Sundar Pichai：Google 与 Alphabet CEO｜Lex Fridman Podcast #471

There was a 5-year waiting list, and we got a rotary telephone, but it dramatically changed our lives. People would come to our house to make calls to their loved ones. I would have to go all the way to the hospital to get blood test records, and it would take 2 hours to go. They would say, “Sorry, it’s not ready. Come back the next day.” Then it was 2 hours to come back. And that became a 5-minute thing.

As a kid, this light bulb went in my head: the power of technology to change people’s lives. We had no running water; it was a massive drought. They would get water in these trucks, maybe 8 buckets per household. So my brother and I, sometimes my mom, would wait in line, get that, and bring it back home. Many years later, we had running water and a water heater, and you could get hot water to take a shower. For me, everything was distinct like that. I’ve always had this firsthand feeling of how technology can dramatically change your life and the opportunity it brings.

I think if p(doom) is actually high at some point, all of humanity is aligned in making sure that’s not the case, right? And so we’ll actually make more progress against it, I think. The irony is, there is a self-modulating aspect there. I think if humanity collectively puts their mind to solving a problem, whatever it is, I think we can get there. Because of that, I think I’m optimistic on the p(doom) scenarios. But that doesn’t mean I think the underlying risk is actually pretty high. I have a lot of faith in humanity rising up to meet that moment.

Take me through that experience when there are all these articles saying you’re the wrong guy to lead Google through this. Google is lost. It’s done. It’s over.

The following is a conversation with Sundar Pachai, the CEO of Google and Alphabet on this the Lex Freedman podcast.

Your life story is inspiring to a lot of people. It’s inspiring to me. You grew up in India, your whole family living in a humble 2-room apartment, with very little—almost no—access to technology. From those humble beginnings, you rose to lead a $2 trillion technology company. If you could travel back in time and tell that, let’s say, 12-year-old Sundar that you’re now leading 1 of the largest companies in human history, what do you think that young kid would say?

I would have probably laughed it off. Probably too far-fetched to imagine or believe at that time. You would have to explain the internet first, for sure. Computers, to me, at that time—I was 12 in 1984. By then, I had started reading about them. I had seen one.

What was that place like? Take me to your childhood.

I grew up in Chennai. It’s in the south of India. It’s a beautiful, bustling city. Lots of people, lots of energy, simple life. I have fond memories of playing cricket outside the home. We used to play on the streets. All the neighborhood kids would come out, and we would play till it got dark and we couldn’t play anymore. Barefoot. Traffic would come; we would just stop the game, everything would drive through, and then you would just continue playing. Just to get the visual in your head.

Before computers, there was a lot of free time. Now that I think about it, now you have to go and seek that quiet solitude or something. Newspapers and books are how I gained access to the vast information at the time.

My grandfather was a big influence. He worked in the post office. He was so good with language. His English, his handwriting—till today, it’s the most beautiful handwriting I’ve ever seen. He would write so clearly. He was so articulate, and he introduced me to books. He loved politics, so we could talk about anything. That was there in my family throughout.

There were lots of books—trashy books, good books—everything from Ayn Rand to books on philosophy to stupid crime novels. Books were a big part of my life. It’s not surprising I ended up at Google, because Google’s mission always resonated deeply with me: this access to knowledge. I was hungry for it.

I definitely have fond memories of my childhood. Access to knowledge was there, so that was the wealth we had. Every aspect of technology I had to wait for a while. I’ve obviously spoken before about how long it took for us to get a phone—about 5 years—but it’s not the only thing.

There was a 5-year waiting list, and we got a rotary telephone, but it dramatically changed our lives. People would come to our house to make calls to their loved ones. I would have to go all the way to the hospital to get blood test records, and it would take 2 hours to go. They would say, “Sorry, it’s not ready. Come back the next day.” Then it was 2 hours to come back. And that became a 5-minute thing.

As a kid, this light bulb went in my head: the power of technology to change people’s lives. We had no running water; it was a massive drought. They would get water in these trucks, maybe 8 buckets per household. So my brother and I, sometimes my mom, would wait in line, get that, and bring it back home. Many years later, we had running water and a water heater, and you could get hot water to take a shower. For me, everything was distinct like that. I’ve always had this firsthand feeling of how technology can dramatically change your life and the opportunity it brings. That was kind of a subliminal takeaway for me throughout growing up. I observed it and felt it.

We had to convince my dad for a long time to get a VCR. Do you know what a VCR is?

Yeah.

I’m trying to date you now. Before that, you only had 1 TV channel. That’s it. You could watch movies or something like that, but this was by the time I was in 12th grade, when we got a VCR. It was a Panasonic, which we had to go to some shop that had smuggled it in, I guess. That’s where we bought a VCR.

Being able to record a World Cup football game, or put in videotapes and watch movies—all of that left me with these distinct memories growing up. It always left me with the feeling of how getting access to technology drives that step change in your life.

I don’t think you’ll ever be able to equal the first time you get hot water, to have that convenience of going and opening a tap and having hot water come out. Yeah, it’s interesting. We take for granted the progress we’ve made.

If you look at human history, just those plots that look at GDP across 2,000 years, you see that exponential growth, where most of the progress happened since the Industrial Revolution. We just take it for granted. We forget how far we’ve gone. So our ability to understand how great we have it, and also how quickly technology can improve, is quite poor.

It’s extraordinary. I go back to India now. The power of mobile—it’s mind-blowing to see the progress through the arc of time. It’s phenomenal.

What advice would you give to young folks listening to this all over the world who look up to you and find your story inspiring, who want to be maybe the next Pichai, who want to start creating companies and build something that has a lot of impact in the world?

Look, you have a lot of luck along the way, but you obviously have to make smart choices. When you’re thinking about what you want to do, your brain is telling you something. But when you do things, I think it’s important to listen to your heart and see whether you actually enjoy doing it. That feeling of, if you love what you do, it’s so much easier, and you’re going to see the best version of yourself.

It’s easier said than done. I think it’s tough to find things you love doing. But I think listening to your heart a bit more than your mind, in terms of figuring out what you want to do, is 1 of the best things I would tell people.

The second thing is trying to work with people who you feel are better than you. At various points in my life, I worked with people who I felt were better than me. You almost are sitting in a room talking to someone, and they’re like, “Wow.” You want that feeling a few times.

Trying to get yourself in a position where you’re working with people who you feel are stretching your abilities is what helps you grow, I think. So putting yourself in uncomfortable situations—I think often you’ll surprise yourself. Being open-minded enough to put yourself in those positions is maybe another thing I would say.

What lessons can we learn, maybe from an outsider perspective? For me, looking at your story and having gotten to know you a bit, you’re humble, you’re kind. Usually, when I think of somebody who has had a journey like yours and climbed to the very top of leadership in a cutthroat world, they’re usually going to be a bit of an—

So what wisdom are we supposed to draw from the fact that your general approach is one of balance, humility, kindness, and listening to everybody? What’s your secret?

I do get angry. I do get frustrated. I have the same emotions all of us do, in the context of work and everything. But a few things: I think, over time, I figured out the best way to get the most out of people.

You find mission-oriented people who are on this shared journey, who have this inner drive to excellence, to do the best, and you motivate people, and you can achieve a lot that way. It often tends to work out that way.

But have there been times when I lose it? Yeah. Maybe less often than others, and maybe over the years less and less so, because I find it’s not needed to achieve what you need to do.

So losing your temper has not been productive.

Yeah. Less often than not, I think people respond to that. They may do things to react to that, but you actually want them to do the right thing. Maybe there’s a bit of sports. I’m a sports fan. In football coaches—in soccer, that’s football—people often talk about man-management, right? Coaches do, right? I think there is an element of that in our lives: how do you get the best out of the people you work with?

At times, you’re working with people who are so committed to achieving that, if they’ve done something wrong, they feel it more than you do, right? So you treat them differently. Occasionally, there are people who you need to clearly let know, “That wasn’t okay,” or whatever it is, but I’ve often found that not to be the case. Sometimes the right words at the right time, spoken firmly, can reverberate through time.

Also, sometimes the unspoken words—you know, people can sometimes see that you're unhappy without you saying it. Sometimes the silence can deliver that message even more. Sometimes less is more.

Who's the greatest soccer player of all time: Messi, Ronaldo, Pelé, or Maradona? I'm going to make you answer. Is this going to be a political answer?

No, I will tell the truthful answer. It's been interesting because my son is a big Cristiano Ronaldo fan, and we've had to watch El Clásicos together with that dynamic in there. I so admire CR7. I've never seen an athlete more committed to that kind of excellence. He's one of the all-time greats, but for me, Messi is it.

Yeah. When I see Lionel Messi, you just are in awe that humans are able to achieve that level of greatness, genius, and artistry. When we talk about AI, maybe robotics and this kind of stuff, that level of genius—I'm not sure you can possibly match it with AI for a long time. It's just an example of greatness.

You have that kind of greatness in other disciplines, but in sport, you get to visually see it unlike anything else. The timing, the movement—this is genius. I had the chance to see him a couple of weeks ago. He played in San Jose against the Quakes, so I went to see the game. I was a fan and had good seats. I knew where he would play in the second half, hopefully. Even at his age, just watching him when he gets the ball—that movement, you're right, that special quality, it's tough to describe, but you feel it when you see it.

Yeah, he still has it.

If we rank all the technological innovations throughout human history—let's go back to the history of human civilizations 12,000 years ago—and rank them by how much of a productivity multiplier they've been, we can go to electricity or the labor mechanization of the Industrial Revolution, or we can go back to the first agricultural revolution 12,000 years ago. In that long list of inventions, do you think AI, when history is written 1,000 years from now, has a chance to be the number-one productivity multiplier?

It's a great question. Look, many years ago, I think it might have been 2017 or 2018, I said at the time, "AI is the most profound technology humanity will ever work on. It'll be more profound than fire or electricity." So I have to back myself. I still think that's the case.

When you ask this question, I was thinking, well, do we have a recency bias? Like in sports, it's very tempting to call the current person you're seeing the greatest player, right? Is there a recency bias? From first principles, I would argue AI will be bigger than all of those.

I didn't live through those moments. Two years ago, I had to go through a surgery, and then I processed that there was a point in time when people didn't have anesthesia when they went through these procedures. At that moment, I was like, "That has got to be the greatest invention humanity has ever done," right?

We don't know what it was like to have lived through those times. Many of what you're talking about were these general things that pretty much affected everything—electricity or the internet, et cetera. But I don't think we've ever dealt with a technology that is both progressing so fast and becoming so capable. It's not clear what the ceiling is, and the main unique thing is that it's recursively self-improving, right? It's capable of that.

The fact that it's going—it's the first technology that will dramatically accelerate creation itself, creating things and building new things, and can improve and achieve things on its own—I think puts it in a different league. I think the impact it'll end up having will far surpass everything we've seen before. Obviously, with that comes a lot of important things to think about and wrestle with, but I definitely think that'll end up being the case, especially if it gets to the point where we can achieve superhuman performance on the AI research itself.

So it's a technology that may—that's an open question—but it may be able to achieve a level where the technology itself can create itself better than it could yesterday. It's like Move 37 of AlphaGo, or whatever it is, right? When it can do novel, self-directed research, obviously, for a long time, we'll hopefully always have humans in the loop and all that stuff. These are complex questions to talk about, but yes, I think the underlying technology—I've said this—if you watched AlphaGo start from scratch, be clueless, and become better through the course of a day, it really hits you when you see that happen.

Even our Veo 3 models, if you sampled the models when they were 30% done and 60% done and looked at what they were generating, you kind of see how it all comes together. It's inspiring and a little bit unsettling as a human. So all of that is true, I think.

The interesting thing about the Industrial Revolution, electricity, like you mentioned—you can go back again to the first agricultural revolution. There's what's called the Neolithic package of the first agricultural revolution. It wasn't just that the nomads settled down and started planting food; all these other kinds of technologies were born from that and are included in this package. It wasn't 1 piece of technology. There are these ripple effects, second- and third-order effects that happen, everything from something silly like pottery that can store liquids and food to something profound like social hierarchies and political hierarchy.

Early government was formed because it turns out that if humans stop moving and have some surplus food, they get bored and start coming up with interesting systems. Then trade emerges, which turns out to be a really profound thing, and, like I said, government. There are just second- and third-order effects from that. That package is incredible and probably extremely difficult to predict. If you asked one of the people in the nomadic tribes to predict that, it would be impossible. It's difficult to predict.

All that said, what do you think are some of the early things we might see in the so-called AI package?

I mean, most of it we probably don't know today, but the one thing we can tangibly start seeing now is, obviously, with the coding progress, you've got a sense of it. It's going to be so easy to imagine thoughts in your head translating into things that exist. That'll be part of the package, right? It's going to empower almost all of humanity to express themselves.

Maybe in the past you could have expressed yourself with words, but you could kind of build things into existence, right? Maybe not fully today. We are at the early stages of vibe coding. I've been amazed at what people have put out online with Veo 3, but it takes a bit of work, right? You have to stitch together a set of prompts. But all this is going to get better. The thing I always think about is, this is the worst it'll ever be, right? At any given moment in time.

Yeah, it's interesting you went there as kind of a first thought. So the exponential increase of access to creativity and software creation—whether you're creating a program, a piece of content to be shared with others, or games down the line—all of that just becomes infinitely more possible.

Well, I think the big thing is that it makes it accessible. It unlocks the cognitive capabilities of the entire 8 billion.

No, I agree. Look, think about 40 years ago. Maybe in the U.S. there were 5 people who could do what you were doing—go do an interview, you know? But today, think about YouTube and other products, et cetera. How many more people are doing it?

I think this is what technology does, right? When the internet created blogs, you heard from so many more people. But with AI, I think that number won't be in the few hundreds of thousands. It'll be tens of millions of people, maybe even 1 billion people, putting out things into the world in a deeper way.

I think it'll change the landscape of creativity and make a lot of people nervous. For example, Fox, MSNBC, and CNN are really nervous about this part. You mean, this dude who could just do this and use YouTube—and thousands of others, tens of thousands, millions of other creators—can do the same kind of thing.

That makes him nervous. And now you get a podcast from NotebookLM that's about 5 to 10 times better than any podcast I've ever done. True. I'm joking at this time, but maybe not.

And that changes. You have to evolve because, on the podcasting front, I'm a fan of podcasts much more than I am a fan of being a host or whatever. If there's a great podcast that's by AIs, I'll just stop doing this podcast. I'll listen to that podcast. But you have to evolve and you have to change, and that makes people really nervous, I think. But it's also a really exciting future.

The only thing I might say is, in a world in which there are 2 AIs, I think people value and choose. Just like in chess, you and I would never watch Stockfish 10 or whatever and AlphaGo play against each other. It would be boring for us to watch, but Magnus Carlsen and Gukesh playing that game would be much more fascinating to watch.

So it's tough to say. One way to say it is you'll have a lot more content, and so you will be listening to AI-generated content because sometimes it's efficient. But the premium experiences you value might be a version of the human essence, wherever it comes through, going back to what we talked about earlier about watching Messi dribble the ball.

I don't know. One day, I'm sure a machine will dribble much better than Messi, but I don't know whether it would evoke that same emotion in us. So I think that'll be fascinating to see. I think the element of podcasting or audiobooks that's about information gathering might be removed, or that might be done more efficiently and in a compelling way by AI.

But then it'll be nice to hear humans struggle with the information, contend with the information, try to internalize it, combine it with the complexity of our own emotions and consciousness, and all that kind of stuff. If you actually want to find out about a piece of history, you go to Gemini. If you want to see Lex struggle with that history, then you look at that—or other humans, you look at that.

The point is, it's going to continue to change the nature of how we discover information, how we consume information, and how we create that information. The same way that YouTube changed everything completely and changed news—it changed everything—and that's something our society is struggling with.

Yeah. YouTube—look, YouTube enabled so many creators. You know this better than anyone else. There's no doubt in my mind that we will enable more filmmakers than there have ever been, right? You're going to empower a lot more people.

So I think there's an expansionary aspect of this which is underestimated. I think it'll unleash human creativity in a way that hasn't been seen before. It's tough to internalize. The only way is if you brought someone from the 1950s or 1940s and just put them in front of YouTube. I think it would blow their mind.

Similarly, I think we would be blown away by what's possible in a 10- to 20-year time frame.

Do you think there's a future—how many years out is it, let's put a marker on it—until 50% of compelling, good content, 50% of good content, is generated by Veo 4, 5, or 6?

You know, I think it depends on what it's for. Maybe if you look at movies today with CGI, there are great filmmakers. You still look at who the directors are and who uses it. There are filmmakers who don't use it at all. You value that. There are people who use it incredibly.

Think about somebody like James Cameron and what he would do with these tools in his hands. But I think there'll be a lot more content created. Just like writers today use Google Docs and don't think about the fact that they're using a tool like that, people will be using future versions of these things. It won't be a big deal at all to them.

I've gotten a chance to get to know Darren Aronofsky well. He's been really leaning in and trying to figure out how AI can be used to create compelling films. It's fun to watch a genius who came up before any of this was even remotely possible. He created Pi, one of my favorite movies, and from there just continued to create a really interesting variety of movies.

You have people like that. You have people I've gotten to know who are AI-first, like the Duffer Brothers. Both Aronofsky and the Duffer Brothers create at the edge of the Overton window of society. They push, whether it's sexuality or violence. It's edgy like artists are, but it's still classy. It doesn't cross that line—whatever that line is.

Hunter S. Thompson has this line: the only way to find out where the edge, where the line is, is by crossing it. I think for artists that's true. That's kind of their purpose. Sometimes comedians and artists just cross that line.

I wonder if you can comment on the weird place that puts Google, because Google's line is probably different from some of these artists. What do you think specifically about Veo and Flow, and how to allow artists to do something crazy, but also about the responsibility of it not being too crazy?

It's a great question. Look, you mentioned Darren. He's a clear visionary, right? Part of the reason we started working with him early on with Veo is that he's one of those people who was able to see that future, get inspired by it, and show the way for how creative people can express themselves with it.

Look, I think when it comes to allowing artistic free expression, that's one of the most important values in a society, right? Artists have always been the ones to push boundaries and expand the frontiers of thought. I think that's going to be an important value we have.

We will provide tools and put them in the hands of artists for them to use and put out their work. Those APIs—I almost think of that as infrastructure. Just like when you provide electricity to people or something, you want them to use it, and you're not thinking about the use cases on top of it.

So it's a paintbrush.

Yeah. And so I think that's how it is. Obviously, there have to be some things, and society needs to decide at a fundamental level what's okay and what's not. We'll be responsible with it. But I do think when it comes to artistic free expression, that's one of those values we should work hard to defend.

I wonder if you can comment on maybe earlier versions of Gemini being a little bit careful about the kinds of things you would be willing to answer. I was really surprised—and pleasantly surprised—and enjoyed the fact that Gemini 2.5 Pro is a lot less careful, in a good sense.

Don't ask me why, but I've been doing a lot of research on Genghis Khan and the Aztecs, so there's a lot of violence there in that history. It's a very violent history. I've also been doing a lot of research on World War I and World War II. Earlier versions of Gemini were very much like, “Are you sure you want to learn about this?”

Now it's actually very factual and objective. It talks about very difficult parts of human history and does so with nuance and depth. It's been really nice. But there's a line there that I guess Google has to walk, and it's also an engineering challenge: how to do that at scale across all the weird queries that people ask.

Can you speak to that challenge? How do you allow Gemini to be, again, forgive—pardon my French—crazy, but not too crazy?

I think one of the good insights here has been that as the models are getting more capable, the models are really good at this stuff, right? In some ways, maybe a year ago, the models weren't fully there, so they would also do stupid things more often. You were trying to handle those edge cases, but then you would make a mistake in how you handled those edge cases and it would compound.

But I think with 2.5, what we particularly found is that once the models cross a certain level of intelligence and sophistication, they're able to reason through these nuanced issues pretty well. I think users really want that, right? You want as much access to the raw model as possible.

I think it's a great area to think about. Over time, we should allow closer and closer access to it, maybe obviously let people customize prompts if they wanted to, and experiment with it. I think that's an important direction. But the first principle we want to think about is, from a scientific standpoint—I'm saying “scientific” in the sense of how you would approach math or physics or something like that—having the models reason about the world and be nuanced from the ground up is the right way to build these things, right?

It's not like some subset of humans is hard-coding things on top of it. I think it's the direction we've been taking, and I think you'll see us continue to push in that direction.

Yeah. I actually asked—I took extensive notes and gave them to Gemini—and said, “Can you ask a novel question that's not in these notes?” Gemini continues to really surprise me. It's been really beautiful. It's an incredible model.

The question it generated was: “You, meaning Sundar, told the world Gemini is turning out 480 trillion tokens a month. What's the most life-changing 5-word sentence hiding in that vast stack?”

That's a Gemini question, but it gave me a sense—it woke me up to the fact that all of these tokens are providing little aha moments for people across the globe. Those tokens are people being curious.

They ask a question and they find something out, and it truly could be life-changing.

Oh, it is. Look, I know I had the same feeling about search many, many years ago. You definitely know this: tokens per month have grown 50 times in the last 12 months. Is that accurate, by the way?

Yeah, it is. It is accurate. I'm glad it got it right. But that number was 9.7 trillion tokens per month 12 months ago, right? It's gone up to 480 trillion. It's a 50x increase. So there's no limit to human curiosity. I think it's one of those moments. Maybe one day there's a five-word phrase which says what the actual universe is, or something like that, and something very meaningful. But I don't think we are quite there yet.

Do you think the scaling laws are holding strong? There are a lot of ways to describe the scaling laws for AI, but on the pre-training and post-training fronts? So, the flip side of that: Do you anticipate AI progress will hit a wall? Is there a wall?

It's a cherished micro-kitchen conversation. Once in a while, I have it, like when Demis is visiting, or Dennis, Koray, Jeff, Noam, Sergey—a bunch of our people. We sit and talk about this, right? Look, we see a lot of headroom ahead. I think we've been able to optimize and improve on all fronts: pre-training, post-training, test-time compute, and tool use.

Over time, we've been making these models more agentic, getting them to be more general world models. In that direction, like Veo 3, the physics understanding is dramatically better than what Veo 1 was. You see progress on all those dimensions. I feel progress is very obvious to see, and I feel like there is significant headroom. More importantly, I'm fortunate to work with some of the best researchers on the planet. They think there is more headroom to be had here, and so I think we have an exciting trajectory ahead.

It's tougher to say. Each year I sit and say, "Okay, we're going to throw 10x more compute over the course of next year at it," and ask, "Will we see progress?" Sitting here today, I feel like the year ahead will have a lot of progress.

Do you feel any limitations? Are you compute-limited, data-limited, idea-limited? Do you feel any of those limitations, or is it full steam ahead on all fronts?

I think it's compute-limited in this sense. Part of the reason you've seen us do Nano, Flash, and Pro models, but not an Ultra model, is that for each generation, we feel like we've been able to get the Pro model at, I don't know, 80% or 90% of Ultra's capability. But Ultra would be a lot slower and a lot more expensive to serve.

What we've been able to do is go to the next generation and make the next generation's Pro as good as the previous generation's Ultra, but be able to serve it in a way that it's fast and you can use it. So I do think scaling laws are working. But at any given time, the models we all use the most are maybe a few months behind the maximum capability we can deliver, because that won't be the fastest or easiest to use.

Also, that's in terms of intelligence. It becomes harder and harder to measure performance, in quotes, because you could argue Gemini Flash is much more impactful than Pro just because of the latency. It's super intelligent already. Sometimes latency is maybe more important than intelligence, especially when the intelligence is just a little bit less in Flash. It's still an incredibly smart model.

You have to now start measuring impact, and it feels like benchmarks are less and less capable of capturing the intelligence of models, the effectiveness of models, the usefulness, and the real-world usefulness of models. Another kitchen question: Lots of folks are talking about timelines for AGI or ASI—artificial superintelligence. AGI, loosely defined, is basically human-expert level in a lot of the main fields of pursuit for humans. Then ASI is what AGI becomes, presumably quickly, by being able to self-improve—becoming far superior in intelligence across all these disciplines than humans. When do you think we'll have AGI? Is 2030 a possibility?

There's one other term we should throw in there. I don't know who used it first. Maybe Karpathy. Have you heard of AJI? Artificial jagged intelligence. Sometimes it feels that way, right? There is progress, and you see what they can do, and then you can trivially find that they make numerical errors, or have trouble counting the Rs in “strawberry,” or whatever it is. Most models seem to trip up on those things.

Maybe we should throw that term in there. I feel like we are in the AJI phase, where there is dramatic progress, some things don't work well, but overall you're seeing lots of progress. But if your question is, will it happen by 2030? Look, we constantly move the line of what it means to be AGI. There are moments today, sitting in a Waymo on a San Francisco street with all the crowds and people and watching it work its way through, where I see glimpses of it. The car is sometimes kind of impatient, trying to work its way through.

Using Astra, like in Gemini Live, or asking questions about the world—“What's this skinny building doing in my neighborhood?”—“It's a streetlight, not a building”—you see glimpses. That's why I use the word AJI, because then you see things which obviously show we're far from AGI, too. So you have both experiences simultaneously happening to you.

I'll answer your question, but I'll also throw out this: I almost feel the term doesn't matter. What I know is that by 2030 there'll be such dramatic progress that we'll be dealing with the consequences of that progress—the positive externalities and the negative externalities that come with it—in a big way. So that I strongly feel. Whatever we may be arguing about the term, maybe Gemini can answer what that moment is in time in 2030, but I think the progress will be dramatic. That I believe in.

Will AI think it has reached AGI by 2030? I would say we will just fall short of that timeline. I think it'll take a bit longer. It's amazing that in the early days of Google DeepMind, in 2010, they talked about a 20-year time frame to achieve AGI. It's fascinating to see.

For me, the whole thing—seeing what Google Brain did in 2012 and when we acquired DeepMind in 2014, right close to where we're sitting—in 2012, Jeff Dean showed the image of when the neural networks could recognize a picture of a cat and identify it. This was the early version of Brain. We all talked about a couple of decades. I don't think we'll quite get there by 2030, so my sense is it's slightly after that.

But I would stress that it doesn't matter what that definition is, because you will have mind-blowing progress on many dimensions. Maybe AI can create videos. We have to figure out, as a society, how to create a system by which we all agree that something is AI-generated and disclose it in a certain way, because how do you distinguish reality otherwise?

Yeah, there are so many interesting things you said. First of all, just looking back at this recent history, which now feels like distant history, with Google Brain: That was before TensorFlow, before TensorFlow was made public and open-sourced. So the tooling matters, too, combined with GitHub's ability to share code.

Then you have the ideas of attention, Transformers, and diffusion. Now there might be a new idea that seems simple in retrospect but will change everything, and that could be post-training or inference-time innovations. I think shadcn tweeted that Google is just one great UI away from completely winning the AI race, meaning UI is a huge part of it—how that intelligence manifests itself.

I think Logan Kilpatrick likes to talk about this right now. It's an LLM, but when is it going to become a system where you're talking about shipping systems versus shipping the particular model?

Yeah, that matters, too. How the system manifests itself and how it presents itself to the world really, really matters. There are simple UI innovations which have changed the world, and I absolutely think we'll see a lot more progress in the next couple of years. I think AI itself is on a self-improving track for UI itself.

Today, we're constraining the models. The models can't quite express themselves in terms of the UI to people. If you think about it, we've kind of boxed them in that way. But given that these models can code, they should be able to write the best interfaces to express their ideas over time. That is an incredible idea.

The APIs are already open. You create a really nice agentic system that continuously improves the way you can talk to an AI.

Yeah. But a lot of that is the interface, and then, of course, the incredible multimodal aspect of the interface that Google has been pushing. These models are natively multimodal. They can easily take content from any format and put it in any format. They can write a good user interface. They probably understand your preferences better over time. All of this is the evolution ahead.

That goes back to where we started the conversation. I think there'll be dramatic evolutions in the years ahead. Maybe one more kitchen question: This even further ridiculous concept of P(doom). The philosophically minded folks in the AI community think about the probability that AGI, and then ASI, might destroy all of human civilization.

I would say my P(doom) is about 10%. Do you ever think about this kind of long-term threat of ASI, and what would your P(doom) be?

Look, for sure. I've been very excited about AI, but I've always felt this is a technology where we have to actively think about the risks and work very, very hard to harness it in a way that it all works out well.

On the P(doom) question, it won't surprise you to say that's probably another micro-kitchen conversation that pops up once in a while, right? Given how powerful the technology is, maybe stepping back, when you're running a large organization, if you can align the incentives of the organization, you can achieve pretty much anything, right? If you can get people all marching toward a goal in a very focused, mission-driven way, you can pretty much achieve anything.

But it's very tough to organize all of humanity that way. I think if P(doom) is actually high at some point, all of humanity is aligned in making sure that's not the case, right? And so we'll actually make more progress against it, I think.

The irony is, there is a self-modulating aspect there. I think if humanity collectively puts its mind to solving a problem, whatever it is, we can get there. Because of that, I think I'm optimistic on the P(doom) scenarios, but that doesn't mean I think the underlying risk is actually pretty high. I have a lot of faith in humanity rising up to meet that moment.

That's really well put. I mean, as the threat becomes more concrete and real, humans do really come together and get their act together.

The other thing I think people don't often talk about is the probability of doom without AI. There are all these other ways that humans can destroy themselves. It's very possible, at least I believe so, that AI will help us become smarter, kinder to each other, and more efficient. It'll help more parts of the world flourish where they would be less resource-constrained, which is often the source of military conflict and tensions.

So we also have to factor in what's the P(doom) without AI, with AI—P(doom) with AI, P(doom) without AI—because it's very possible that AI will be the thing that saves us, saves human civilization from all the other threats.

I agree with you. I think it's insightful. Look, I felt like, to make progress on some of the toughest problems, it would be good to have an AI pair helping you, right? And so that resonates with me for sure.

Quick pause. Bathroom break. You know, let's do that.

If NotebookLM was as compelling as what I saw today with Beam, it blew my mind. It was incredible. I didn't think it was possible. My mind was like, can you imagine the US president and the Chinese president being able to do something like Beam, with the live AI translation working well? So they're both sitting and talking, making a bit more progress.

Just for people listening, we took a quick bathroom break, and now we're talking about the demo I did. We'll probably post it somewhere somehow, maybe here.

I got a chance to experience Beam, and it's hard to describe in words how real it felt with just 6 cameras. It's incredible. It's one of the toughest products to describe to people. Even when we show it in slides, you don't know what it is. You have to experience it in the real world.

On the world leaders front, on politics and geopolitics, there's something really special, again, with studying World War II and how much could have been saved if Chamberlain had met Stalin in person. I sometimes also struggle explaining to people—articulating why I believe meeting in person for world leaders is powerful. It just seems naive to say that, but there is something there in person, and with Beam I felt that same thing.

I'm unable to explain it. All I kept doing was what a child does: “You look real.” I don't know if that makes meetings more productive or so on, but it certainly makes them more human, for the same reason you want to show up to work versus working remotely sometimes: that human connection. I don't know what that is. It's hard to put into words.

There's something beautiful about great teams collaborating on a thing that's not captured by the productivity of that team or by whatever is on paper. Some of the most beautiful moments you experience in life are at work, pursuing a difficult thing together for many months. There's nothing like it. You're in the trenches, and you do form bonds that way for sure. To be able to do that somewhat remotely, with that same personal touch, I don't know, that's a deeply fulfilling thing.

I know a lot of people—I personally hate meetings because a significant percentage of meetings, when done poorly, don't serve a clear purpose. But that's a meeting problem. That's not a communication problem. If you can improve the communication for the meetings that are useful, that's just incredible.

I was blown away by the great engineering behind it, and then we get to see what impact that has. That's really interesting. Just incredible engineering. Really impressive.

Oh, it is. Obviously, we'll work hard over the years to make it more and more accessible. Even on a personal front, outside of work meetings, a grandmother who's far away from her grandchild being able to have that kind of interaction—all of that, I think, will end up being very meaningful.

Nothing substitutes for being in person, but it's not always possible. You could be a soldier deployed, trying to talk to your loved ones. So, that's what inspires us.

When you and I hung out last year and took a walk, I remember—I don't think we talked about this—but I remember seeing dozens of articles written by analysts and experts saying that Sundar Pichai should step down, because the perception was that Google was definitively losing the AI race, had lost its magic touch in the rapidly evolving technological landscape.

Now, a year later, it's crazy. You showed this plot of all the things that were shipped over the past year. It's incredible, and Gemini Pro is winning across many benchmarks and products as we sit here today.

Take me through that experience, when there were all these articles saying you're the wrong guy to lead Google through this—Google's lost, it's done, it's over—to today, where Google is winning again. What were some low points during that time?

Look, lots to unpack. Obviously, the main bet I made as a CEO was to really make sure the company was approaching everything in an AI-first way—really setting ourselves up to develop AGI responsibly and making sure we're putting out products that embody that, things that are very, very useful for people.

I knew, even through moments like that last year, that I had a good sense of what we were building internally. I'd already made many important decisions, bringing together teams of the caliber of Brain and DeepMind and setting up Google DeepMind. There were things like the decision we made to invest in TPUs 10 years ago. We knew we were scaling up and building big models.

Anytime you're in a situation like that, there are a few aspects. I'm good at tuning out noise, separating signal from noise. Do you scuba dive?

No.

You know, it's amazing. I'm not good at it, but I've done it a few times. Sometimes you jump in the ocean and it's so choppy, but you go down 1 foot under and it's the calmest thing in the entire universe, right? So there's a version of that.

Running Google, you may as well be coaching Barcelona or Real Madrid, right? You have a bad season. There are aspects to that, but I'm good at tuning out the noise. I do watch out for signals. It's important to separate the signal from the noise. There are good people sometimes making good points outside, so you want to listen to it. You want to take that feedback in.

Internally, you're making a set of consequential decisions. As leaders, you're making a lot of decisions. Many of them are inconsequential. It feels like they matter, but over time you learn that most of the decisions you're making on a day-to-day basis don't matter. You have to make them to keep things moving, but you have to make a few consequential decisions.

We had set up the right teams, the right leaders. We had world-class researchers. We were training Gemini internally. There were factors that people outside may not have appreciated. TPUs are amazing, but we had to ramp up TPUs, too. That took time, right? To scale, you actually need enough TPUs to get the compute needed.

I could see internally the trajectory we were on, and I was so excited internally about the possibility. To me, this moment felt like one of the biggest opportunities ahead for us as a company. The opportunity space ahead over the next decade, the next 20 years, is bigger than what has happened in the past. I thought we were set up better than most companies in the world to realize that vision.

You had to make some consequential, bold decisions, like you mentioned the merger of DeepMind and Brain. Maybe it's my perspective, just knowing humans, but I'm sure there were a lot of egos involved. It's very difficult to merge teams, and I'm sure there were some hard decisions to be made.

Can you take me through your process of how you think through that? Do you just pull the trigger and make that decision? What were some painful points? How do you navigate those turbulent waters?

Look, we were fortunate to have two world-class teams.

But you’re right. It’s like somebody coming and telling you, “Take Stanford and MIT, put them together, and create a great department.” Easier said than done. But we were fortunate to have phenomenal teams. Both had their strengths, and they were run very differently.

Brain had a lot of diverse, bottoms-up projects, and out of that came a lot of important research breakthroughs. DeepMind, at the time, had a strong vision of how they wanted to build AGI, and so they were pursuing their direction.

Luckily, Jeff had expressed a desire to go back to more of a scientific individual-contributor role. He felt like management was taking up too much of his time. Demis, naturally, was running DeepMind and was a natural choice there.

It took us a while to bring the teams together. Credit to Demis, Jeff, Koray, and all the great people there—they worked super hard to combine the best of both worlds. There were a few sleepless nights here and there as we put that thing together. We were patient in how we did it so that it would work well for the long term.

With things moving fast, you definitely felt the pressure, but I think we pulled off that transition well. They’re obviously doing incredible work, and there are a lot more incredible things ahead coming from them.

Like we talked about, you have a very calm, even-tempered, respectful demeanor. During that time, whether it was the merger or just dealing with the noise, were there times when frustration boiled over? Did you have to go a bit more intense on everybody than you usually would?

Probably. I think it was a moment where we were all driving hard, but when you’re in the trenches working with passion, you’re going to have days where you disagree and argue. All that is just part of the course of working intensely.

At the end of the day, all of us are doing what we’re doing because of the impact it can have. We’re motivated by it. For many of us, this has been a long-term journey, and it’s been super exciting. The positive moments far outweigh the stressful moments.

Just early this year, I had a chance to celebrate back-to-back over 2 days: a Nobel Prize for Geoffrey Hinton, and the next day, a Nobel Prize for Demis Hassabis and John Jumper.

You work with people like that. All that is super inspiring. Is there something with you where you had to put your foot down, maybe with less versus more, where it’s like, “I’m the CEO, and we’re doing this”?

Probably. I think, in the sense that it was a moment where we were all driving hard, but when you’re in the trenches working with passion, you’re going to have days where you disagree and argue. All that is just part of the course of working intensely.

To my earlier point about consequential decisions, there are decisions you make where people can disagree pretty vehemently. At some point, you make a clear decision and ask people to commit. You can disagree, but it’s time to disagree and commit so that we can get moving.

Whether it’s putting your foot down or not, it’s a natural part of what all of us have to do. You can do that calmly and be very firm in the direction you’re making the decision. If you’re clear, people over time respect that. If you can make decisions with clarity, I find it very effective.

In meetings where you’re making such decisions, it’s important, when you can, to hear everyone out. Sometimes what you’re hearing actually influences how you think about it, and you’re wrestling with it while making a decision. Sometimes you have a clear conviction, and you state, “Look, this is how I feel. This is my conviction.” You place the bet and move on.

Some other big decisions like that—I’m intuitively assuming the merger was the big one.

I think that was a very important decision for the company to meet the moment. We had to make sure we were doing that and doing that well. I think that was a consequential decision.

There were many other things. We set up an AI infrastructure team to really meet the moment, scale up the compute we needed, and bring together teams from disparate parts of the company. We created that team to move forward.

We brought people together and got them to work together physically, both in London with DeepMind and at what we call Gradient Canopy, which is where the Mountain View Google DeepMind teams are. One of my favorite moments is that I routinely walk multiple times per week to the Gradient Canopy building, where our top researchers are working on the models.

Sergey is often there among them, getting an update on the model and looking at the loss curves. That cultural part of getting the teams together and bringing back that energy ended up playing a big role, too.

What about the decision to recently add AI Mode? Google Search is, as they say, the front page of the internet. It’s a legendary minimalist thing with 10 blue links. When people think of the internet, they think of that page, and now you’re starting to mess with that.

AI Mode is a separate tab, and then you’re integrating AI into the results. I’m sure there were some battles in meetings on that one.

Look, in some ways, when mobile came, people wanted answers to more questions, so we’re constantly evolving it. But you’re right: this moment is an evolution because the underlying technology is becoming much more capable. You can have AI give a lot of context.

One of our important design goals, though, is that when you come to Google Search, you’re going to get a lot of context, but you’re also going to go and find a lot of things out on the web. That will be true in AI Mode, in AI Overviews, and so on.

To our earlier conversation, we are still giving you access to links. Think of the AI as a layer that is giving you context and a summary. Maybe in AI Mode, you can have a dialogue with it back and forth.

Mhm.

On your journey, you’re learning what’s out there in the world. Those core principles don’t change, but I think AI Mode allows us to push the technology further. We have our best models there—models that are using Search as a deep tool. For every query you’re asking, they’re fanning out, doing multiple searches, and assembling that knowledge in a way so you can consume what you want to. That’s how we think about it.

I got a chance to listen to Liz Reid describe this. Two things stood out to me that you mentioned. One thing is what you were talking about: query fan-out, which I didn’t even think about before. It’s the powerful aspect of integrating a bunch of stuff on the web for you in one place. So, yes, it provides that context so that you can decide which page to then go on to.

The other really big thing speaks to the earlier point about the productivity multiplier we were talking about. She mentioned language. One of the things you don’t quite understand is that through AI Mode, you make English-language websites accessible to non-English speakers in the reasoning process as you try to figure out what you’re looking for.

Of course, once you show up to a page, you can use basic Translate. But if you empathize with a large part of the world that doesn’t speak English, their web is much smaller in that original language. This unlocks that huge cognitive capacity that we take for granted here, with all the bloggers and journalists writing about AI Mode. You forget that this now unlocks that, because Gemini is really good at translation.

It is. I mean, the multimodality, the translation, and its ability to reason—we are dramatically improving tool use. By putting that power in the flow of Search, I think—I’m super excited about AI Overviews.

We’ve seen the product get much better. We measure it using all kinds of user metrics, and it has obviously driven strong growth of the product. We’ve been testing AI Mode; it’s now in the hands of millions of people, and the early metrics are very encouraging. I’m excited about this next chapter of Search.

For people who aren’t thinking through or aware of this, there are the 10 blue links, with AI Overview on top providing a nice summarization. You can expand it, and you have sources and links now embedded.

Yep. Embedded.

Yeah, I believe at least Liz said so. I actually didn’t notice it, but there are ads in the AI Overview. I also don’t think there are ads in AI Mode. When will there be ads in AI Mode?

Two things. The early part of AI Mode will obviously focus more on the organic experience, to make sure we’re getting it right. I think the fundamental value of ads is that they enable us to deploy these services to billions of people.

The second is that the reason we've always taken ads seriously is that we view ads as commercial information. But it's still information, and so we bring the same quality metrics to it. Going back to our earlier conversation, I think AI itself will help us over time figure out the best way to do it.

Given that we're giving context around everything, I think it'll give us more opportunities to also explain, “Okay, here's some commercial information.” Like today, as a podcaster, you do it at certain spots, and you probably figure out what's best in your podcast. I think there are aspects of that, but the underlying need—people value commercial information, businesses are trying to connect to users—all that doesn't change in an AI moment.

But look, we will rethink it. You've seen us on YouTube now do a mixture of subscriptions and ads. Obviously, we are now introducing subscription offerings across everything, and so as part of that, the optimization point will end up being in a different place as well.

Do you see a trajectory in the possible future where AI Mode completely replaces the 10 blue links plus AI Overview?

Our current plan is that AI Mode is going to be there as a separate tab for people who really want to experience that, but it's not yet at the level where our main search page is. As features work, we'll keep migrating it to the main page, and so you can view it as a continuum.

AI Mode will offer you the bleeding-edge experience, but it will keep overflowing into AI Overviews and the main experience.

And the idea is that AI Mode will still take you to the web, to the human-created web?

Yes, that's going to be a core design principle for us.

So really, if users decide, right, they drive this. Yeah, it's just exciting, a little bit scary that it might change the internet, because Google has been dominating with a very specific look and idea of what it means to have the internet. As you move to AI Mode, I mean, it's just a different experience.

I think Liz was talking about—I think you've mentioned that you ask more questions, you ask longer questions, dramatically different types of questions.

Yeah, it actually fuels curiosity. For me, I've been asking a much larger number of questions of this black-box machine, let's say, whatever it is.

With AI Overview, it's interesting because I still value the human. I still ultimately want to end up on the human-created web, but, like you said, the context really helps us deliver higher-quality referrals, where people have a much higher likelihood of finding what they're looking for. They're exploring, they're curious, their intent is getting satisfied more. So all that is what all our metrics show.

It makes the humans that create the web nervous. The journalists are getting nervous. They've already been nervous. Like we mentioned, CNN is nervous because of podcasts. It makes people nervous.

Look, I think news and journalism will play an important role in the future. We're pretty committed to it, right? And so I think making sure that ecosystem—in fact, I think we'll be able to differentiate ourselves as a company over time because of our commitment there.

It's something I definitely value a lot, and as we are designing, we'll continue prioritizing approaches.

I'm sure for the people who want it, they can have a fine-tuned AI model that produces clickbait hit pieces that will replace current journalism. That's a shot at journalism, forgive me. But I find that if you're looking for really strong criticism of things, Gemini is very good at providing that.

Oh, absolutely.

It's better than anything else for now. People are concerned that there would be bias introduced, that as the AI systems become more and more powerful, there's an incentive from sponsors to roll in and try to control the output of the AI models. But for now, the objective criticism that's provided is way better than journalism.

Of course, the argument is that journalists are still valuable, but then, I don't know, the crowdsourced journalism that we get on the open internet is also very, very powerful. I feel like they're all super important things.

I think it's good that you get a lot of crowdsourced information coming in, but I feel like there is real value for high-quality journalism, right? And I think these are all complementary.

I view it as—I find myself constantly seeking out, also, trying to find objective reporting on things. Sometimes you get more context from the crowdsourced sources you read online, but I think both end up playing a super important role.

So you've spoken a little about this. You talked about this sort of slice of the web that will increasingly become about providing information for agents. So we can think about it as two layers of the web: one is for humans, one is for agents.

Do you see the one that's for AI agents growing over time? Do you see there still being long-term, 5-to-10-year value for the human-created web—the web created for the purpose of human consumption—or will it all be agents in the end?

Today, not everyone does, but you go to a big retail store. You love walking the aisles, you love shopping, or you go to a grocery store, picking out food, et cetera. But you're also online shopping and they're delivering, right? So both are complementary, and that's true for restaurants, et cetera.

I do feel like over time websites will also get better for humans. They will be better designed. AI might actually design them better for humans. So I expect the web to get a lot richer and more interesting and better to use.

At the same time, I think there'll be an agentic web, which is also making a lot of progress, and you have to solve the business value and the incentives to make that work well, right? For people to participate in it. But I think both will coexist.

Obviously, the agents won't need the same design and UI paradigms that humans need to interact with, but I think both will be there.

I have to ask you about Chrome. I have to say, for me personally, Google Chrome is probably—I don't know, I'd like to see where I would rank it—but this is not a recency bias, although it might be a little bit. I think it's up there in the top 3, maybe the number-one piece of software for me of all time.

It's just incredible. It's really incredible. The browser is our window to the web. Chrome really continued for many years to push innovation on that front when it was stale, and it continues to challenge, it continues to make it more performant, so efficient, just innovating constantly.

And the Chromium aspect of it, anyway—you were one of the pioneers of Chrome, pushing for it when it was an insane idea, probably one of the ideas that was criticized and doubted and so on. Can you tell me the story of what it took to push for Chrome? What was your vision?

Look, it was such a dynamic time around 2004 and 2005. With Ajax, the web suddenly became dynamic in a matter of a few months. Flickr, Gmail, Google Maps all kind of came into existence, right? The fact that you had an interactive, dynamic web—the web was evolving from simple text pages, simple HTML, to rich, dynamic applications.

But at the same time, you could see the browser was never meant for that world, right? JavaScript execution was super slow. The browser was far away from being an operating system for that rich, modern web which was coming into place. So that's the opportunity we saw.

It was an amazing early team. I still remember the day we got a shell on WebKit running and how fast it was. We had a clear vision for building a browser: We wanted to bring core OS principles into the browser. So we built a secure browser sandbox; each tab was its own process. These things are common now, but at the time it was pretty unique.

We found an amazing team in Denmark, with a leader who built V8, the JavaScript VM, which at the time was 25 times faster than any other JavaScript VM out there. And by the way, you're right: We open-sourced it all and put it in Chromium too.

But we really thought the web could work much better, much faster, and you could be much safer browsing the web. The name Chrome came from the fact that we literally felt the chrome of the browser was getting clunkier. We wanted to minimize it. And so that was the origin of the project.

Definitely, obviously, a highly biased person here talking about Chrome, but it's the most fun I've had building a product from the ground up, and it was an extraordinary team. I had my co-founders on the project—they were terrific. So, definite fond memories.

So for people who don't know, Sundar, it's probably fair to say you're the reason we have Chrome. I know there are a lot of incredible engineers, but pushing for it inside a company that probably was opposing it because it's a crazy idea—because, as everybody probably knows, it's incredibly difficult to build a browser.

Yeah, look, Eric, who was the CEO at the time, I think it was less that he was opposed to it. He firsthand knew what a crazy thing it is to go build a browser, and so he definitely was like, “This is…” There was a crazy aspect to actually wanting to go build a browser.

But he was very supportive. Everyone—the founders—were. I think once we started building something and we could use it and see how much better it was, from then on, you're really tinkering with the product and making it better. It came to life pretty fast.

What wisdom do you draw from pushing through on a crazy idea in the early days that ends up being revolutionary? What can we learn for future crazy ideas like it?

This is something Larry and Sergey have articulated clearly. I really internalized this early on, which is their whole feeling around working on moonshots. When you work on something very ambitious, first of all, it attracts the best people, so that's an advantage. Second, because it's so ambitious, you don't have others working on something crazy, so you pretty much have the path to yourselves. It's way more self-driving.

Third, even if you end up not quite accomplishing what you set out to do and you end up doing 60% or 80% of it, you'll end up being a terrific success. That's the advice I would give people. Aiming for big ideas has all these advantages, and it's risky, but it also has all these advantages that people, I don't think, fully internalize.

You mentioned one of the craziest, biggest moonshots, which is Waymo. When I first saw, over a decade ago, a Waymo vehicle—a Google self-driving car—it was an aha moment for robotics. It made me fall in love with robotics even more than before. It gave me a glimpse into the future, so it's incredible. I'm truly grateful for that project and for what it symbolizes.

But it's also a crazy moonshot. For a long time, it has been, just like you mentioned with scuba diving, not listening to anybody and calmly improving the system—better and better, more testing, and expanding the operational domain more and more. First of all, congrats on 10 million paid robotaxi rides. What lessons do you take from Waymo about the perseverance and persistence on that project?

I'm really proud of the progress we have had with Waymo. One of the things we were very committed to is that the final 20% can look like—I mean, we always say the first 80% is easy, and the final 20% takes 80% of the time. I think we were definitely working through that phase with Waymo, but I was aware of that. We knew we were at that stage.

There were many other self-driving companies, but we knew the technology gap was there. In fact, right at the moment when others were doubting Waymo is when we made the decision to invest more in Waymo. In some ways, it's counterintuitive, but we have always been a deep-technology company, and Waymo is a version of building an AI robot that works well.

The caliber of the teams there is phenomenal. I know you followed the space super closely, so I'm talking to someone who knows the space well, but it was very obvious that it was going to get there. There's still more work to do, but it's a good example of how we always prioritized being ambitious and safety at the same time—equally committed to both—and pushed hard. We couldn't be more thrilled with how it's working and how much people love the experience. This year, we've scaled up a lot, and we'll continue scaling up in 2026.

That said, the competition is heating up. You've been friendly with Elon, even though Tesla is a competitor, but you've been friendly with a lot of tech CEOs in that way, showing respect toward them and so on. What do you think about the robotaxi efforts that Tesla is doing? Do you see it as competition? What do you think? Do you like the competition?

We are one of the earliest and biggest backers of SpaceX as Google, so we're thrilled with what SpaceX is doing and fortunate to be investors as a company there. We don't compete with Tesla directly. We are not making cars. We're building Level 4 and Level 5 autonomy. We're building the Waymo Driver, which is general-purpose and can be used in many settings. They're obviously working on making Tesla self-driving, too.

I've just assumed it's a fait accompli that Elon would succeed in whatever he does. That's not something I question. I think we are still so far from the limits of these spaces. They're such vast spaces. I think about the opportunity space in transportation. The Waymo Driver is a general-purpose technology we can apply in many situations, so we have a vast green space.

In all future scenarios, I see Tesla doing well and Waymo doing well. Like we mentioned with the Neolithic package, I think it's very possible that in the quote-unquote AI package, when history is written, autonomous vehicles and self-driving cars will be the big thing that changes everything. Imagine, over a period of a decade or 2, just a complete transition from manually driven to autonomous in ways we might not predict. It might change the way we move about the world completely.

The possibility of that, and then the second- and third-order effects, as you're seeing now with Tesla, very possibly you would see some of that internally with Alphabet. Maybe Waymo, maybe some of the Gemini Robotics stuff, might lead you into the other domains of robotics, because we should remember that Waymo is a robot. It just happens to be on 4 wheels.

You said that the next big thing—we can also throw that into the AI package—the big aha moment might be in the space of robotics. What do you think that would look like?

The DeepMind team is very focused on Gemini Robotics. We are definitely building the underlying models. We have a lot of investments there, and I think we're also pretty cutting-edge in our research there, so we're definitely driving in that direction. We are obviously thinking about applications in robotics. We'll work seriously on it. We are partnering with a few companies today.

It's an area where I would say, stay tuned. We are yet to fully articulate our plans outside, but it's an area we are definitely committed to driving a lot of progress in. I think AI ends up driving that massive progress in robotics. The field has been held back for a while. The hardware has made extraordinary progress; the software has been the challenge.

With AI now and the generalized models we are building, getting these models to work in the real world in a safe, generalized way is the frontier we're pushing pretty hard on.

It's really nice to see the models and the different teams integrated, where all of them are pushing toward one world model that's being built from all these different angles—multimodal. You're ultimately trying to get Gemini. The same thing that would make AI Mode really effective at answering your questions, which requires a kind of world model, is the same kind of thing that would help a robot be useful in the physical world. So everything is aligned.

That is what makes this moment so unique, because, running a company, for the first time you can do one investment in a very deep, horizontal way. On top of it, you can drive multiple businesses forward. That's effectively what we are doing in Google and Alphabet.

Yeah, it's all coming together like it was planned ahead of time, but it's not, of course. It's all distributed. If Gmail and Sheets and all these other incredible services—I can sing Gmail's praises for years. It's just revolutionized email—but the moment you start to integrate AI, Gemini, into Gmail, that's the other thing.

Speaking of productivity multipliers, people complain about email, but that changed everything. The invention of email changed everything, and it's been ripe. There have been a few folks trying to revolutionize email, some of them on top of Gmail, but that's ripe for innovation—not just spam filtering, but you demoed a really nice demo of personalized responses, right?

At first, I felt really bad about that. But then I realized there's nothing wrong to feel bad about, because the example you gave is when a friend asks, “You went to whatever hiking location. Do you have any advice?” It just searches through all your information to give them good advice, and then you put the cherry on top, maybe some love or whatever camaraderie.

But the transformational aspect—the knowledge transfer it does for you—I think there'll be important moments. It should be like today: if you write a card in your own handwriting and send it to someone, that's a special thing. Similarly, there'll be a time—I mean, to your friends, maybe your friend wrote and said he's not doing well or something. Those are moments you want to save your time for, writing something and reaching out.

But saying, “Give me all the details of the trip you took,” to me makes a lot of sense for an AI assistant to help you. I think both are important, but I think I'm excited about that direction.

Yeah, I think ultimately it gives more time for us humans to do the things we humans find meaningful. I think it scares a lot of people, because we're going to have to ask ourselves the hard question of what we find meaningful. I'm sure there are answers.

It's the old question of the meaning of existence. You have to try to figure that out. That might ultimately be parenting, or being creative in some domains of art or writing. It challenges you to ask yourself a good question: in my life, what is the thing that brings me the most joy and fulfillment? If I'm able to actually focus more time on that, that's really powerful.

I think that's the holy grail. If you get this right, I think it allows more people to find that. I have to ask you on the programming front: AI is getting really good at programming. Gemini, both the agentic version and just the LLM, has been incredible.

A lot of programmers are really worried that they will lose their jobs. How worried should they be, and how should they adjust so they can thrive in this new world where more and more code is written by AI?

I think a few things. Looking at Google, we've given various stats around 30% of code now using AI-generated suggestions or whatever it is, but the most important metric—and we carefully measure it—is how much our engineering velocity has increased as a company due to AI. It's tough to measure, and we rigorously try to measure it. Our estimates are that number is now at 10%.

Across the company, we've accomplished a 10% engineering velocity increase using AI, but we plan to hire more engineers next year because the opportunity space of what we can do is expanding too. So, hopefully, at least in the near to midterm, for many engineers it frees up more and more of the engineering and coding work.

There are aspects of engineering and coding that are so much fun: you're designing, you're architecting, you're solving a problem. There's a lot of grunt work, which all goes hand in hand, but hopefully it takes a lot of that away, makes it even more fun to code, and frees up more time to create, problem-solve, brainstorm with your fellow colleagues, and so on. That's the opportunity there.

Second, I think it'll put the creative power in more people's hands, which means people will create more. That means there'll be more engineers doing more things. So it's tough to fully predict, but in general, in this moment, it feels like people will adopt these tools and be better programmers. There are more people playing chess now than ever before, right? It feels positive that way to me, at least. Speaking from within a Google context, that's how I would talk to them about it.

I just know anecdotally that a lot of great programmers are generating a lot of code. They're not always using all the code; there's still a lot of editing. But even for me—it's still programming as a side thing—I think I'm 5 times more productive. I think that's true even for a large codebase that's touching a lot of users, like Google's does. I'm imagining that very soon, that productivity should be going up even more.

The big unlock will be as we make the agent capabilities much more robust, right? I think that's what unlocks that next big wave. I think the 10% is a massive number. If tomorrow I showed up and said, "You can improve a large organization's productivity by 10%," when you have tens of thousands of engineers, that's a phenomenal number.

That's different from what others cite as statistics, saying that this percentage of code is now written by AI. I'm talking more about overall productivity—the actual engineering productivity—which are 2 different things, and which is the more important metric. I think it'll get better. There's no engineer who, if they magically became 2 times more productive tomorrow, wouldn't just create more things. They'd create more value-added things, and I think they'd find more satisfaction in their job.

There are a lot of aspects to it. The actual Google codebase might just improve because it'll become more standardized, easier for people to move around the codebase, because AI will help with that. Therefore, that will also allow the AI to understand the entire codebase better, which makes the engineering aspect better.

I've been using Cursor a lot as a way to program with Gemini and other models. One of its powerful things is that it's aware of the entire codebase, and that allows you to ask questions of it. It allows the agents to move around that codebase in a really powerful way.

I mean, that's a huge unlock. Think about migrations and refactoring old codebases. I think, once we can do all this in a much better, more robust way than where we are today, everything will be written in JavaScript and run in Chrome. I think it's all going in that direction.

I mean, just for fun, Google has legendary coding interviews—rigorous interviews for engineers. How can you comment on how that has changed in the era of AI? It's such a weird thing. The whiteboard interview, I assume, doesn't allow prompts.

Such a good question. Look, I do think we're making sure we'll introduce at least 1 round of in-person interviews for people, just to make sure the fundamentals are there. I think that'll end up being important, but it's an equally important skill: if you can use these tools to generate better code, I think that's an asset. So overall, I think it's a massive positive.

For vibe-coding engineers, do you recommend that students interested in programming still get an education in computer science and a college education? What do you think?

If you have a passion for computer science, I wouldn't change what you pursue. Computer science is obviously a lot more than programming alone. I think AI will horizontally impact every field. It's pretty tough to predict in what ways, so any education in which you're learning good first-principles thinking, I think, is good education.

You've revolutionized web browsing, and you've revolutionized a lot of things over the years. Android changed the game. It's an incredible operating system, and we could talk for hours about Android.

What does the future of Android look like? Is it possible for it to become more and more AI-centric, especially now that you throw Android XR into the mix, with its ability to do augmented reality, mixed reality, and virtual reality in the physical world?

Yeah. The best innovations in computing have come when you go through a paradigm I/O change, like with a graphical user interface, then with multitouch in the context of mobile, and voice later on. Similarly, I feel like AR is that next paradigm.

I think it was held back by the system-integration challenges of making good AR—it's very, very hard. The second thing is that you need AI to actually work; otherwise, the I/O is too complicated for you to have natural, seamless I/O for that paradigm. AI ends up being super important, and so this is why Project Astra ends up being super critical for that Android XR world.

I think glasses are a real opportunity for Android, and I've always been amazed at how useful these things are going to be. XR is 1 way it will really come to life. But I think there's an opportunity to rethink the mobile OS too.

We've been living in this paradigm of apps and shortcuts. All that won't go away, but if you're trying to get stuff done at an operating-system level, it needs to be more agentic, so that you can describe what you want to do, or it proactively understands what you're trying to do, learns from how you're doing things over and over again, and adapts to you. All that is the unlock we need to pursue with a basic, efficient, minimalist UI.

I've gotten a chance to try the glasses, and they're incredible. It's the little stuff. It's hard to put into words, but there's no latency; it just works. Even that little map demo, where you look down and you look up, has a very smooth transition between the 2, and a very small amount of useful information is shown to you—enough not to distract from the world outside, but enough to provide a bit of context when you need it.

In order to bring that into reality, you have to solve a lot of the OS problems to make sure it works when you're integrating the AI into the whole thing. So everything you do launches an agent that answers some basic question. A good moonshot. It's crazy.

No, but I think we are much closer to reality than with other moonshots. We expect to have glasses in the hands of developers later this year, and in consumers' hands next year. So it's an exciting time.

Yeah. Well, Beam is extremely well-executed—all the stuff. Sometimes you don't know. Somebody commented on a top comment on one of the demos of Beam, "This will either be killed off in 5 weeks or revolutionize all meetings in 5 years."

Google tries so many things, and sometimes sadly kills off very promising projects because there are so many other things to focus on. I use so many Google products. I still use Google Voice. I'm so glad that's not being killed off. It's still alive. Thank you to whoever is defending that, because it's awesome and it's great that you keep innovating.

I just want to list them off as a big thank-you. Search, obviously, Google revolutionized. Chrome—and all of these could be multi-hour conversations. Gmail: I've been singing Gmail's praises forever. Maps—an incredible technological innovation that's revolutionized mapping. Android, like we talked about. YouTube, like we talked about. AdSense. Google Translate. For the academic-minded, Google Scholar is incredible, along with Google Books and the scanning of the books.

So making all the world's knowledge accessible, even with that knowledge being a kind of niche thing—which Google Scholar is. And then obviously with DeepMind, with AlphaZero, AlphaFold, and AlphaEvolve—I could talk forever about AlphaEvolve. That's mind-blowing.

All of that was released as part of the set of things you've released this year, when those brilliant articles were written about Google being done. And, like we talked about, pioneering self-driving cars and quantum computing could be another thing that is low-key scuba-diving its way to changing the world forever.

So, another Pichai-slash-micro-kitchen question: If you build AGI, what kind of question would you ask it? What would you want to talk about? Definitely, Google has created AGI that can basically answer any question. What topic are you going to? Where is it? Where are you going?

It's a great question. Maybe it's proactive by then and should tell me a few things I should know. But I think if I were to ask it, it'll help us understand ourselves much better, in a way that'll surprise us, I think. Maybe you already see people do it with the products, but in an AGI context, I think that'll be pretty powerful at a personal level, or in terms of general human nature.

You talking to AGI—I think there's some chance it'll kind of understand you in a very deep way. I think, in a profound way, that's a possibility. I think there's also the obvious thing of maybe it helps us understand the universe better, in a way that expands the frontiers of our understanding of the world. That is something super exciting.

But look, I really don't know. I haven't had access to something that powerful yet, but I think those are all possibilities. On the personal level, asking questions about yourself—a sequence of questions like, “What makes me happy?”—I think we would be very surprised to learn those kinds of things.

A sequence of questions and answers might explore some profound truths in the way that sometimes art reveals to us, great books reveal to us, or great conversations with loved ones reveal to us things that are obvious in retrospect but are nice when they're said. But for me, the number-one question is: How many alien civilizations are there?

100 percent.

That's going to be your first question?

Number one: How many living and dead alien civilizations? Maybe a bunch of follow-ups, like how close are they? Are they dangerous? If there are no alien civilizations, why? Or if there are no advanced alien civilizations but bacteria-like life everywhere, why? What is the barrier preventing it from getting to that?

Is it because, when you get sufficiently intelligent, you end up destroying yourselves? You need competition in order to develop an advanced civilization, and when you have competition, it's going to lead to military conflict, and conflict eventually kills everybody. I don't know. I'm going to have that kind of discussion—get an answer to the Fermi paradox.

Yeah, exactly.

And have a real discussion about it. I'm realizing now that your answer is a more productive answer, because I'm not sure what I'm going to do with that information. But maybe it speaks to the general human curiosity that Liz talked about—that we're all just really curious.

Making the world's information accessible allows our curiosity to be satiated somewhat. With AI even more, we can be more and more curious and learn more about the world and about ourselves.

Many years ago, there was an MIT study. They estimated the impact of Google Search, and they basically said it's the equivalent of, on a per-person basis, a few thousand dollars per year per person—the value that got created per year. But it's tough to capture these things. You kind of take it for granted as these things come, and the frontier keeps moving.

How do you measure the value of something like AlphaFold over time, and so on? Also, the increase in quality of life when you learn more.

I have to say, with some of the programming I've done with AI, for some reason I'm more excited to program. The same is true with knowledge and discovering things about the world. It makes you more excited to be alive. It makes you more curious, too, and the more curious you are, the more exciting it is to live and experience the world.

It's very hard to—I don't know if that makes you more productive. Probably not nearly as much as it makes you happy to be alive, and that's a hard thing to measure. The quality-of-life increase—some of these things do.

As AI continues to get better and better at everything that humans do, what do you think is the biggest thing that makes us humans special?

Look, I think it's stuff that's at the essence of humanity. There's something about the consciousness we have, what makes us uniquely human. Maybe the lines will blur over time, and it's tough to articulate, but hopefully we live in a world where, if you make resources more plentiful and make the world less of a zero-sum game over time—which it's not, but in a resource-constrained environment people perceive it to be—the values of what makes us uniquely human—empathy, kindness, all that—surface more. That's the aspirational hope I have.

Yeah, it multiplies the compassion, but also the curiosity—the banter, the debates we'll have about the meaning of it all. I also think, in the scientific domains, all the incredible work that DeepMind is doing will continue to explore scientific questions, mathematical questions, and physics questions, even as AI gets better and better at helping us solve some of the questions.

Sometimes the question itself is a really difficult thing: both the right new questions to ask and the answers to them, and the self-discovery process which it'll drive. I think our early work with AI co-scientist and AlphaEvolve is just super exciting to see.

What gives you hope about the future of human civilization?

I've always—I'm an optimist, and I look at it this way: If you were to take the journey of human civilization, we've relentlessly made the world better in many ways. At any given moment in time, there are big issues to work through.

I always ask myself the question: Would you have been born now or at any other time in the past? I almost always would rather be born now. That's the extraordinary thing human civilization has accomplished, and we've constantly made the world a better place.

Something tells me that, as humanity, we always rise collectively to drive that frontier forward. So I expect it to be no different in the future.

I agree with you totally. I'm truly grateful to be alive in this moment, and I'm also really excited for the future. The work you and the incredible teams here are doing is one of the big reasons I'm excited for the future, so thank you.

Thank you for all the cool products you've built, and please don't kill Google Voice.

Thank you. We won't.

Yeah. Thank you for talking today. This was incredible.

Thank you. Real pleasure. I appreciate it.

Thanks for listening to this conversation with Sundar Pachchai. To support this podcast, please check out our sponsors in the description or at lexfreedman.com/sponsors.

Shortly before this conversation, I got a chance to get a couple of demos that frankly blew my mind. The engineering was really impressive. The first demo was Google Beam, and the second demo was the XR glasses. Some of it was caught on video, so I thought I would include here some of those video clips.

Andrew Edelman

Hey, Lex. My name is Andrew. I lead the Google Beam team, and we're excited to show you a demo. We're going to show you, I think, a glimpse of something new. That's the idea: a way to connect, a way to feel present from anywhere with anybody you care about.

Here's Google Beam. This is a development platform that we've built. There's a prototype here of Google Beam. There's one right down the hallway. I'm going to go down and turn that on in a second. We're going to experience it together. We'll be back in the same room.

Wonderful. Wow. Okay, here we are. All right. This is real already. Wow. This is real. Good to see you.

Andrew Edelman

This is Google Beam. We're trying to make it feel like you and I could be anywhere in the world, but when these magic windows open, we're back together. I see you exactly the same way you see me. It's almost like we're sitting at a table together. I could learn from you, talk to you, share a meal with you, and get to know you.

So you can feel the depth of this.

Yeah. Great to meet you. Wow. Wow. For people who probably can't even imagine what this looks like, there's a 3D version of it. It looks real. You look real. It looks real to me; it looks real to you. It looks like you're coming out of the screen.

Andrew Edelman

We quickly believe, once we're in Beam, that we're just together. You settle into it. You're naturally attuned to seeing the world like this, and you just get used to seeing people this way—literally from anywhere in the world with these magic screens.

This is incredible. It's a neat technology. Wow. So, I saw demos of this, but they don't come close to the experience of this. I think one of the top YouTube comments on one of the demos I saw was, “Why would I want a high-definition video call?” I'm trying to turn off the camera, but this actually feels like the camera has been turned off and we're just in the same room together. This is really compelling.

Andrew Edelman

That's right. I know it's kind of late in the day, too, so I brought you a snack just in case you're a little bit hungry. But you can push it further, and it just becomes—let's try to float it between rooms. You know, it kind of fades from my room into your room, and then you see my hand, the depth of my hand.

Of course. Yes, of course. It feels like you—try this. Try, give me a high five.

Andrew Edelman

And there's almost a sensation of feeling touch. You almost feel, because you're so attuned to that, that it should be a high five. It feels like you could connect with somebody that way. So, it's kind of a magical experience.

Oh, this is really nice. How much does it cost?

Andrew Edelman

Yeah, we've got a lot of companies testing it. We just announced that we're going to be bringing it to offices soon as a set of products. We've got some companies helping us build these screens. But eventually, I think this will be on almost every screen.

I'm not wearing anything. Well, I'm wearing a suit and tie, to clarify. I ain't wearing glasses. This is not CGI. But outside of that, cool. The audio is really good, and you can see me in the same three-dimensional way.

Andrew Edelman

Yeah, the audio is spatialized. So, if I'm talking from here, of course it sounds like I'm talking from here. If I move to the other side of the room here, these little subtle cues really matter to bring people together. All the nonverbals, all the emotion, the things that are lost today—we put them back into the system.

You pulled this off. Holy crap. They pulled it off and integrated it into this. I saw the translation also. This is the—

Andrew Edelman

Yeah, we've got a bunch of things. Let me show you a couple of cool things. Let's do a little bit of work together. Maybe we could critique one of your latest episodes. You and I are working together, so of course we're in the same room, but with this superpower, I can bring other things in here with me. It's nice. We could sit together, we could watch something, we could work. We've shared meals as a team together in this system, but once you do the presence aspect of this, you want to bring some other superpowers to it. You could review code together.

Yeah, exactly.

Andrew Edelman

I've got some slides I'm working on. Maybe you could help me with this. Keep your eyes on me for a second. I'll slide back into the center. I didn't really move, but the system just puts us in the right spot and knows where we need to be.

Oh, so you just turn to your laptop, the system moves you, and then it does the overlay automatically.

Andrew Edelman

It morphs the room to put things in the spot that they need to be in. Everything has a place in the room. Everything has a sense of presence or spatial consistency, and that makes it feel like we're together with each other and other things.

I should also say you're not just three-dimensional. It feels like you're leaning out of the screen. You're coming out of the screen. You're not just in that world three-dimensionally.

Andrew Edelman

Yeah, exactly.

Holy crap. Move back to center.

Andrew Edelman

Okay, okay, okay. Let me tell you how this works. You probably already have the premise of it, but there are 2 really hard things that we put together. One is an AI video model. There's a set of cameras—you asked about those earlier. There are 6 color cameras, just like webcams that we have today, taking video streams and feeding them into our AI model and turning that into a 3D video of you and me.

It's effectively a light field. It's an interactive 3D video that you can see from any perspective. That's transmitted over to the second thing, and that's a light-field display, and it's happening bidirectionally. I see you and you see me, both in our light-field displays.

These are effectively flat televisions, or flat displays, but they have a sense of dimensionality. The depth and size are correct. You can see shadows and lighting are correct, and everything's correct from your vantage point. So, if you move around ever so slightly and I hold still, you see a different perspective here. You see things that were occluded become revealed. You see shadows that move in the way they should move.

All of that's computed and generated using our AI video model for you. It's based on your eye position: Where does the right scene need to be placed in this light-field display for you just to feel present? It's real time, with no latency.

I'm not seeing latency. You weren't freezing up at all.

Andrew Edelman

No, no, I hope not. I think it's you and I together, real time. That's what you need for real communication, and at a quality level that's realistic.

Is it possible to do 3 people? Is that going to move that way also?

Andrew Edelman

Yeah, let me show you. If she enters the room with us, you can see her and you can see me. If we had more people, you would eventually lose the sense of presence. You kind of shrink people down and lose a sense of scale.

Think of it as the window fitting a certain number of people. If you want to fit a big group of people, you want the boardroom or the big room; you need a much wider window. If you want to see just grandma and the kids, you can do smaller windows, so everybody has a seat at the table or a sense of where they belong. There's a sense of presence that's obeyed. If you have too many people, you go back to the 2D metaphors that we're used to.

People in tiles placed anywhere, for the image I'm seeing? Did you have to get scanned? I see you without being scanned, so it's just so much easier if you don't have to wear anything. You don't have to prescan. You just do it the way it's supposed to happen without anybody having to learn anything or put anything on. I thought you had to solve the scanning problem, but here you don't. It's just cameras. It's just vision. It's video.

Andrew Edelman

Yeah, we're not trying to make an approximation of you, because everything you do every day matters. I cut myself shaving. I put on a pin. All the little aspects of you just happen. We don't have the time to scan or capture those or dress avatars. We appear as we appear, and so all that's transmitted truthfully as it's happening.

How are you doing? Good to meet you. Nice to meet you.

Shahram Izadi

So, as Max mentioned, I've got the eyeglasses here. We start with a foundation of great glasses: something stylish, lightweight, and wearable. We say, how can we build great technology and experiences on top of that?

One of the core tenets of the Android XR platform is this idea of a multimodal conversational device: see what you see, hear what you hear. We've got a camera, speakers, and multiple microphones for speaker isolation. I'll give you a chance to try these yourself.

Sorry, I woke it up there. Whoa.

Shahram Izadi

The first thing you see is a super-simple, straightforward home screen. You probably see the time, the weather, the calendar, and appointments there. This is designed to be your one-stop shop for quick, glanceable information throughout the day.

We want to be something that's easy to get what you need from and then go back to what you're doing. You can imagine turning on the display, getting that quick glance, and then continuing on your way. You can be fully conversational at this point. You can ask questions, for example, about the paintings. You can interrupt, and you can ask follow-up questions. If you want to pause at any time, you just tap there on the right.

Gemini, how much does this painting cost?

Gemini

The painting is called The Bridge of Belief Leap, 1875, by Arman Gome[?]. It appears to be a print, and it would be impossible to know its cost without finding the seller.

Okay, so this is fake. I understand. Why do humans pay so much money for paintings?

Gemini

People pay a lot for paintings because of their artistic merit, the artist, the decent rarity, and the appreciation.

Shahram Izadi

A few more things that we want to show you, just for the sake of time. Go ahead and long-press on the side again to put Gemini to sleep. There you go. Did you catch Google I/O last week by any chance? You might have seen the Google Maps experience onstage very briefly. I want to give you a chance to get a sense of what that feels like today.

You can imagine you're walking down the street. If you look up, like you're walking straight ahead, you get quick turn-by-turn directions, so you have a sense of what the next turn is like while keeping your phone in your pocket.

Oh, that's so intuitive.

Shahram Izadi

Sometimes you need that quick sense of which way is the right way. So, let's say you're coming out of the subway or getting out of a cab—you can just glance down at your feet.

We have it set up to translate from Russian to English.

I think I get to wear the glasses.

Shahram Izadi

You speak to me, if you don't mind.

I can speak Russian. I'm doing well. How are you doing? I'm tempted to swear. I'm tempted to say inappropriate things. I see it transcribed in real time, and so, obviously, based on the different languages and the sequence of subjects and verbs, there's a slight delay sometimes. But it's really just like subtitles for the real world.

Thank you for this. All right, back to me. Hopefully, watching videos of me having my mind blown like the apes in 2001: A Space Odyssey playing with a monolith was somewhat interesting. Like I said, I was very impressed. And now I thought, if it's okay, I could make a few additional comments about the episode and just in general.

In this conversation with Sundar Pichai, I discussed the concept of the Neolithic package, which is the set of innovations that came along with the first agricultural revolution about 12,000 years ago. This included the formation of social hierarchies, early, primitive forms of government, labor specialization, the domestication of plants and animals, early forms of trade, and large-scale cooperation among humans, like that required to build the pyramids and temples like Göbekli Tepe.

I think this may be the right way to talk about the inventions that changed human history—not just as a single invention, but as a kind of network of innovations and transformations that came along with it. The productivity multiplier framework that I mentioned in the episode is a nice way to try to concretize the impact of each of these inventions under consideration. We have to remember that each node in the network of the sort of fast-follow-on inventions is itself a productivity multiplier. Some are additive, and some are multiplicative.

So, in some sense, the size of the network in the package is the thing that matters when you're trying to rank the impact of inventions on human history. The easy picks for the period of biggest transformation, at least in modern-day discourse, are the Industrial Revolution or, even in the 20th century, the computer or the internet. I think it's because it's easiest for modern-day humans to intuit the exponential impact of those technologies.

Recently—and I suppose this changes week to week—I have been doing a lot of reading on ancient human history. So, recently, my pick for the number one invention would have to be the first agricultural revolution: the Neolithic package that led to the formation of human civilizations. That's what enabled the scaling of the collective intelligence machine of humanity and for us to become the early bootloader for the next 10,000 years of technological progress, which, yes, includes AI and the technology that builds on top of AI.

Of course, it could be argued that the word “invention” doesn't properly apply to the agricultural revolution. I think Yuval Harari argues that it wasn't humans who were the inventors, but a handful of plant species—namely, wheat, rice, and potatoes. This is a fair perspective, but I'm having fun, like I said, with this discussion.

I just think of the entire Earth as a system that continuously transforms, and I'm using the term “invention” in that context, asking the question: When was the biggest leap on the log-scale plot of human progress? Will AI, AGI, or ASI eventually take the number one spot in this ranking? I think it has a very good chance to do so, again due to the size of the network of inventions that will come along with it.

I think we discussed in this podcast the kinds of things that would be included in the so-called AI package, but I think there are a lot more possibilities, including things discussed in many previous podcasts, including with Dario Amodei, on the biological innovation side and the science progress side. In this podcast, I think we talk about something that I'm particularly excited about in the near term, which is unlocking the cognitive capacity of the entire landscape of brains that is the human species.

That means making it more accessible through education and through machine translation, making information, knowledge, and the rapid learning and innovation process accessible to more humans—to the entire 8 billion, if you will. So, I do think language, or machine translation applied to all the different methods that we use on the internet to discover knowledge, is a big unlock.

There are a lot of other things in the so-called AI package, like curing all major human diseases, as I discussed with Dario. He really focuses on that in the “Machines of Loving Grace” essay. I think there will be huge leaps in productivity for human programmers and semi-autonomous human programmers—humans in the loop, but with most of the programming done by AI agents—and then moving that toward a superhuman AI researcher that's doing the research that develops and programs the AI system itself.

I think there would be huge transformative effects from autonomous vehicles. These are the things that we maybe don't immediately understand, or that we understand from an economics perspective. But there will be a point when AI systems are able to interpret, understand, and interact with the human world to a sufficient degree that many of the manually controlled, human-in-the-loop systems we rely on become fully autonomous.

I think mobility is such a big part of human civilization that there will be effects from that that are not just economic, but social, cultural, and so on. There are a lot more things I could talk about for a long time. Obviously, there's the integration and utilization of AI in the creation of art, film, and music; the digitization and automation of basic functions of government; and the integration of AI into that process, thereby decreasing corruption and cost and increasing transparency and efficiency.

I think we, as individual humans, will continue to transition further and further into cyborgs. There's already AI in the loop of the human condition, and that will become increasingly so as AI becomes more powerful. The thing I'm obviously really excited about is major breakthroughs in science—not just on the medical front, but in physics and fundamental physics—which would then lead to energy breakthroughs, increasing the chance that we actually become a Kardashev Type I civilization and, in so doing, enabling us to carry out interstellar exploration and colonization of space.

I think there's also, in the near term, much like with the Industrial Revolution that led to rapid specialization of skills and expertise, a great sort of despecialization. As AI systems become superhuman experts in particular fields, there might be greater and greater value to being the integrator of AIs for humans—to being generalists.

The great value of the human mind will come from generalists, not specialists. That's a real possibility that changes the way we think about the world: We want to know a little about a lot of things and move about the world in that way. That could, when passing a certain threshold, completely shift who we are as a collective intelligence, as a human species.

Also, as an aside, when thinking about the invention that was the greatest in human history—again, for a bit of fun—we have to remember that all of them build on top of each other. So, we need to look at the delta, the step change, on the impossibly-to-perfectly-measure plot of exponential human progress.

Really, we can go back to the entire history of life on Earth. A previous podcast guest, Nick Lane, does a great job of this in his book Life Ascending, listing 10 major inventions throughout the evolution of life on Earth, like DNA, photosynthesis, complex cells, sex, movement, sight, and all those kinds of things. I forget the full list that's in there, but I think that's so far from the human experience that my intuition about, let's say, the productivity multipliers of those particular inventions completely breaks down, and a different framework is needed to understand the impact of these inventions of evolution, the origin of life on Earth, or even the Big Bang itself.

The Big Bang, of course, is the OG invention that set the stage for all the rest of it. There are probably many more turtles under that which are yet to be discovered.

So, anyway, we live in interesting times, fellow humans. I do believe the set of positive trajectories for humanity outnumbers the set of negative trajectories, but not by much. So, let's not mess this up.

Now let me leave you with some words from French philosopher Jean de La Bruyère: “Out of difficulties grow miracles.”

Thank you for listening, and I hope to see you next time.