years old and get their sides of the story?

Well, it would have to have a lot of capability that it doesn’t have now to be able to generate a view of something that we don’t have now. So I’m using large language models a little bit to try to—but actually, my memory is not bad of things that happened a long time ago.

All right. When’s the biography coming out?

I can’t wait.

It’s about ready. It should be out within a year.

Yeah, I’ve had a chance to read it. It’s pretty amazing. The thing I’m really eager to read in that biography is the role of the futurist. You made all these really bold predictions, and I’m sure at the time everyone was like, “You’re a crackpot. You’re a crackpot.” I suspect by now everyone’s like, “Wow, what incredible foresight.”

I assume you’re at an all-time high now, but maybe not. The role of being a futurist is fraught with this hindsight bias, where you get 3 things wrong. The self-driving car is not out yet, our clothes are not made by nanotechnology, and computing isn’t done on biological systems. We don’t have DNA computers.

I’m getting less of that now. Before, if I would make a whole bunch of predictions and one of them was wrong, everybody would focus on that.

Yeah. Yeah.

But now people are more generous in their views of me.

To me, the most amazing thing is when you read, at the time, “Everyone’s going to have a computer in their pocket, in their clothes, and it’s going to be almost like an extension of their life.” At the time it sounded like nuts, and now everyone’s like, “Oh, that’s just an iPhone.”

Well, no, it’s not just an iPhone. It’s a total cultural phenomenon that’s changed our—it changes you much more than you ever know. If you go to a conference and there are several hundred people, every single person has a cell phone in their pocket.

Yes.

And it’s actually an extension of your mind.

It is. It is.

If you don’t have your cell phone, you left three-quarters of your mind.

And I’ll tell you what else. The headmaster of the school that my kids went to took all of the kids, I think in 7th grade or 6th grade, to an island without their phones for 3 days and said, “You have to learn to live without your phone.”

The new headmaster came in and said, “That’s inhumane. We can’t do this anymore.”

I think there’s a book called Lord of the Flies that was written about that.

Lord of the Flies. That’s funny. But it’s so innate, and we’re talking about 7th graders here, but it’s so attached to their mentality, their mind, their body, whatever, that they can’t.

We’re going to replace this. I mean, carrying around a physical object like this is difficult. Where do you put it? How do you not lose it?

He’s got 2 chips in his hand now. [laughter]

What do you think replaces the phone?

We’ll have something besides this. It’ll be something like virtual reality.

Yeah, that’s a good question.

Yeah.

So, basically, you look out and you can see a screen, and it will be interfacing with your computer, but it’ll be on all the time. You’ll be able to interact with it, and you won’t be carrying something around or leave it at your apartment.

Yeah.

Beyond that, it’ll actually go inside our nervous system and interact with our biological neurons.

I’ve got this thing now. We’re starting to record everything, basically. Peter’s got the wearable, and now we’ve got these omnidirectional mics. They’re the size of a credit card. You just throw them on the table, and everything that happens is not only recorded, but it’s assigned to whoever said it with these omnidirectional mics. They’re starting to pop up everywhere.

This year, Dave, at the Abundance Summit, we’re giving everybody 2 devices. One is a ring format that we talked about on one of the WTF episodes that Pebble is putting out, where you can quickly record a message, go to an LLM, and then we’re giving everybody something called Plaud. I guess I’m spoiling the secret for our abundance members.

Ray, can we talk about one of the concerns you raised earlier that people have, which is people’s attachment to their employment? Thoughts on the future of work? You’ve spoken eloquently about the need for universal basic income and even universal high income that Elon has spoken about. So, what are your thoughts on the future of work, and when do we start having UBI? Should people be worried about their future income?

Well, we relate having an income to having the means to deal with our financial system. But if we separate that, and you’re going to be able to deal with your financial needs without having a conventional job, that’s actually liberating.

Why do people retire? To me, retirement doesn’t make sense, because what I’m doing, I enjoy doing. But if you look at most jobs, people don’t like them so much that they want to be able to do them forever. It’s actually liberation not to have to do that and to find something within their means that gives them gratification without having to work in a way that’s unpleasant. We’re basically overcoming that.

You know, 79% of corporate employees do not find meaning in their work, so this might be an easier transition than most people think.

Yeah.

Do you think we’re going to develop UBI soon?

We’re going to have to do something that’s equivalent to it, because if people don’t have enough money, the economic system won’t work for anybody. I made a prediction at TED that we would develop UBI by the 2030s, and I think that’s still true.

I’m going to do a quick separate thing. Imagine you’re in a courtroom, right? The prosecutor is saying to you, “You have made absurdly accurate predictions for 30 years. We don’t believe you’re human.” How would you defend that? You actually feel to me like a time traveler from somewhere who’s popped in to deliver and inject into humanity all of these insights. It blows my mind that 60 times I’ve heard you speak, I’ve never not learned something. So, if I was the litigator, I’d say, “You must be a time-traveling something.” How would you defend against that?

Hopefully, I would appear enough like a human to convince people. Maybe that won’t be true in the future. You can’t really tell if someone’s a human or not a human, because they’ll still act human. Then I wouldn’t have a defense.

Alex, over to you, buddy.

I could say something mildly snarky about looking at Ray’s immunome to see if he’s been exposed to future diseases, as a way of determining whether he’s a time traveler or not. But instead, I’d like to shift gears, Ray, and maybe talk about the past and future of the nature of the mind.

One of the many striking performances, and I think just incredible accomplishments, of yours goes all the way back—this is more than 60 years back now—to your appearance on I’ve Got a Secret on television with Steve Allen in February of 1965. It’s incredible to think that that was 60-plus years ago. You demonstrated an AI-based music generator on television.

Yeah, that was actually the first music composition by AI anywhere.

We should show that clip.

Absolutely. We should show that clip. It’s such an incredible accomplishment.

[music]

Very nicely played. And now, your performance, of course, leads into your secret. So, if you’ll whisper it to me, we’ll let everybody at home know what’s up.

Well, that certainly deserves applause, but it’s all the subways leaving that deserves applause. But what has it got to do with the music? I don’t understand that. Ah, I see. Ray’s secret concerns something that he did. And we’ll start the game this time with Bess Myerson.

Ray, that’s a very unlikely-sounding piece of music. Am I being supercritical?

No.

Did you compose it?

No, I didn’t.

Oh. Did you, however, use some kind of formulas or letters or something unusual to compose, to make up the notes of this piece?

You could say that, I guess.

Mm-hmm. Well, for example, would the notes spell out a name, or would they be a mathematical formula or anything like that?

Not spell out a name. Nothing like that, ma’am.

$20 down, $60 to go. Henry.

Was that thing written by a computer?

Wow. Is it writing music at this moment?

Right now, it’s writing.

Writing tones. I have a feeling that, as a nonscientist, I’m not going to understand this too well. But perhaps you can explain how it works. First of all, I want the folks to see some of this. This nest of spaghetti-like wire here is connected to a bunch of little whatnots. What are these black things over here, Ray?

Well, those are relays. That’s what does the trick. That’s what writes the music.

I see. The relays write the music. They feed it into this white cheese box here, whatever that is. And there are 3 little—are these wires or just pieces of string?

Pieces of string or wires?

I mean, does the message go through there, or are they just—

No, that’s just recording what music the computer says.

I see. And then the typewriter does the final part of the process.

Right.

So, 60 years ago, you demonstrated what I understand to be the first, at least on television, AI music generator. I’d like to ask you now, 60-plus years from now. We’re now in 2026, so we’re talking 2086. What form do you think most intelligence in our solar system will take?

I’ll offer you a few options, and I’ll deny you one option. The option that I’ll deny you is that you’re not allowed to say, “It’s past the singularity, so I have no idea.” You have to condition on having a real opinion on this topic.

I’ll offer you a few options and an escape valve for maybe something that I haven’t thought of.

Say the question again, Alex.

Yes. The question is, again, 60 years from now, in the year 2086, what form will most intelligence in our solar system take?

A few options: meat bodies substantially similar to the way human intelligence is embodied now. That’s option 1. Cyborgs, which are some sort of human-machine hybrid, inclusive of nanorobots in the human bloodstream. Uploads—that’s option 3—so human minds have been uploaded to the cloud. Foundation models or pure AIs, not dissimilar to GPT-type models that we have right now.

Some sort of unrecognizable life form, maybe an unrecognizable arrangement of matter or energy that’s far more efficient. In the past on the podcast, I’ve talked about how black holes, for example, are amazing computers in principle. So maybe something like that, or something totally different—maybe uploads to the gravitational field or something else entirely. I’m laying out a few options, plus an escape valve.

What do you think?

We’re going to have things like computronium by certainly 2045, if not sooner, and I know people who are working on this.

You want to define computronium, Ray?

Yeah. It’s basically taking what we know is feasible and creating something out of matter that can perform the maximum computation that we can conceive of. One analysis has basically 1 liter cube being more intelligent than all people—like 10 billion people combined—in one setting.

That’s going to be happening by 2045. So, you talk about 2086; it’s going to be beyond what we can imagine, but it’ll be even more so. We’ll be able to create something that’s very exciting.

If I listen to, let’s say, some of the things I have on the web that go with the book—my father playing the Fifth Brandenburg Concerto, which was done several hundred years ago by Bach—it’s actually quite amazing to listen to that. It’ll be something like that, only more fantastic, that will generate fantastic emotions and will be as intelligent as all people combined, or more so. We really can’t imagine what that would be like, but we can state it mathematically by comparing it to what we can do today.

If I may ask a follow-up question on this: It sounds, Ray, unless I’m misunderstanding, that you do, in fact, have a prediction for what most intelligence would look like. Namely, if I heard correctly, you think that in 60 years most intelligence in the solar system will be basically software running on computronium. I think you referenced some work by Seth Lloyd, with the reference to 1 liter of computronium, and Seth Lloyd’s work from 25 years ago on the ultimate computer and the physical limit of the maximum amount of computation.

Since that’s going to be feasible well before 2086, any kind of intelligent being is going to contain that. Then I have to ask you, I guess, the obvious question: If you think 60 years from now most intelligence in our solar system is going to be software running on computronium, what happens to our solar system? Do we disassemble the planets? Do we starlift our Sun? Do we convert our solar system to computronium to run the software?

Yes. So what it’ll be is even beyond that, but certainly that will be the capability that it will have.

Saturn had it coming.

Saturn.

Yeah. Actually, I think Ray is right about it in this instance.

I don’t know. We’ll have to think about that.

But the point, Alex and Ray, that you’re both making is that humanity as we know it today, as biological forms, is either in the vast minority or absolutely displaced by a digital or quantum version of intelligence. So will some people choose to maintain an enhanced meat body, or are the overwhelming benefits of going digital so much that it will wash away all previous versions?

Well, I didn’t say the meat bodies would go away. But certainly it will have the capability of computronium running the ultimate software, certainly by 2086.

Since you’ve been inside Google for so long, and Google is really kind of like AT&T Bell Labs or universities times 1,000, this computronium shift—back in your early books, you made the point that Moore’s law isn’t really Moore’s law. It goes back—you go all the way back to switches, telecom switches, then vacuum tubes, then transistors, then integrated circuits.

So there’s been a shift in the compute platform that keeps this curve going. But now we’re at this stage where we’re just pushing silicon to its limit and scaling horizontally, with half a trillion dollars we’re going to put into NVIDIA chips. So we’re kind of at this flat spot, waiting for that next breakthrough in how we compute. Is there anything imminent, anything that’s going to fill that gap? And I know AI will help us innovate very quickly here.

Well, it’s a different issue, but I think we’ll actually generate slower computational bodies. If you look at the brain, it uses about 2 watts of power, and that’s because it’s very, very slow. Our neurons compute between 1 calculation per second and about 200 calculations per second.

But both of those are extremely slow compared to the millions or billions—or actually trillions—of computations per second that current computers are capable of. What I wrote about a couple of decades ago was that it would make sense to slow it down and introduce parallel processing, because every single neuron in the brain is computing at the same time.

Twenty years ago, basically, a computer would do one thing at a time. So we actually have done that. We now have millions—or actually billions—of computations that occur at the same time, but we actually haven’t slowed down the speed of the circuits. If we slow them down a little bit, we’d use much less power, and I think that would actually solve the power problem.

Well, so it solves the chip fab bottleneck problem. I think there’s imminent innovation in exactly that vein you’re talking about. So that buys you another few years, but it doesn’t switch you to a new computronium kind of paradigm.

I know you were kind of like, quantum isn’t really going to change the curve here. I don’t know if you still feel that way about quantum computing, but is there anything else on the horizon that you know of, from either inside Google or elsewhere?

Well, I think going toward circuits that use a completely different paradigm, that are actually done at the molecular level and can be done in 3 dimensions. Right now we’re using the 3rd dimension in a very limited way. So we can actually create 3-dimensional circuits at the atomic level, where 1 liter of computronium will match 10 billion human beings.

Salim, when you look at what’s coming over the next year, is there anything that you’re incredibly excited about? One of the things I’ve heard you talk about is the intersection between these—you intersect synthetic biology or neuroscience with AI and computing, and all sorts of new fields get instigated by that. What is most exciting to you, and what are you anticipating most excitedly in the next year or 2?

Well, robotics has not really been something that has affected us very much. I think that’s going to begin to take place in 2026 and 2027. But you look at robots—I mean, they can do certain things, like a very fast dance, but they really have not been practical.

If you eat a meal and leave your dishes, there’s no robot that can actually pick them up and clean them up the way a human being can. That’s going to happen over the next couple of years. So that’s one area that’s been behind, and I think there’s going to be a lot of debate on that.

Large language models are pretty fantastic, but we’ve got to bring that to the real world of actually being able to handle physical things using robots.

Salim, you had some questions, I think, on society that were important.

Yeah. If you were advising a 25-year-old today, how would you set about giving them a sense of how to manage their life in this radical uncertainty? How would you advise them to think? What mindset should they have, et cetera? What advice would you give to a 25-year-old today?

My son Ethan is involved with venture capital, and most, if not all, of his investments are in AI—actually bringing the practice of AI to all kinds of things that haven’t been done yet. There are a tremendous number of opportunities to apply AI to all kinds of things that we do and create businesses that would be effective.

So I think the opportunities to create a new business and do things that have not been done before are higher than they’ve ever been before.

You talk a lot about entrepreneurship being really the biggest modality you could go after. I think there’s a great comment by Kevin Kelly, where he said, “The next 10,000 business plans will be: Take a domain and add AI to it.”

Yeah, Ray, do you ever feel like you were just born in the wrong era? If you think about what you did early on with the keyboard, the company around it, then the omni-font optical character recognition, the same person today would probably be looking at a billion-dollar valuation within a year, a year and a half of founding.

Well, I enjoyed bringing some of the concepts that we use today in decades past, though.

Let’s do a quick speed round to close out this session with Ray. Alex, you want to kick it off?

All right, Ray, here’s a really fast question. The cliché is that every American male thinks about ancient Rome at least once per day. So here’s my cliché question for you.

Really?

We’re going to go there. The question, Ray, is: Why didn’t ancient Rome have an Industrial Revolution? And what does the answer to that question teach us about technical revolutions that we could be having today but otherwise aren’t?

Well, they did have a technical revolution, given the capabilities of that time. We can only create things that are feasible and in keeping with the rate of progress that was feasible at that time. So I think they did okay.

I feel like I’m seeing the passing of the torch of the futurist here, from Ray to Peter to Alex. But I’m really curious if you’re happy with your life as a great futurist, because you were already a great entrepreneur before that.

There were many, many years in the middle there where everyone I talked to around MIT or elsewhere was like, “Yeah, I think Ray’s wrong. I think Ray’s wrong. I think Ray’s wrong.” Now, obviously, you’re on top of the world again, but there are a lot of years of just the pain and suffering that goes along with anyone who tries to predict the future. Any regrets? Any advice for future futurists?

I got used to it. There were certain people who were able to think in the future. For example, Singularity University, which Peter and I started, could think about how to go beyond what conventional people were thinking. But it didn’t really bother me that people were not able to think in an exponential manner at the time.

Really? Thanks again. Okay.

See, the fact that it didn’t bother you is why I think you’re a time-traveling avatar from the future. Here’s my question: If right now you’ve said that intelligence and energy are the 2 things that will become abundant in the future...

It seems right now that energy is the limiting factor. Are you excited about what’s coming with nuclear and fusion, or are there other forms of energy generation that you’re looking at? When do you think we’ll have a major breakthrough around some of that?

I’m not that enthusiastic about nuclear. I still think it’s dangerous.

There are two things we can do about energy. We can use reversible computing, which most of the computation would use. In theory, it uses no energy at all because it reverses itself and gives back the energy that it took. We haven’t actually experimented with that, but that seems feasible.

I also mentioned the other thing, where we could dramatically reduce the speed and the amount of energy it requires, and therefore overcome the excessive use of energy. Right now, we’re running things at the very maximum speed, and it uses a great deal of energy. We could reduce that a little and really overcome the energy use at that point. But ultimately, we will go to reversible computing, using computation at atomic levels, which don’t require any energy, at least in theory.

Ray, I want to take a second and say thank you for the extraordinary partnership we’ve had over these last number of decades. I remember our first lunch together, where we kicked around the idea of Singularity University, and I think you waited a nanosecond before saying yes. It’s been a great joy, and a shout-out to all the Singularity alumni out there who have been part of this journey.

The Singularity Is Near has sort of been our mantra and our war cry here.

On a 10-point scale, how optimistic are you about the future of humanity?

I’d say I’m a 10.

All right. Well, that’s a good place to wrap it up. Ray, on behalf of the Moonshot Mates, thank you for all of your wisdom. Thank you for charting the path for us.

Yeah. Well, this was a great discussion. I appreciate it very much.

Appreciate it.

Wait for the biography, too. Everybody keep an eye out for that.

I look forward to seeing you in May for our follow-on book launch event. Dave, safe travels to the World Economic Forum. I’ll see you—I’ll come and pick you up and see you in an hour. Then we head to the XPRIZE board meeting. Alex, enjoy Paris and Switzerland.

Yeah.

Amazing. All right, guys.

See you all.