During the period when outsiders said Google was losing the AI race and that Sundar should step down, he was calm because he could see the internal trajectory — the TPU investments, the DeepMind-Brain merger, and the models being trained — and felt the opportunity ahead was bigger than anything in the company's past.

(00:00:00) The following is a conversation with Sundar Pichai, the CEO of Google and Alphabet. And now a quick few second mention of your sponsor. Check them out in the description or at lexfreedman.com slash sponsors. (00:00:16) It's the best way to support this podcast. We got Tax Network USA for taxes, BetterHelp for mental health, Element for electrolytes, Shopify for selling stuff online, and AG1 for your daily multivitamin drink. Choose wisely, my friends. And now onto the philateries. You can skip them if you like, but if you do, please still check out our sponsors. I enjoy their stuff. Maybe you will too. If you want to get in touch with me for whatever reason, go to lexfreeman.com slash contact. (00:00:44) All right, let's go. This episode is brought to you by Tax Network USA, a full-service tax firm focused on solving tax problems for individuals and for small businesses. I remember when I was preparing for the Roman Empire episode, I came across a lot of places where there was a rigorous discussion about the intricate tax collection algorithms used by the Roman Empire. (00:01:14) The reason I use the word algorithms is basically there's a systematic process for determining how much you owe based on your location, based on your status, based on your job, based on all these kinds of factors. It's sad, but those rules in the early days initially give power to the individual because they protect the individual. But when they become too complicated, then the bureaucracy, the centralized power starts to abuse (00:01:43) its power by using the rules and then the individual loses power because they can't figure out the complexity of the rules. And that's essentially why you need the CPAs and the firms to figure out the complexity. Anyway, these guys are good. Talk with one of their strategies for free today. Call 1-800-958-1000 or go to tnusa.com slash Lex. (00:02:10) This episode is brought to you by BetterHelp, spelled H-E-L-P, help. I got to recently meet a lot of interesting people when I visited San Francisco. I was there in part to celebrate Yoshi Bach and the newly launched California Institute for Machine Consciousness. I, by the way, encourage you to check it out. I think it's CIMC.ai. And there I talked to a lot of brilliant people, and one of them was a grad student studying the so-called dark triad. (00:02:38) These are the three personality traits of narcissism, Machiavellianism, and psychopathy. A little bit for a brief moment made me wish I took that path of studying the human mind. And perhaps that is the indirect way, through all the AI, through all the programming, through all the building of systems, and now the podcast. Maybe I somehow sneaked up to that dream in the end. Anyway, I say all that because (00:03:06) These topics are studying the extremes of the human mind. But of course, the extremes are just the edges of an incredibly complicated system. That's just so fascinating to study, to reflect on, to put a mirror to all those processes that you do through talk therapy. It's just fascinating. Anyway, you can check them out at betterhelp.com/lex and save on your first month. That's betterhelp.com/lex. This episode is also brought to you by Element, my daily zero sugar and delicious electrolyte mix. (00:03:36) I'm not going to go down the rabbit hole, but there's a lot of interesting studies that measure the decreased performance of the human brain. So cognitive processing speed, for example, by what amount does it decrease? Reaction time, by what amount does it decrease when you decrease the brain's sodium levels, for example? Sodium and potassium really are important on a chemical level for the functioning of the human brain. Now, obviously, all throughout human history, people understood the value of water, but (00:04:06) As a medical concept, the concept of dehydration only came about in the 19th century. If we just look at the history of medicine, it's kind of hilarious how little we knew before. And it makes me think we know very little now relative to what we will know in a hundred and a thousand years. The human body, the biological system of the human body is incredibly complicated. So for us, (00:04:30) to have the certainty that we sometimes exude about the human body, about what we understand, about disease, about health. It's kind of funny. Anyway, get a simple pack for free with any purchase. Try it. DrinkElement.com slash flex. This episode is also brought to you by Shopify, a platform designed for anyone to sell anywhere with a great looking online store. Once again, I do this often. (00:04:57) where I don't just or at all talk about Shopify, but instead talk about the CEO of Shopify, Toby. He once again, like I mentioned, with Yoshi Bach and the newly launched CIMC, California Institute of Machine Consciousness. He's a big supporter of that too. And a bunch of people have asked me why I have not done a podcast with him yet. I don't know either. I'm sure it's going to happen soon. And I haven't seen him in quite a while. (00:05:27) A lot of people from a lot of walks of life deeply respect him for his intellect, for the way he does business, and just for the human being he is. So anyway, not sure why I mentioned that here, but back to what this is supposed to be. You can sell shirts online, like I did, lxfreatment.com slash shop. It's super easy to set up a store. I did it in a few minutes. What else can I say? (00:05:51) You should do it too. Sign up for $1 per month trial period at shopify.com/lex. That's all lowercase. Go to shopify.com/lex to take your business to the next level today. This episode is also brought to you by AG1, an all-in-one daily drink to support better health and peak performance. I was training jiu-jitsu the other day in that wonderful Texas heat, and I was reminded, first of all, how long my journey with jiu-jitsu has been. (00:06:19) and how fulfilling it has been. How interesting the exploration of the puzzle of two humans trying to break each other's arms and legs, plus the wrestling and the grappling component. Really interesting. Leverage, power, speed, how all that could be neutralized. How to control a human body with leverage, with technique, as opposed to raw, generally, misapplied strength, I should say. (00:06:49) Anyway, because there are times where there's long stretches of weeks where I don't train, you feel it in the cardio. You do a bunch of rounds and you just, the breaths are shallow. You feel like the mind is hazy from exhaustion, that you're a little bit more risk averse because you don't want to end up in a bad position, have to battle out of that bad position after many rounds of exhausting battles. And after that training session, when I got home, I enjoyed (00:07:18) and nice cold AG1. They'll give you a one-month supply of fish oil when you sign up at drinkag1.com/lex. This is the Lex Freeman Podcast. To support it, please check out our sponsors in the description or at lexfreeman.com/sponsors. And now, dear friends, here's Sundar Buchai. (00:07:56) Your life story is inspiring to a lot of people. It's inspiring to me. You grew up in India, whole family living in a humble two-room apartment, very little, almost no access to technology. And from those humble beginnings, you rose to lead a $2 trillion (00:08:17) technology company. So if you could travel back in time and told that, let's say, 12-year-old Sundar that you're now leading one of the largest companies in human history, what do you think that young kid would say? I would have probably laughed it off. You know, probably too far-fetched to imagine or believe at that time. You would have to explain the internet first. For sure. I mean, computers to me at that time, you know, I was 12 in 1984. (00:08:47) So probably, by then, I had started reading about them. I hadn't seen one. What was that place like? Take me to your childhood. I grew up in Chennai. It's in south of India. It's a beautiful, bustling city. Lots of people, lots of energy. You know, simple life, definitely. (00:09:07) fond memories of playing cricket outside the home. We just used to play on the streets. All the neighborhood kids would come out and we would play until it got dark and we couldn't play anymore barefoot. Traffic would come. We would just stop the game. Everything would drive through and you would just continue playing, right? Just to kind of get the visual in your head. You know, pre-computers, there's a lot of free time now that I think about it. Now you have to go and seek that (00:09:36) quiet solitude or something. Newspapers, books, is how I gained access to the world's information at the time, you will. My grandfather was a big influence. He worked in the post office. He was so good with language. His English, his handwriting, till today, is the most beautiful handwriting I've ever seen. He would write so clearly. He was so articulate. And so he kind of got me introduced into books. (00:10:05) He loved politics, so we could talk about anything. And that was there in my family throughout. So lots of books, trashy books, good books, everything from Ayn Rand to books on philosophy to stupid crime novels. So books was a big part of my life, but that kind of (00:10:26) This whole, it's not surprising I ended up at Google because Google's mission kind of always resonated deeply with me. This access to knowledge, I was hungry for it, but definitely have fond memories of my childhood. Access to knowledge was there. So that's the wealth we had. You know, 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 five years, but it's not the only thing. A telephone. There was a five-year waiting list. (00:10:57) 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 two hours to go, and they would say, sorry, it's not ready. Come back the next day, two hours to come back. And that became a five-minute thing. So as a kid, like, I mean, this light bulb went in my head, you know, this power of technology to kind of change people's lives. (00:11:26) We had no running water, it was a massive drought. So they would get water in these trucks, maybe 8 buckets per household. So me and my brother, sometimes my mom, we would wait in line, get that and bring it back home. Many years later, like we had running water and we had a water heater and you could get hot water to take a shower. I mean, like, so, you know, for me, everything was discreet like that. (00:11:56) And so I've always had this thing, firsthand feeling of like how technology can dramatically change like your life and like the opportunity it brings. So, that was kind of a subliminal takeaway for me throughout growing up. And, I kind of actually observed it and felt it, so (00:12:19) 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. But, you know, because before that, you only had like kind of 1 TV channel, right? That's it. And so, you know, you can watch movies or something like that, but this is by the time I was in 12th grade, we got a VCR, you know, it was a, (00:12:47) like a Panasonic, which we had to go to some like shop, which had kind of smuggled it in, I guess, and that's where we bought a VCR. But then being able to record like a World Cup football game and then, or get boot like videotapes and watch movies, like all that. So like, you know, I had these discreet memories growing up and so. (00:13:09) always left me with the feeling of like 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 have 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, (00:13:36) And you see that exponential growth to where most of the progress happened since the Industrial Revolution. And we just take 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. Oh, I mean, it's extraordinary. You know, I go back to India now, the power of mobile, you know, it's mind-blowing to see the progress through the arc of time. It's phenomenal. (00:14:05) 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 Sundar Pichai, who want to start, create companies, 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. 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 kind of get that (00:14:34) listen to your heart and see whether you actually enjoy doing it, right? 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 kind of listening to your heart a bit more than your mind in terms of figuring out what you want to do, I think is one of the best things I would tell people. (00:15:04) The second thing is, I mean, trying to work with people who you feel at various points in my life, I've worked with people who I felt were better than me. I kind of like, you know, you almost are sitting in a room talking to someone and they're like, wow, like, you know, and you want that feeling a few times. Trying to get yourself in a position where you're working with people who you feel are kind of like stretching your abilities is what helps you grow, I think. (00:15:34) So putting yourself in uncomfortable situations. And I think often you'll surprise yourself. So I think being open-minded enough to kind of put yourself in those positions is maybe another thing I would say. Well, lessons can we learn maybe from an outsider perspective? For me, looking at your story and 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 climbs to the very top of leadership, they're in a cut (00:16:03) throat world, they're usually going to be a bit of an asshole. So what wisdom are we supposed to draw from the fact that your general approach is of balance, of humility, of kindness, listening to everybody? What's your secret? I do get angry. I do get frustrated. I have the same emotions all of us do, right, in the context of work and everything. But a few things, right? I think, you know, I (00:16:33) Over time, I figured out the best way to get the most out of people. you kind of find mission-oriented people who are on the shared journey, who have this inner drive to excellence, to do the best, and you kind of motivate people and you can achieve a lot that way. (00:16:54) Right. And so it often tends to work out that way. But have there been times like, I lose it? but not 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 shit has not been productive. Yeah, less often than not. I think people respond to that. Yeah. (00:17:19) they may do stuff to react to that, like what you actually want them to do the right thing. And so, maybe there's a bit of like sports, I'm a sports fan. In football coaches, in soccer, that football, people often talk about like man management, right? Great coaches too, 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? (00:17:46) at times you're working with people who are so committed to achieving, if they've done something wrong, they feel it more than you do, right? So you treat them differently than, occasionally there are people who you need to clearly let them know, like that was an okay or whatever it is. But I've often found that not to be the case. And sometimes the right words at the right time, spoken firmly, can reverberate through time. Also, sometimes the unspoken words. (00:18:15) people can sometimes see that, like, you're unhappy without you saying it. And so sometimes the silence can deliver that message even more. Sometimes less is more. (00:18:28) Who's the greatest soccer player of all time, Messi or Ronaldo or Pele or Maradona? I'm going to make, you know, in this question. Is this going to be a political answer? No, I will tell the truthful answer, because it's Messi. It is. You know, it's been interesting because my son is a big Cristiano Ronaldo fan. And so we've had to watch El Clasicos together, you know, with that dynamic in there. (00:18:55) I so admire C.R. Simmons. I mean, I've never seen an athlete more committed to that kind of excellence. And so he's one of the all-time greats. But you know, for me, Messi is it. Yeah, when I see Leonardo Messi, you just are in awe that humans are able to achieve that level of greatness and genius and artistry. When we talk, we'll talk about AI, maybe robotics and this kind of stuff. That level of genius, I'm not sure you can possibly match. (00:19:25) by AI in a long time. It's just an example of greatness. And you have that kind of greatness in other disciplines. But in sport, you get to visually see it, unlike anything else. And just the timing, the movement, it's just genius. I had the chance to see him a couple of weeks ago. He played in San Jose, so against the Quake. So I went to see it, see the game. I was a fan on the (00:19:52) had good seats, knew where he would play in the second half, hopefully. And 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's still got it. If we rank all the technological innovations throughout human history, let's go back. Maybe the history of human civilization is 12,000 years ago. And you rank them by the (00:20:22) how much of A productivity multiplier there have been. So we can go to electricity or the labor mechanization of the Industrial Revolution, or we can go back to the first agricultural revolution to 1000 years ago. In that long list of inventions, do you think AI, when history is written 1000 years from now, do you think (00:20:43) it 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, you know, I said at the time, like, you know, 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, you know, I still think that's the case. You know, when you asked this question, I was thinking, well, (00:21:07) Do we have a recency bias, right? Like in sports, it's very tempting to call the current person you're seeing the greatest player, right? And so is there a recency bias? And I do think 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 people didn't have anesthesia when they went through these procedures. (00:21:37) At that moment, I was like, that has got to be the greatest invention humanity has ever done, right? So look, we don't know what it is to have lived through those times. But, you know, many of what you're talking about were kind of this general things which pretty much affected everything, you know, electricity or internet, et cetera. But I don't think we've ever dealt with the technology, both which is (00:22:04) progressing so fast, becoming so capable, it's not clear what the ceiling is. And the main unique, it's recursively self-improving, right? It's capable of that. And so the fact it is going, it's the first technology will kind of dramatically accelerate creation itself, like creating things, building new things, can improve and achieve things on its own. (00:22:33) Right, I think puts it in a different league, right? And so different league. And so 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 and wrestle with, but I definitely think that'll end up being the case. Especially if it gets to the point of where we can achieve superhuman performance on the AI research itself. So it's the technology (00:23:00) that may, it's an open question, but it may be able to achieve a level to where the technology itself can create itself better than it could yesterday. It's like the move 37 of alpha research or whatever it is, right? Like, you know, and when, yeah, you're right, when it can do novel, self-directed research, obviously for a long time, we'll have hopefully always humans in the loop and all that stuff. And these are complex questions to (00:23:30) talk about, but yes, I think the underlying technology, I've said this, like if you watched seeing AlphaGo start from scratch, be clueless and like become better through the course of a day, you know, like, you know, kind of like, kind of like, you know, really hits you when you see that happen. Even our like the VO3 models, if you sample the models when they were like 30% done and 60% done and looked at what they were generating, (00:24:00) And you kind of see how it all comes together. It's kind of like, I would say, it's kind of inspiring, a little bit unsettling, right, as a human. So all of that is true, I think. Well, the interesting thing of the Industrial Revolution, electricity, like you mentioned, you can go back to the, again, the agriculture, the first agricultural revolution. There's what's called the Neolithic package of the first agricultural revolution. (00:24:29) It wasn't just that the nomads settled down and started planting food, but all this other kinds of technology was born from that and it's included in this package. It wasn't one piece of technology. It's there's these ripple effects, 2nd and 3rd order effects that happen. Everything from something silly, like silly, profound, like pottery, it can store liquids and food. (00:24:54) to something we kind of take for granted, but social hierarchies and political hierarchy. So like early government was formed. Because it turns out if humans stop moving and have some surplus food, they start coming up with, they get bored and they start coming up with interesting systems. And then trade emerges, which turns out to be a really profound thing. And like I said, government. I mean, there's just (00:25:21) Second and 3rd order effects from that, including that package, is incredible and probably extremely difficult. If you'll ask one of the people in the nomadic tribes to predict that, it would be impossible. It's difficult to predict. But all that said, what do you think are some of the early things we might see in the quote unquote AI package? I mean, most of it probably we don't know today, but like, you know, the one thing which we can tangibly start seeing now is (00:25:52) Obviously, with the coding progress, you got a sense of it. It's going to be so easy to imagine, like thoughts in your head, translating that into things that exist. That'll be part of the package, right? Like it's going to empower almost all of humanity to kind of express themselves. Maybe in the past, you could have expressed with words, but like, (00:26:17) you could kind of build things into existence, right? maybe not fully today. We are at the early stages of vipe coding. You know, I've been amazed at what people have put out online with VO3, 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, this is the worst it'll ever be, right? Like at any given moment in time. Yeah, it's interesting you went there as kind of a first thought. So the exponential increase (00:26:46) of access to creativity. Software creation, are you creating a program, a piece of content to be shared with others, 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. (00:27:15) Maybe in the US, there were five people who could do what you were doing, like go do an interview. But today, think about with YouTube and other products, et cetera, like how many more people are doing it? So I think this is what technology does, right? Like when the internet created blogs, you heard from so many more people. So I think (00:27:40) But with AI, I think that number won't be in the few hundreds of thousands. It'll be 10s of millions of people, maybe even a billion people, like putting out things into the world in a deeper way. And I think it'll change the landscape of creativity. And it makes a lot of people nervous. Like for example, (00:28:01) whatever. Fox, MSNBC, CNN are really nervous about this part. Like, you mean this dude in a suit could just do this and YouTube and thousands of others, 10s of thousands, millions of other creators can do the same kind of thing? That makes them nervous. And now you get a podcast from Notebook LM. It's about 5 to 10 times better than any podcast I've ever done. (00:28:25) I'm joking at this time, and 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 great podcasts that are both 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. (00:28:47) But it's also a really exciting future. The only thing I may say is I do think like in a world in which there are two AI, 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, like it would be boring for us to watch. But Magnus Carlsen and Gukesh, that game would be much more fascinating to watch. So it's tough to say, like one way to say is, (00:29:16) You'll have a lot more content, and so you will be listening to AI-generated content because sometimes it's efficient, et cetera. But the premium experiences you value might be a version of the human essence wherever it comes through. Going back to what we talked 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 (00:29:46) or audiobooks that is about information gathering, that part might be removed or that might be more efficiently and in a compelling way done by AI. But then it'll be just 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. But if you actually want to find out about a piece of history, (00:30:15) you go to Gemini. If you want to see Lex struggle with that history, then you look, or other humans, you look at that. But the point is it's going to change the nature, continue to change the nature of how we discover information, how we consume the information, how we create that information. The same way that YouTube changed everything completely, changed news. And that's something our society is struggling with. Yeah, YouTube, look, YouTube enabled (00:30:43) I mean, this better than anyone else. It's enabled so many creators. There is no doubt in me that like we will enable more filmmakers than that have ever been, right? You're going to empower a lot more people. So I think there is 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. (00:31:10) The only way it is if you brought someone from the 50s or 40s and just put them in front of YouTube, I think it would blow their mind away. Similarly, I think we would get 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 that, let's say, let's put a mark on it, 50% of content, good content, 50% of good content is generated by VO4, 5, 6? You know, I think it depends on what it is for. (00:31:42) maybe if you look at movies today with CGI, there are great filmmakers. Like you still look at like who the directors are and who use it. There are filmmakers who don't use it at all. You value that. There are people who use it incredibly. You know, think about somebody like a James Cameron, like what he would do with these tools in his hands. But I think there'll be a lot more content created, like just like writers today use Google Docs. (00:32:07) and not think about the fact that they're using a tool like that, like people will be using the future versions of these things, like 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, it's fun to watch a genius who came up before (00:32:30) 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. And now he's trying to see how can AI be used to create compelling films. You have people like that. You have people, I've got to just know edgier folks that are AI first, like Door Brothers. Both Aronofsky and Door Brothers create at the edge of (00:32:57) 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. You know, Hunter S. Thompson has this line that the only way to find out where the edge, where the line is, by crossing it. And I think for artists, that's true, that's kind of their purpose sometimes, comedians and artists just cross that line. (00:33:27) I wonder if you can comment on the weird place that puts Google, because Google's line is probably different than some of these artists. What's your, how do you think about specifically VO and Flow about like how to allow artists to do crazy shit, but also like the responsibility of like not for it not to be too crazy? I mean, it's a great question. Look, part of, you mentioned Darren, you know, (00:33:56) He's a clear visionary, right? Part of the reason we started working with them early on VO is he's one of those people who's able to kind of see that future, get inspired by it, and kind of showing the way for how creative people can express themselves with it. Look, I think when it comes to allowing artistic free expression, (00:34:20) It's one of the most important values in a society, right? I think artists have always been the ones to push boundaries, expand the frontiers of thought. And so, look, I think that's going to be an important value we have. So I think we will provide tools and put it in the hands of artists for them to use and put out their work. (00:34:47) Those APIs, I mean, 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. It's a paint brush. Yeah. And so I think that's how obviously there have to be some things and society needs to decide at a fundamental level what's okay, what's not. We'll be responsible with it. But I do think (00:35:14) when it comes to artistic free expression, I think that's one of those values we should work hard to defend. I wonder if you can comment on maybe earlier versions of Gemini were a little bit careful on the kind of things you would be willing to answer. I just want to comment on 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. (00:35:43) 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. And earlier versions of Gemini were very, basically this kind of sense, are you sure you want to learn about this? And now it's actually very factual, objective. (00:36:09) 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 kind of walk. I wonder if it's, and it's also an engineering challenge, how to do that at scale across all the weird queries that people ask. Can you just speak to that challenge? How do you allow Gemini to say, again, forgive, pardon my French, crazy shit, but not too crazy? I think one of the (00:36:39) good insights here has been, as the models are getting more capable, the models are really good at this stuff, right? And so I think in some ways, maybe a year ago, the models weren't fully there. So they would also do stupid things more often. And so, you know, you're trying to handle those edge cases, but then you make a mistake in how you handle those edge cases and it compounds. But I think with 2.5, what we particularly found is once the models (00:37:08) across a certain level of intelligence and sophistication, they are able to reason through these nuanced issues pretty well. And I think users really want that, right? Like, you want as much access to the raw model as possible, right? But I think it's a great area to think about, like, you know, over time, you know, we should allow more and more closer access to it. (00:37:31) maybe obviously let people custom prompts if they wanted to and like, and experiment with it, et cetera. I think that's an important direction. But look, the first principles we want to think about it is, from a scientific standpoint, like making sure the models, and I'm saying scientific in the sense of like how you would approach math or physics or something like that, from first principles, having the models reason about the world, be nuanced, (00:38:00) et cetera, from the ground up, is the right way to build these things, right? Not like some subset of humans kind of hard coding things on top of it. So 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 gave these notes, I took extensive notes, and I gave them to Gemini and said, can you ask a novel question that's not in these notes? (00:38:28) And it wrote, Gemini continues to really surprise me, 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 churning out 480 trillion tokens a month. (00:38:48) What's the most life-changing five-word sentence hiding in that haystack? That's A Gemini question. But it gave me a sense, I don't think you can answer that, but it gave me, it woke me up to like, all of these tokens are providing little aha moments for people across the globe. So that's like learning. Those tokens are, people are curious, they ask a question, and they find something out. And it truly could be life-changing. Oh, it is. Say, look, you know. (00:39:17) I had the same feeling about search many, many years ago. You definitely, this tokens per month has like grown 50 times in the last 12 months. Is that accurate by the way? Yeah, it is. You know, it is. It is accurate. I'm glad it got it right. But you know, that number was 9.7 trillion tokens per month 12 months ago. Right. It's gone up to 480. You know, it's a 50x increase. So there's no limit to human curiosity. (00:39:46) And I think it's one of those moments. Maybe, I don't think it is there today, but maybe one day there's a five-word phrase which says what the actual universe is or something like that and something very meaningful. (00:40:00) But I don't think we are quite there yet. (00:40:02) Do you think the scaling laws are holding strong on? (00:40:07) There's a lot of ways to describe the scaling laws for AI, but on the pre-training, on the post-training fronts. (00:40:13) So the flip side of that, do you anticipate AI progress will hit a wall? (00:40:18) Is there a wall? (00:40:19) Yeah, it's a cherished micro kitchen conversation once in a while. (00:40:23) I have it, you know, like when Demis is visiting or, you know, (00:40:28) Demis, Korai, Jeff, Noam, Sergey, a bunch of our people, like we sit and talk about this, right? (00:40:35) And look, we see a lot of headroom ahead, right? (00:40:40) I think we've been able to optimize and improve on all fronts, right? (00:40:46) Pre-training, post-training, test time, compute, tool use, right, over time, making these more agentic. (00:40:55) So (00:40:56) getting these models to be more general world models in that direction, like VO3, you know, the physics understanding is dramatically better than what VO1 or something like that was. (00:41:08) So you kind of see on all those dimensions, I feel, you know, progress is very obvious to see. (00:41:15) And I feel like there is significant headroom. (00:41:21) More importantly, you know, I'm fortunate to work with some of the (00:41:25) best researchers on the planet, right? (00:41:27) They think there is more headroom to be had here. (00:41:31) And so I think we have an exciting trajectory ahead. (00:41:34) It's tougher to say, you know, each year I sit and say, okay, we're going to throw 10X more compute over the course of next year at it. (00:41:42) And like, will we see progress? (00:41:45) Sitting here today, I feel like the year ahead will have a lot of progress. (00:41:49) And do you feel any limitations like that or the bottlenecks (00:41:54) Compute limited, data limited, idea limited. (00:41:57) Do you feel any of those limitations or is it full steam ahead on all fronts? (00:42:01) I think it's compute limited in this sense, right? (00:42:03) Like, you know, we can all, part of the reason you've seen us do flash, nano flash and pro models, but not an ultra model. (00:42:12) It's like for each generation, we feel like we've been able to get the pro model at like, I don't know, 80, 90% of ultra's capability, but ultra would be a, (00:42:23) a lot more slow and a lot more expensive to serve. (00:42:30) But what we've been able to do is to 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 and so on. (00:42:42) So I do think scaling laws are working, but it's tough to get at any given time the models we all use the most. (00:42:53) is maybe like a few months behind the maximum capability we can deliver, right? (00:43:00) Because that won't be the fastest, easiest to use, et cetera. (00:43:04) Also, that's in terms of intelligence. (00:43:06) It becomes harder and harder to measure performance in quotes. (00:43:11) Because, you know, you could argue Gemini Flash is much more impactful than Pro. (00:43:17) Just because of the latency, it's super intelligent already. (00:43:21) I mean, sometimes like latency is maybe more important than intelligence, especially when the intelligence is just a little bit less and flash not, it's still incredibly smart model. (00:43:33) And so you have to now start measuring impact. (00:43:35) And then it feels like benchmarks are less and less capable of capturing the intelligence of models, the effectiveness of models, the usefulness, the real world usefulness of models. (00:43:45) Another kitchen question. (00:43:47) So lots of folks are talking about timelines for AGI. (00:43:51) or ASI, artificial superintelligence. (00:43:54) So AGI loosely defined is basically human expert level at a lot of the main fields of pursuit for humans. (00:44:02) And ASI is what AGI becomes presumably quickly by being able to self-improve. (00:44:10) So becoming far superior in intelligence across all these disciplines in humans. (00:44:15) When do you think we'll have AGI? (00:44:16) Is 2030 a possibility? (00:44:19) There's one other term we should throw in there. (00:44:21) I don't know who used it first. (00:44:23) Maybe Karpati did AJI. (00:44:25) Have you heard AJI, the artificial jagged intelligence? (00:44:29) Sometimes feels that way, right? (00:44:31) Both there are progress and you see what they can do. (00:44:34) And then you can trivially find they make numerical errors or counting Rs in strawberry or something which seems to trip up most models or whatever it is, right? (00:44:44) So (00:44:46) So maybe we should throw that term in there. (00:44:47) I feel like we are in the AGI phase where like dramatic progress, some things don't work well, but overall, you're seeing lots of progress. (00:44:57) But if your question is, will it happen by 2030? (00:45:01) Look, we constantly move the line of what it means to be AGI. (00:45:07) There are moments today, you know, like sitting in a Waymo in a San Francisco street with all the crowds and the people and kind of work its way through. (00:45:16) I see glimpses of it there. (00:45:18) The car is sometimes kind of impatient, trying to work its way, using Astra, like in Gemini Live or seeing, you know, asking questions about the world. (00:45:27) What's this skinny building doing in my neighborhood? (00:45:29) It's a streetlight, not a building. (00:45:32) You see glimpses. (00:45:34) That's why I use the word AGI, because then you see stuff which obviously, you know, we are far from AGI too. (00:45:41) So you have both experiences simultaneously happening to you. (00:45:45) I'll answer your question, but I'll also throw out this. (00:45:47) I almost feel the term doesn't matter. (00:45:49) What I know is by 2030, there'll be such dramatic progress. (00:45:54) We'll be dealing with the consequences of that progress, both the positives, both the positive externalities and the negative externalities that come with it in a big way by 2030. (00:46:06) So that I strongly feel. (00:46:08) right? (00:46:09) Whatever, we may be arguing about the term, or maybe Gemini can answer what that moment is in time in 2030. (00:46:16) But I think the progress will be dramatic, right? (00:46:19) So that I believe in. (00:46:20) Will the AI think it has reached AGI by 2030? (00:46:24) I would say we will just fall short of that timeline, right? (00:46:27) So I think it'll take a bit longer. (00:46:29) It's amazing in the early days of Google DeepMind in 2010, they talked about a 20-year timeframe to achieve (00:46:35) AGI, so which is kind of fascinating to see. (00:46:39) But, for me, the whole thing, seeing what Google Brain did in 2012 and when we acquired DeepMind in 2014, right close to where we are sitting in 2012, Jeff Dean showed the image of when the neural networks could recognize a picture of a cat, right, and identify it. (00:47:00) You know, this is the early versions of Brain, right? (00:47:02) And so (00:47:04) we all talked about a couple of decades. (00:47:07) I don't think we'll quite get there by 2030. (00:47:10) So my sense is it's slightly after that. (00:47:13) But I would stress it doesn't matter what that definition is, because you will have mind-blowing progress on many dimensions. (00:47:22) Maybe AI can create videos. (00:47:25) We have to figure out as a society, how do we, need some system by which (00:47:29) We all agree that this is AI generated and we have to disclose it in a certain way because how do you distinguish reality otherwise? (00:47:36) Yeah, there's so many interesting things you said. (00:47:37) So first of all, just looking back at this recent, now it feels like distant history with Google Brain. (00:47:43) I mean, that was before TensorFlow, before TensorFlow was made public and open sourced. (00:47:48) So the tooling matters too, combined with GitHub ability to share code. (00:47:53) Then you have the ideas of attention transformers and the diffusion now. (00:47:57) And then there might be a new idea that seems simple in retrospect, but will change everything. (00:48:03) And that could be the post-training, the inference, time, innovations. (00:48:06) And I think Shad CN tweeted that Google is just one great UI from completely winning the AI race, meaning like UI is a huge part of it. (00:48:17) Like how that intelligence (00:48:21) I think Logan Kopratrick likes to talk about this right now. (00:48:23) It's an LLM, but it become, like, when is it going to become a system where you're talking about shipping systems versus shipping a particular model? (00:48:32) Yeah, that matters too, how the system manifests itself and how it presents itself to the world. (00:48:38) That really, really matters. (00:48:40) Oh, hugely so. (00:48:42) There are simple UI innovations which have changed the world, right? (00:48:46) And I absolutely think so. (00:48:50) We will see a lot more progress in the next couple of years as I think AI itself on a self-improving track for UI itself. (00:48:59) Like today, we are like constraining the models. (00:49:03) The models can't quite express themselves in terms of the UI to people. (00:49:10) But that is like, you know, if you think about it, we've kind of boxed them in that way. (00:49:15) But given these models can code, (00:49:19) they should be able to write the best interfaces to express their ideas over time, right? (00:49:25) That is an incredible idea. (00:49:27) So the API is already open. (00:49:29) So you create a really nice agentic system that continually improves the way you can be talking to an AI. (00:49:38) But a lot of that is in the interface. (00:49:41) And then, of course, the incredible multimodal aspect of the interface that Google has been pushing. (00:49:46) These models are natively multimodal. (00:49:49) They can easily take content from any format, put it in any format. (00:49:53) They can write a good user interface. (00:49:55) They probably understand your preferences better over time. (00:49:59) Like, you know, and so all this is like the evolution ahead, right? (00:50:03) And so that goes back to where we started the conversation, right? (00:50:08) Like, I think there'll be dramatic evolutions in the years ahead. (00:50:12) Maybe one more kitchen question. (00:50:14) This even further ridiculous concept. (00:50:18) of P doom. (00:50:19) So the philosophically minded folks in the AI community think about the probability that AGI and then ASI might destroy all of human civilization. (00:50:30) I would say my P doom is about 10%. (00:50:32) Do you ever think about this kind of long-term threat of ASI? (00:50:39) And what would your P doom be? (00:50:41) Look, I mean, for sure. (00:50:42) Look, I've both been very excited about AI, but I've always felt this is a technology, you know, 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. (00:51:00) On the PDOM question, look, it's, you know, wouldn't surprise you to say that's probably another micro kitchen conversation that pops up once in a while, right? (00:51:08) And (00:51:09) Given how powerful the technology is, maybe stepping back, when you're running a large organization, if you can kind of align the incentives of the organization, you can achieve pretty much anything, right? (00:51:19) Like, if you can get kind of people all marching in towards like a goal in a very focused way, in a mission-driven way, you can pretty much achieve anything. (00:51:28) But it's very tough to organize all of humanity that way. (00:51:32) But I think if speedom is actually high, at some point, all of humanity is like, (00:51:38) aligned in making sure that's not the case, right? (00:51:40) And so we'll actually make more progress against it, I think. (00:51:44) So the irony is, so there is a self-modulating aspect there. (00:51:50) Like I think if humanity collectively puts their mind to solving a problem, whatever it is, I think we can get there. (00:51:57) So because of that, you know, I think I'm optimistic on the P-Doom scenarios, but that doesn't mean (00:52:07) I think the underlying risk is actually pretty high. (00:52:11) But I have a lot of faith in humanity kind of rising up to meet that moment. (00:52:17) That's really, really what I put. (00:52:18) I mean, as the threat becomes more concrete and real, humans do really come together and get their shit together. (00:52:26) Well, the other thing I think people don't often talk about is probability of doom. (00:52:31) without AI. (00:52:32) So there's all these other ways that humans can destroy themselves, and it's very possible, at least I believe so, that AI will help us become smarter, kinder to each other, more efficient. (00:52:46) It'll help more parts of the world flourish where it wouldn't be less resource constrained, which is often the source of military conflict and tensions and so on. (00:52:56) So we also have to load into that, what's the P doom without AI? (00:53:01) with AI, be doing with AI, be doing without AI, because it's very possible that AI will be the thing that saves us, saves human civilizations from all the other threats. (00:53:10) I agree with you. (00:53:11) I think it's insightful. (00:53:12) Look, I felt like to make progress on some of the toughest problems would be good to have AI like pair helping you, right? (00:53:21) And like, you know, so that resonates with me for sure. (00:53:25) Yeah. (00:53:26) Quick pause, bathroom break? (00:53:27) I know. (00:53:28) Let's do that. (00:53:31) If Notebook LM was the same, like what I saw today with Beam, if it was compelling in the same kind of way. (00:53:36) How was Beam? (00:53:39) Blew my mind. (00:53:40) It was incredible. (00:53:41) I didn't think it's possible. (00:53:42) I didn't think it's possible. (00:53:44) Can you imagine like the US president and the Chinese president being able to do something like Beam with the live Meet translation working well? (00:53:52) So they're both sitting and talking, make progress a bit more. (00:53:58) Yeah, just for people listening, we took a quick bathroom break and now we're talking about the demo I did. (00:54:03) We'll probably post it somewhere, somehow, maybe here. (00:54:07) I got a chance to experience Beam and it was, it's hard to describe in words how real it felt with just, what is it, 6 cameras. (00:54:18) It's incredible, it's incredible. (00:54:20) It's one of the toughest products of, you can't quite describe it to people. (00:54:25) even when we show it in slides, et cetera, like you don't know what it is. (00:54:30) You have to kind of experience it. (00:54:32) On the world leaders front, on politics, geopolitics, there's something really special, again, with studying World War II and how much could have been saved if Chamberlain met Stalin in person. (00:54:45) And I sometimes also struggle explaining to people, articulating why I believe meeting in person for world leaders is powerful. (00:54:53) It just seems naive to say that, but there is something there in person. (00:54:58) And with Beam, I felt that same thing. (00:55:00) And then I'm unable to explain. (00:55:04) All I kept doing is what like a child does. (00:55:06) You look real. (00:55:08) You know, and I mean, I don't know if that makes meetings more productive or so on, but it certainly makes them more (00:55:17) The same reason you want to show up to work versus remote sometimes, that human connection. (00:55:23) I don't know what that is. (00:55:24) It's hard to put into words. (00:55:28) There's something beautiful about great teams collaborating on a thing that's not captured by the productivity of that team or by whatever on paper. (00:55:41) Some of the most beautiful moments you experience in life is at work. (00:55:45) pursuing a difficult thing together for many months, there's nothing like it. (00:55:51) You're in the trenches and yeah, you do form bonds that way for sure. (00:55:55) And to be able to do that like somewhat remotely in that same personal touch, I don't know, that's a deeply fulfilling thing. (00:56:01) I know a lot of people, I personally hate meetings because a significant percent of meetings when done poorly don't serve a clear purpose. (00:56:11) But that's a meeting problem, that's not a communication problem. (00:56:15) If you can improve the communication for the meetings that are useful, it's just incredible. (00:56:19) So yeah, I was blown away by the great engineering behind it. (00:56:23) And then we get to see what impact that has. (00:56:26) That's really interesting, but just incredible engineering. (00:56:28) Really impressive. (00:56:29) It is. (00:56:29) And obviously, we'll work hard over the years to make it more and more accessible. (00:56:34) But yeah, even on a personal front, outside of work meetings, you know, a grandmother who's far away from our grandchild and being able to (00:56:43) have that kind of an interaction, right? (00:56:46) All of that, I think, will end up being very mean. (00:56:48) Nothing substitutes being in person. (00:56:51) You know, it's not always possible. (00:56:53) You know, you could be a soldier deployed, try trying to talk to your loved ones. (00:56:58) So I think, you know, so that's what inspires us. (00:57:02) 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 (00:57:13) outside of that, seeing dozens of articles written by analysts and experts and so on, that Sundar Pichai should step down because the perception was that Google was definitively losing the AI race, has lost its magic touch in the rapidly evolving technological landscape. (00:57:34) And now a year later, it's crazy. (00:57:36) You showed this plot (00:57:38) of all the things that were shipped over the past year. (00:57:41) It's incredible. (00:57:43) And Gemini Pro is winning across many benchmarks and products as we sit here today. (00:57:47) So take me through that experience when there's all these articles saying, you're the wrong guy to lead Google through this. (00:57:55) Google is lost, it's done, it's over, to today where Google is winning again. (00:58:02) What were some low points during that time? (00:58:05) Look, I (00:58:08) I mean, lots to unpack. (00:58:10) obviously, the main bet I made as a CEO was to really, make sure the company was approaching everything in a AI-first way, really, setting ourselves up to develop AGI responsibly, right? (00:58:29) And make sure we're putting out products which embodies that, things that are very, very useful for people. (00:58:38) So look, I knew even through moments like that last year, I had a good sense of what we were building internally, right? (00:58:49) So I'd already made many important decisions, bringing together teams of the caliber of Brain and DeepMind and setting up Google DeepMind. (00:59:01) There were things like we made the decision to invest in TPUs 10 years ago. (00:59:07) So we knew we were scaling up and building big models. (00:59:11) Anytime you're in a situation like that, a few aspects. (00:59:17) I'm good at tuning out noise, right? (00:59:19) Separating signal from noise. (00:59:21) Do you scuba dive? (00:59:22) Like, have you? (00:59:23) No. (00:59:23) You know, it's amazing, like, I'm not good at it, but I've done it a few times. (00:59:29) But sometimes you jump in the ocean, it's so choppy. (00:59:34) but you go down 1 feet under, it's the calmest thing in the entire universe, right? (00:59:41) So there's a version of that, right? (00:59:43) Like, running Google, you may as well be coaching Barcelona or Real Madrid, right? (00:59:51) Like, you know, you have a bad season. (00:59:54) So there are aspects to that. (00:59:55) But, you know, like, look, I'm good at tuning out the noise. (01:00:00) I do watch out for signals. (01:00:01) You know, it's important to separate the signal from the noise. (01:00:04) So there are good people sometimes making good points outside, so you want to listen to it, you want to take that feedback in. (01:00:11) But internally, you're making a set of consequential decisions. (01:00:17) As leaders, you're making a lot of decisions. (01:00:21) Many of them are inconsequential. (01:00:24) It feels like, but over time, you learn that. (01:00:27) Most of the decisions you're making on a day-to-day basis doesn't matter. (01:00:33) You have to make them and you're making them just to keep things moving. (01:00:36) But you have to make a few consequential decisions, right? (01:00:39) And we had set up the right teams, right leaders. (01:00:47) We had world-class researchers. (01:00:50) We were training Gemini. (01:00:53) Internally, there are factors which were, for example, outside people may not have appreciated. (01:00:57) I mean, TPUs are amazing, but we had to ramp up TPUs too. (01:01:02) That took time, right? (01:01:04) And to scale, actually having enough TPUs to get the compute needed. (01:01:11) But I could see internally the trajectory we were on. (01:01:15) And I was so excited internally about the possibility. (01:01:21) To me, this moment felt like one of the biggest opportunities ahead for us as a company. (01:01:26) that the opportunity space ahead over the next decade, next 20 years is bigger than what has happened in the past. (01:01:35) And I thought we were set up like better than most companies in the world to go realize that vision. (01:01:42) I mean, you had to make some consequential, bold decisions. (01:01:47) Like you mentioned the merger of DeepMind and Brain. (01:01:53) Maybe it's my perspective, just knowing humans. (01:01:56) I'm sure there's a lot of egos involved. (01:01:58) It's very difficult to merge teams, and I'm sure there are some hard decisions to be made. (01:02:03) Can you take me through your process of how you think through that? (01:02:06) Do you go to pull the trigger and make that decision? (01:02:09) Maybe what were some painful points? (01:02:10) How do you navigate those turbulent waters? (01:02:14) Look, we were fortunate to have two world-class teams, but you're right. (01:02:17) It's like somebody coming and telling to you, (01:02:20) take Stanford and MIT, and then put them together and create a great department, right? (01:02:25) And easier said than done. (01:02:28) But we were fortunate, phenomenal teams, both had their strengths, but they were run very differently, right? (01:02:35) Like brain was kind of a lot of diverse projects, bottoms up, and out of it came a lot of important research breakthroughs. (01:02:45) DeepMind at the time had a strong, (01:02:49) vision of how you want to build AGI. (01:02:50) And so they were pursuing their direction. (01:02:54) But I think through those moments, luckily tapping into, you know, Jeff had expressed a desire to be more, to go back to more of a scientific individual contributor roots. (01:03:06) You know, he felt like management was taking up too much of his time. (01:03:10) And Demis naturally, I think, you know, was running DeepMind and was a natural choice there. (01:03:19) But I think it was, you're right, it took us a while to bring the teams together. (01:03:22) Credit to Demis, Jeff, Korai, all the great people there. (01:03:27) They worked super hard to combine the best of both worlds when you set up that team. (01:03:34) A few sleepless nights here and there as we put that thing together. (01:03:38) We were patient in how we did it so that it works well for the long term, right? (01:03:44) And some of that in that moment, I think, yes, (01:03:48) With things moving fast, I think you definitely felt the pressure, but I think we pulled off that transition well, and I think they've obviously doing incredible work, and there's a lot more incredible things I had coming from them. (01:04:04) Like we talked about, you have a very calm, even-tempered, respectful demeanor. (01:04:09) During that time, whether it's the merger or just dealing with the noise, (01:04:16) Did were there times where frustration boiled over? (01:04:19) Like, did you have to go a bit more intense on everybody than you usually would? (01:04:26) Probably, you're right. (01:04:28) I think in the sense that, there 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, right? (01:04:39) You disagree, you argue, but like all that, I mean, just (01:04:44) part of the course of working intensely, right? (01:04:47) And at the end of the day, all of us are doing what we are doing because the impact it can have, we are motivated by it. (01:04:57) It's like, for many of us, this has been a long-term journey. (01:05:03) And so it's been super exciting. (01:05:05) The positive moments far outweigh the kind of stressful moments just early this year. (01:05:11) I had a chance to celebrate back-to-back over two days, like, you know, Nobel Prize for Jeff Finton and the next day a Nobel Prize for Dennis and John Jumper. (01:05:22) You know, you worked with people like that. (01:05:25) All that is super inspiring. (01:05:26) Is there something like with you where you had to like put your foot down maybe with less versus more or like I'm the CEO and we're doing this? (01:05:39) To my earlier point about consequential decisions you make, there are decisions you make, people can disagree pretty vehemently. (01:05:45) But at some point, you make a clear decision and you just ask people to commit, right? (01:05:55) Like, you can disagree, but it's time to disagree and commit so that we can get moving. (01:06:02) whether it's putting the foot down or, it's a natural part of what all of us have to do. (01:06:07) And, I think you can do that calmly and be very firm in the direction you're making the decision. (01:06:14) And I think if you're clear, actually people over time respect that, right? (01:06:18) Like, if you can make decisions with clarity. (01:06:21) I find it very effective in meetings where you're making such decisions to hear everyone out. (01:06:28) I think it's important. (01:06:30) when you can to hear everyone out. (01:06:32) Sometimes what you're hearing actually influences how you think about and you're wrestling with it and making a decision. (01:06:39) Sometimes you have a clear conviction, and you state so. (01:06:42) Look, this is how I feel, and this is my conviction, and you kind of place the bet and you move on. (01:06:51) Are there big decisions like that? (01:06:52) I'm kind of intuitively assumed the merger was the big one. (01:06:57) I think that was a very important decision. (01:06:59) for the company to meet the moment, I think we had to make sure we were doing that and doing that well. (01:07:07) I think there was a consequential decision. (01:07:09) There were many other things. (01:07:10) We set up an AI infrastructure team, like to really go meet the moment, to scale up the compute we needed to, and really brought teams from disparate parts of the company, kind of created it to move forward. (01:07:27) You know, bringing people, like, (01:07:29) getting people to kind of work together physically, both in London with DeepMind and what we call Gradient Canopy, which is where the Mountain View, Google DeepMind teams are. (01:07:41) But one of my favorite moments is I routinely walk multiple times per week to the Gradient Canopy building where our top researchers are working on the models. (01:07:53) Sergey is often there amongst them, right? (01:07:56) Just looking at (01:08:00) getting an update on the model, seeing loss curve, so all that. (01:08:02) I think the cultural part of getting the teams together back with that energy, I think ended up playing a big role too. (01:08:10) What about the decision to recently add AI mode? (01:08:14) So Google search is the, as they say, the front page of the internet. (01:08:20) It's like a legendary minimalist (01:08:23) thing with 10 blue links. (01:08:25) Like that's, when people think internet, they think that page. (01:08:29) And now you're starting to mess with that. (01:08:32) So the AI mode, which is a separate tab, and then integrating AI and the results, I'm sure there were some battles in meetings on that one. (01:08:40) Look, you know, in some ways when mobile came, you know, people wanted answers to more questions. (01:08:47) So we're kind of constantly evolving it. (01:08:49) But you're right, this moment, you know, that evolution (01:08:53) because the underlying technology is becoming much more capable. (01:08:58) You can have AI give a lot of context. (01:09:01) But one of our important design goals, though, is when you come to Google search, you're going to get a lot of context, but you're going to go and find a lot of things out on the web. (01:09:11) So that will be true in AI mode, in AI overviews, and so on. (01:09:17) But I think to our earlier conversation, we are still giving you access to links. (01:09:21) But think of the AI as a layer which is giving you context, summary. (01:09:27) Maybe in AI mode, you can have a dialogue with it back and forth on your journey, right? (01:09:34) But through it all, you're kind of learning what's out there in the world. (01:09:37) So those core principles don't change. (01:09:40) But I think AI mode allows us to push the, we have our best models there, right? (01:09:46) Models which are using search as a deep tool. (01:09:50) really for every query you're asking, kind of fanning out, doing multiple searches, like kind of assembling that knowledge in a way so you can go and consume what you want to, right? (01:10:00) And that's how we think about it. (01:10:03) I got a chance to listen to a bunch of Elizabeth, Liz Reed, describe this. (01:10:08) Two things stood out to me that you mentioned. (01:10:10) One thing is what you were talking about is the query fan out, which I didn't even think about before. (01:10:18) is the powerful aspect of integrating a bunch of stuff on the web for you in one place. (01:10:23) So yes, it provides that context so that you can decide which page to then go on to. (01:10:29) The other really, really big thing speaks to the earlier in terms of productivity multiplier that we're talking about that she mentioned was language. (01:10:39) So one of the things you don't quite understand is it through AI mode, you make (01:10:46) For non-English speakers, you make sort of, let's say, English language websites accessible by in the reasoning process as you try to figure out what you're looking for. (01:10:58) Of course, once you show up to a page, you can use a basic translate. (01:11:02) But that process of figuring it out, if you empathize with a large part of the world that doesn't speak English, their like web is much smaller. (01:11:13) in that original language. (01:11:15) And so it unlocks, again, unlocks that huge cognitive capacity there. (01:11:20) We don't, you know, you take for granted here with all the bloggers and the journalists writing about AI mode, you forget that this now unlocks, because Gemini is really good at translation. (01:11:31) No, it is. (01:11:32) I mean, the multimodality, the translation, its ability to reason, we are dramatically improving tool use. (01:11:41) as putting that power in the flow of search, I think, look, I'm super excited with AI overviews. (01:11:50) We've seen the product has gotten much better. (01:11:53) We measure it using all kinds of user metrics. (01:11:57) It's obviously driven strong growth of the product. (01:12:01) And we've been testing AI mode. (01:12:04) It's now in the hands of millions of people. (01:12:09) And the early metrics are very encouraging. (01:12:10) So look, I'm excited about this next chapter of search. (01:12:14) For people who are not thinking through or aware of this. (01:12:16) So there's the 10 blue links with the AI overview on top that provides a nice summarization. (01:12:22) You can expand it. (01:12:23) And you have sources and links now embedded. (01:12:27) I believe at least Liz said so. (01:12:29) I actually didn't notice it, but there's ads in the AI overview also. (01:12:34) I don't think there's ads in AI mode. (01:12:38) When ads in AI mode, when do you think, I mean, it's, okay, we should say that in the 90s, I remember the animated GIFs, banner GIFs that take you to some shady websites that have nothing to do with anything. (01:12:53) AdSense revolutionized advertisement. (01:12:55) It's one of the greatest inventions in recent history because it allows us for free (01:13:04) to have access to all these kinds of services. (01:13:06) So ads fuel a lot of really powerful services. (01:13:10) And at its best, it's showing you relevant ads, but also very importantly, in a way that's not super annoying, in a classy way. (01:13:19) So when do you think it's possible to add ads into AI mode? (01:13:25) And what does that look like from a classy, non-annoying perspective? (01:13:30) Two things. (01:13:31) Early part of AI mode, (01:13:32) We'll obviously focus more on the organic experience to make sure we are getting it right. (01:13:37) I think the fundamental value of ads are it enables access to deploy the services to billions of people. (01:13:46) The second is ads are, the reason we've always taken ads seriously is we view ads as commercial information, but it's still information. (01:13:54) And so we bring the same quality metrics to it. (01:13:57) I think with AI mode to our earlier conversation about, I think AI itself will help us over time figure out the best way to do it. (01:14:08) I think given we are giving context around everything, I think it'll give us more opportunities to also explain, okay, here's some commercial information. (01:14:17) Like today as a podcaster, you do it at certain spots and you probably figure out what's best in your podcast. (01:14:25) I think so there are aspects of that, but I think the underlying need of people value commercial information, businesses are trying to connect to users, all that doesn't change in an AI moment. (01:14:40) But look, we will rethink it. (01:14:42) You've seen us in YouTube now do a mixture of subscription and ads. (01:14:47) Like obviously, you know, we are now introducing subscription offerings. (01:14:53) across everything. (01:14:54) And so as part of that, we can optimize, the optimization point will end up being a different place as well. (01:15:01) Do you see it trajectory in the possible future where AI mode completely replaces the 10 blue links plus AI overview? (01:15:10) Our current plan is 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. (01:15:22) But as features work, we'll keep migrating it to the main page. (01:15:27) And so you can view it as a continuum. (01:15:29) AI mode will offer you the bleeding edge experience. (01:15:33) But things that work will keep overflowing to AI overviews and the main experience. (01:15:39) And the idea that AI mode will still take you to the web, to the human-created web. (01:15:44) Yes, that's going to be a core design principle for us. (01:15:46) So really, if users decide, right, they drive this. (01:15:49) Yeah. (01:15:51) It's just exciting, a little bit scary that it might change the internet. (01:15:56) Because you, Google has been dominating with a very specific look and idea of what it means to have the internet. (01:16:05) And to, as you move to AI mode, I mean, it's just a different experience. (01:16:12) I think Liz was talking about, I think you've mentioned that you ask more questions (01:16:17) You ask longer questions. (01:16:19) Dramatically different types of questions. (01:16:21) Yeah, like it actually fuels curiosity. (01:16:23) Like I think it's for me, I've been asking just a much larger number of questions of this black box machine, let's say, whatever it is. (01:16:33) And with AI overview, it's interesting because I still value the human. (01:16:40) I still ultimately want to end up on the human created web. (01:16:44) But like you said, the context really helps. (01:16:47) It helps us deliver higher quality referrals, right? (01:16:51) where people are like, they have much higher likelihood of finding what they're looking for. (01:16:56) They're exploring, they're curious, their intent is getting satisfied more. (01:17:00) So that's what all our metrics show. (01:17:03) It makes the humans that create the web nervous. (01:17:05) The journalists are getting nervous. (01:17:07) They've already been nervous. (01:17:08) Like we mentioned, CNN is nervous because of podcasts. (01:17:13) It makes people nervous. (01:17:14) Look, I think news and journalism will play an important role in the future. (01:17:22) We're pretty committed to it, right? (01:17:24) 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. (01:17:34) So it's something I think I definitely value a lot. (01:17:39) And as we are designing, we'll continue prioritizing approaches. (01:17:43) I'm sure for the people who want, they can have a fine-tuned AI model that's clickbait hit pieces that will replace current journalism. (01:17:52) That's a shot of journalism. (01:17:53) Forgive me. (01:17:54) But I find that if you're looking for really strong criticism of things, that Gemini is very good at providing that. (01:18:01) Absolutely. (01:18:02) It's better than anything. (01:18:03) For now, I mean, people are concerned that there would be bias that's introduced that as the AI systems become more and more powerful, there's incentive from sponsors to roll in and try to control the output of the AI models. (01:18:19) But for now, the objective criticism that's provided is way better than journalism. (01:18:23) Of course, the argument is the journalists are still valuable. (01:18:26) But then, I don't know, the crowdsource journalism that we get on the open internet is also very, very powerful. (01:18:33) I feel like they're all super important things. (01:18:36) I think it's good that you get a lot of crowdsourced information coming in. (01:18:42) But I feel like there is real value for high quality journalism, right? (01:18:48) And I think (01:18:50) these are all complementary, I think, like I view this, I find myself constantly seeking out also, like try to find objective reporting on things too. (01:19:01) And sometimes you get more context from the crowdfunded sources you read online, but I think both end up playing a super important role. (01:19:10) So there's, you've spoken a little bit about this, Dennis talked about this, sort of the... (01:19:16) The slice of the web that will increasingly become about providing information for agents. (01:19:21) So we can think about it as like 2 layers of the web. (01:19:25) One is for humans, one is for agents. (01:19:27) Do you see the AI agents? (01:19:30) Do you see the one that's for AI agents growing over time? (01:19:34) Do you see there still being long-term 5, 10 years value for the human created, human created for the purpose of human consumption web? (01:19:44) Or will it all be agents in the end? (01:19:46) In today, not everyone does, but you go to a big retail store, you love walking the aisle, you love shopping, or grocery store, picking out food, et cetera.