Google CEO Sundar Pichai and the Future of AI | The Circuit
We are in the Plex. This is the center of Google. Gosh, it's hard to remember life before Google. It changed literally everything. How we live, how we work, how we communicate, how we get literally anywhere on a Googly colored bike.
Google has been the front door of the internet for over two decades, and now there are so many doors. Google may not be the first place you go for answers anymore, so what are they gonna do about it? Hello. Hi, Emily. So good to meet you again. Good to see you again.
Thanks so much for having us here. Oh, likewise. I'm glad you chose a sunny day to come on campus.
It's beautiful. Sundar Pichai is at a pivotal moment. He's the CEO of Google and its parent company, Alphabet, rising to the top after proving his medal as a talented product leader and peacemaker. He runs a tech giant that functions more like a micro country, overseeing businesses as consequential as YouTube, DeepMind, Cloud, and, of course, Search.
Pichai has been carefully crafting a strategy that infuses AI into every corner of the business. That deliberate planning was met with a surprise from longtime rival Microsoft and OpenAI, whose chatbot seemingly knocked Google off its perch and challenged its cultural relevance. ChatGPT set off a code red at Google, and an industry-wide fever pitch over AI not seen since the dot-com boom. But for Pichai, the frenzy is just part of the long game.
I saw it's your 20th anniversary. That's right. It was last week. It crept upon me.
Are you coffee or tea? I'll go for coffee. How about you? I'm gonna have a green tea. Does it feel like you've been here for 20 years? Not quite.
You know, time flies by. You know, I had my kids right when I started Google too, so the whole thing just flew by. Google is famous for those out-of-the box job interview questions.
Do you remember any of yours? Like, did you have to figure out how many golf balls fit in a school bus or something like that? Thank God. No. You know, but, you know, I remember very, very clearly I interviewed on April 1st, so it was April Fool's Day 2004. There was rumor. I didn't know whether it had actually happened, but Gmail apparently had launched. So all my interviews were about Gmail.
People wanted to know what I thought of the product. And it definitely wasn't a joke? They launched it. They actually launched it, yeah.
But I never allow, no products on April Fool's Day. I think it's too confusing. You just had a blowout quarter. The stock jumped more than it has in a long time.
Did that feel like a little bit of vindication, or, as they say in cricket, was that the bat talking? In many ways, you know, it felt we worked very hard to set up the company for that. You know, in 2016, one of the first things I did as CEO was to say the company should be AI first. To me, we are just getting started in the beginning of what I think will be a extraordinary decade of innovation, and so I'm incredibly excited about it. AI has been around for decades, but it seems like everything is happening everywhere, all at once.
How do you make sense of the frenzy and the scale? All tech cycles are this way, right? You know, the- But this one feels different. It feels bigger. Is it? It is bigger. We still have long ways to go, but we are in the early stages of that. And so, you know, you're gonna feel that excitement, that frenzy, but I think we are prepared for it.
So you kind of roll with the flow. You embrace it. How are you and your kids experimenting with AI at home? We use Google Lens for homework. I don't want to get 'em in trouble, but the class allows you to do that. But sometimes he asks me for help on math. I dunno.
Sometimes I'm lazy and I pretend as if I'm thinking, but I'm also using Google Lens to kind of figure out the answer. You grew up in Chennai in India. What was it like at the kitchen table? Like, what got you here? My parents always emphasized, you know, learning and knowledge, and it's worked in some ways, Google with its mission, it always resonated very deeply with me.
You know, I felt this quest for learning knowledge, et cetera. It's what the company is about, too. I grew in a middle-class family. I perceived our life through the arrival of gadgets.
We waited five years for a telephone. It was a rotary phone, but when it came to our home, you know, it changed everyone's lives. I remember getting our first television and suddenly being able to watch sports. You know, I used to bike a long way to school, but, you know, there was no gear in the bike, and I had to go uphill. But then, after many years, I got a bike with gears, and, like, I'm like, "Wow, this thing make, you know, what a dramatic difference."
I never took technology for granted. And so that's what, you know, always been optimistic about how technology can make a difference. You've been CEO now for a decade. How have you changed in that time? I think the main thing is when you've done it for a while, you know, you get to understand patterns, so you're running into something, but there's a sense of deja vu, you know, you've seen it before. And so I think that helps you kind of pattern match and deal with it more effectively. But a lot of this is, you know, this scale, you have so much coming at you, there's a lot of noise, and most of it doesn't matter.
So the ability to separate the signals from the noise, pay attention to the few things that you need to pay attention to. And, you know, I think I've gotten better at that over the years. I heard Sergey is back and working a bit on AI.
What is the involvement of Larry and Sergey these days, and what advice are they giving you? I talk to them regularly. Sergey is actually spending more time in the office. He's literally coding.
And, you know, some of my fondest memories over the last year is sitting with Sergey on large screen looking at lost curves as we train these models. I think one of the advantages they have is they're not caught up in the day-to-day. And so sometimes, when I have conversations with them, it allows all of us to step back and look at the bigger picture, which I think is incredibly important when you run something at this scale. The decisions Pichai and Google make influence how billions of people get information. And the nature of how users do that is starting to change in the age of AI.
This seems like a threat to Google, except the core technology being used by Google's competitors was invented by Google itself. Google researchers invented the transformer, literally the T in GPT. Do you wish you capitalized on that louder and sooner? We use transformers in Search. That's what led to large gaps in Search quality compared to other products. So we have infused transformers across our products.
We have a chance to do that better with generative AI and with the Gemini series of models. And there's gonna be more breakthroughs in this field. But what is more important is we are driving that progress. And if the new Google is only gonna be more and more AI, you know, AI is super helpful sometimes, but sometimes it's still deeply wrong. Where do you draw the line? I think part of what makes Google Search differentiator is while there are times we give answers, it'll always link to a wide variety of sources. Now we've had answers in Search now for many, many years.
We are just now using generative AI to do that. So the links will live on. Yes, and, you know, I think it'll always be an important part of Search. There will be times when they want quick answers. My son is celiac, so we did a quick question to see whether something is gluten-free. We just want to know.
But often it leads to more things, and, you know, then you want to explore more. I think understanding that, meeting all that needs is part of what makes Search unique. Some leading computer scientists have said Search is getting worse, more SEO spam, et cetera. Do you see their point? Anytime there's a transition, you get an explosion of new content, and AI is going to do that. So for us, we view this as the challenge, and I actually think, you know, there'll be people who will struggle to do that, right? So doing that well is what will define a high-quality product, and I think it's gonna be the heart of what makes Search successful.
You make a ton of money on ads next to the links generated by searches. If a chatbot is giving you answers and not links, and maybe more answers than links sometimes, are we in the midst of an assault on Google's business model? So we've always found people want choices, including in commercial areas, and that's a fundamental need. And I think we've always been able to balance it.
As we are rolling out AI or views in Search, we've been experimenting with ads, and the data we see shows that those fundamental principles will hold true during this phase as well. The images that Gemini initially generated of Asian Nazis and Black founding fathers, you've said that was unacceptable. If you look at any pictures of the founding fathers, you're seeing old white men. People are calling this woke AI, and it's not just happening here, it's happening across the industry. How did the model generate something that it never saw? We are a company which serves products to users around the world, and there are generic questions.
For example, people come and say, "Show me images of school teachers, or doctors, or nurses." We have people asking this query from Indonesia or the US, right? How do you get it right for our global user base? Obviously, the mistake was that we overapplied, including cases where it should have never applied. So that was the bug, and, you know, so we got it wrong. Would you say it's like good intentions gone awry? In this particular case, yes, right? I think, you know, we are rightfully held to a high bar, and I think we clearly take responsibility for it, and we are gonna get it right. How concerned are you about AI-generated content ruining Search? For example, the AI-generated selfie of the tank man in Tiananmen Square.
It shows up in Google search results, but it never happened. The challenge for everyone and the opportunity is, how do you have a notion of what's objective and real in a world where there's gonna be a lot of synthetic content? I think it's part of what will define Search in the next decade ahead, right? People often come to Google right away to see whether something they saw somewhere else actually happen. It's a common pattern we see.
We are making progress, but it's gonna be an ongoing journey, right? You've got AI systems that are running out of training data. You've got companies turning to AI-generated data to train their models. Aren't there risks to that? Yes. I think that, you know, through it all, are you creating new knowledge? Are these models developing reasoning capabilities, right? Are you making progress in the intelligence of these models? I think those are the frontiers we need to prove that, you know, you can do that by using these techniques. Is LLM technology nearing a plateau? I would be surprised if LLMs are the only thing we would need to make progress. We are investing a lot of computing and resources.
Our AI research is talent in driving the next generation set of breakthroughs. It seems, when you look at the big picture, like Google missed the big moment, and ChatGPT took it. A new artificial intelligence program called ChatGPT made its debut online. It's as revolutionary as the internet. The topic of today's big take is also Microsoft, and it's big bet on AI and how it just kind of leapfrogged over Google and no one kind of knew about it until now. If you could go back, what would you do differently? To be clear, I take a long-term perspective and say, when the internet just first came about, Google didn't even exist then, right? So we were in the first company to do search, we were in the first company to do email, we were in the first company to build a browser.
So I view this AI as, you know, we are in the earliest possible stages. Your leadership style has been described as slow, and steady, and cautious, sometimes maybe too cautious, and you're often compared to these other tech leaders who are moving fast and breaking things. How would you describe yourself? Look, the reality, I think, is quite different.
One of the first things I did when I became a CEO was to pivot the company sharply to focus on AI as well as really invest more in YouTube and Cloud to build them into big businesses. I think the larger the company is, you are making fewer consequential decisions, but they need to be clear, and you have to point the whole company to that. Part of that, at times, involves bringing the company along. You build consensus because that's what allows you to have maximum impact behind those decisions. I mean, any leader in a position like yours has to be willing to hear the criticism. And I'm not gonna make you read the mean tweets like they do on "Late Night," but I do have a few.
"Where is Google?" "Running things through legal." "Google doesn't have one single visionary leader, not a one." Do you think you're the right person to lead Google? Look, it's a privilege to lead the company.
You know, people tend to focus in this micro moment, but it is so small in the context of what's ahead, and, you know, when I look at the opportunities ahead across everything we do, and for the first time, all of that has a calm and leveraged technology with AI. You know, I put a lot of chips, at least from my perspective on Google. So can you walk through campus without getting stopped? It's definitely been nice to walk and see people, you know, so I enjoy it a lot. I see a dinosaur statue in the distance, which I think is a good reminder. Like, how much do you worry about becoming a dinosaur in a world where technology is moving so quickly? I mean, in technology, I think, if you don't innovate to stay ahead, I think that's the inevitable fate of any company. No dinosaurs? Not yet.
Well, they were great, but you know, you don't want, you don't want to have the same fate. How much has this AI moment forced you to move and think differently? Because it does seem like you're playing defense sometimes. We have been preparing for this for a while. You know, a lot of the foundation of breakthroughs in the field came from Google. So to me, this moment has been over the past year, really channeling the company to meet the moment.
There they are. The I/O tents. I/O tent. For one time a year, we get to pretend as if we are on a concert stage, except we are speaking about technology.
So Google I/O is sort of a state of the Google Union. What is the key message this year? It's less about, you know, particular product, more about the journey we are on, the vision we have for how AI can transform our products, and how we can bring it. Alphabet used to be thought of as this collection of moonshots, but a lot of those projects have been spun off, or folded back into Google, or shut down. How should we think about Alphabet today? What is the new Google? When we think about both across Google and Alphabet, the underlying bet is the same. You're going to invest in deep technology and computer science and apply it to solve problems for people. So I think that part doesn't change.
When you take big, large-scale bets, by definition, you're not aiming big enough if you don't have a few failures. When you go back to the S-1, Google said, "We're not a conventional company. You can spend 20% of your time on personal projects." Google created this bottoms up culture where everyone has a voice, and it's super transparent, and the perks are great too.
But did it go too far? Like, did it become an entitlement culture? Part of what makes Google unique is, I think, there is a culture of vibrant, open debate. And so I think it's super important to preserve that. I think we are also a company where employees have many ways to speak up, and I think that's made the company better.
Google, Google, you can't hide. Google, Google, you can't hide. You recently fired Google employees who were protesting this contract with the Israeli government for cloud services. It seemed like a distinct change in tone for a company that's historically welcomed all kinds of views. Why did you take this stand? I think when we have cases, including in this case where a few employees, you know, cross beyond what's in the code of contact and disrupt the productivity of the workplace, or do so in a way that it makes other people feel uncomfortable, I think we have to take action, and that's what we are doing. It has nothing to do with the topic they are discussing.
It's about the conduct of how they went about it. Over the past years, through the pandemic, the company has grown a lot. There are a lot of new people. I view, particularly in this moment with AI, the opportunity we have ahead of us is immense, but it needs a real focus on our mission. There have been multiple rounds of layoffs. Why take this approach? Why not cut once and cut deep? We are reallocating people to our highest priorities.
There are cases where you're simplifying teams, you're moving people to focus on newer areas, removing layers so that you can improve velocity. So that's why we have taken the time to do it correctly and well. Microsoft is obviously making huge investments in AI as well. OpenAI, Inflection, Mistral. We've reported that their OpenAI investment was actually in part because they were worried about Google and wanted to catch up.
How do you feel about the competition? I've always held a view if you're working in the technology space, there is a lot of competition. We see it all the time. The way you stay ahead is by innovating relentlessly.
It has to be true all the time. It's happening at a faster pace. But, you know, technology, changes tend to get faster over time, so it's not surprising to me at all. Microsoft CEO Satya Nadella has had some fighting words. I hope that with our innovation, they will definitely want to come out and show that they can dance.
And I want people to know that we made them dance, and I think that'll be a great day. Who's really choosing the dance music? I think one of the ways you can do the wrong thing is by listening to noise out there and playing to someone else's dance music. I've always been very clear. I think we have a clear sense of what we need to do. So you're listening to your own music.
That's exactly right. Google is facing a ton of regulatory pressure in the US, abroad, over your dominance in Search, video, ads, the app store. Some other big companies have split themselves up to focus on their core. Has Google thought about that? A lot of our products integrate in a way that provides value for our users. So I think that is important.
Part of what allows us to compete in the Google Cloud market is our, the investment in AI we undertook because of Search is what allows us to take that and compete hard against other larger companies, like Amazon and Microsoft, in Cloud. So I would argue that the way we are approaching it drives innovation and adds choice in the market. What do you think is the future or potential of AI-powered hardware, and what will Google's role in it be? I think with AI, you get a chance to rethink that experience over the next few years. I still see the center of where the AI innovation will happen in smartphones, followed by glasses, right? That's how I see it.
Last time we talked, you told me China will be at the forefront of AI. How should policymakers factor that into their decisions? I think over time we need to develop frameworks by which we achieve global cooperation to achieve AI safety. I know it sounds farfetched now, but we've done it in other areas, like nuclear technology and so on, to some extent.
I think we are gonna need framework like that. And so I would expect, over time, there needs to be engagement with China on important issues like AI safety. Artificial general intelligence.
What does it mean to you? Do we get there and when? It's not a well-defined phrase, it means different things to different people. But I think if you define AGI as AI becoming capable across a wide variety of economic activity and being able to do it well, I think that's one way to look at it. So Google's gonna get us to AGI? You know, we are committed to making foundational progress towards AGI in a bold and responsible way. And so, you know, I'll focus on the effort to do that and do that well. You've said there's even some things about AI that you don't understand.
Will AI always be somewhat in a black box? I have a little bit of a counterintuitive view there. I think, you know, humans are very mysterious, too. Often, when people explain why they did things, you know, they're telling something, it's not entirely clear. Today, we can't make sense of many complex systems. AI will give us more insights and more visibility into many complex things. When I asked OpenAI CEO Sam Altman why we should trust him, he said.
You shouldn't. Why should we trust Google? Well, I share the notion that no one, you shouldn't blind lead, you know? That's why it's important to have systems in place. Regulation has a part to play, you know, test balance innovation.
But as these AI systems get more capable, it shouldn't just be based on a system of trust people or trust companies. We've talked a lot about the opportunities. What is the biggest threat to Google's future? I view for all companies, particularly, you know, at scale, you know, the biggest threat is not executing well.
Is there a healthy dose of paranoia? Like not becoming Stan the T-Rex out there? I think that, you know, there's a part of me which has always internalized, I think, the old Andy Grove phase, "Only the paranoid survive." That is important. And I think this moment is no different. Are we gonna look back on this LLM era and laugh? Like, is this gonna all look so basic and rudimentary? I hope we do because that shows, you know, you know, my kids aren't impressed by touchscreens or the fact that they have this extraordinary amount of computing in their hands.
So similarly, you know, there's no reason we wouldn't scale up our computing a hundred thousand times in a few years. So, yes, you know, I hope some of this looks like a toy in the future. I hope it is that way, otherwise, we didn't do our job well. Thank you so much. That was so awesome. She's more difficult than any AI to deal with, right? Oh, man.
2024-05-10 20:18
