Google CEO's In-Depth Interview: The Future of Search, AI Transformation, and the Backstory of Merging DeepMind

[baoshi.rao]

On the day after the Google I/O developer conference, Nilay Patel, editor-in-chief of tech media The Verge, had an in-depth conversation with Sundar Pichai, CEO of Google and Alphabet, which was packed with insights.

Sundar discussed Google's new Search Generative Experience, the dual nature of generative AI, why Google wasn't the initiator of the latest AI platform revolution, and how Google views the impact of generative AI on job losses and fake songs. Recently, Google announced the integration of DeepMind, the UK-based AI company under Alphabet, into Google, merging it with Google Brain to form Google DeepMind. Sundar also explained the reasoning behind this decision and how it was executed.

Google has invented many of the core technologies behind the current AI boom. The "T" in ChatGPT, the popular AI chatbot, stands for Transformer, a large language model technology invented by Google. However, in the generative AI wave, Google has lagged behind. OpenAI and many other companies have already launched generative AI products. Microsoft's New Bing is seen as the first real competitor to Google Search in a long time.

Therefore, Sundar's views on the future of search and his plans to reshape search with AI are not only crucial for Google's development but also highly relevant to the search industry. During the conversation, Sundar responded to Microsoft CEO Satya Nadella's comment that "Microsoft made Google dance," discussing his vision for Google and the ambition driving the company forward.

Below is the full transcript of the interview, lightly edited and supplemented for clarity:

Nilay: Sundar Pichai, you are the CEO of Alphabet and Google. Welcome to Decoder.

Sundar: Niraj, it's great to be here.

Niraj: I want to start with the news. At yesterday's Google I/O developer conference, you delivered the keynote. You announced generative AI features in almost every Google product I can think of. What's your favorite?

Sundar: It has to be the new search generative experience we're bringing to Labs. This is our most widely used and most important product. The opportunity to improve the product through such evolution is one of the most exciting product challenges. I think the team has risen to the challenge, so I'm very excited about it.

Niraj: I think it's incredibly exciting. I'd love to talk more about search. Two demos caught my eye. First, you used the compose feature in Gmail to request a refund from an airline, and Gmail wrote the email for you. Later, Dave Burke wrote an email to Rick Osterloh saying Rick did a great job. How would you feel if your employees used generative AI to write you a flattering email?

Sundar: This is something I've been thinking about, especially in a personal context. You'll eventually... I think social norms will develop over time. People will decide what's appropriate and what's not. The last thing you'd want is an AI-generated email being replied to by AI. I think it's fine for the airline voucher case, but definitely not in personal situations. That said, some friends have told me they're not the best at writing such emails and might need some help. But over time, as a society, we'll figure out where the right norms lie.

Niraj: Do you think, in the airline case, there's a procedural element where if you say the right words to customer service, you might get a refund? The AI might know these right words, and the airline might say, "Look, we'll let the AI scan emails for these right words and issue refunds." Could this become a loop?

Sundar: I'm concerned—perhaps airlines are already using AI to review your emails, which might give people a pass. Sometimes it can be seen as having effective proxies in these situations, allowing two people to efficiently complete transactions. I think that's acceptable. But it depends on the use case. You're right, I believe there will be situations where people find better, more efficient ways to handle this back-and-forth communication, and maybe that's fine.

Nira: You're at the forefront of this field, so I'm curious—when you see these norms changing, does it affect your work at Google? Or do you say, "Let's have many people try it and see what happens?"

Sundar: I've encountered these issues with Smart Reply and Smart Compose. At first, I found it strange to use. Later, they incorporated emotions into these suggestions. So you'd see emotions, and I was like... But over time, I think I've become better at using emotions I genuinely feel. People adapt to these things faster than we imagine. They quickly learn how to use these technologies, making it distinctive. But I think we'll go through a similar journey.

Nira: I want to broaden the perspective. Something you said in your keynote caught my attention—that AI is a platform shift, which I agree with. But what struck me is how crucial it is to accurately understand what you mean by platform shift. Why do you consider AI a platform shift, and what does that mean to you?

Sundar: Absolutely, I firmly believe this is an extraordinary platform shift. It will touch everything: every sector, every industry, every aspect of life. One way to think about it is no different than how we view the PC revolution, the internet revolution, or the mobile revolution. From that perspective, I see this as a monumental shift.

But I think it goes deeper. In some ways, I consider it the most profound technology humanity is working on. It touches the essence—even your initial question reflects this. It shows the nature of AI. Therefore, I believe it will impact everything we do.

So I see it as one of those deeper shifts. It's hard to find the right words, but I think it's... Even in a very traditional industry, you wouldn't say the internet has had a major impact on healthcare. Has it affected healthcare? I'm not entirely sure. But with AI, I believe it will have a significant impact on healthcare over time. From this perspective, I think the term 'platform shift' here carries a deeper meaning.

Niraj: On a narrower definition, there was no PC before, and then there was; there was no internet before, and then there was.

Sundar: Exactly.

Niraj: There were no mobile phones, no cloud before. Now, especially mobile and cloud, they've changed the way we behave in every aspect you can think of. I think that's what a platform shift means. Right? It's very narrow... much narrower than your view.

Sundar: Still big, still big.

Niraj: We can talk philosophy.

Sundar: Sure.

Niraj: The philosophical questions we can explore include: Can computers communicate with us? We can. But at this level, it's a platform shift. Many people will change their behavior. This is usually when new companies emerge and institutions tend to decline. Google is an emerging company, and I think, with the development of the internet and the shift to mobile, Google became a dominant player. Do you see risks for Google in this platform shift?

Sundar: I think the risks were greater in mobile. The reason is that we developed Android, and I believe we had to adapt to mobile as a company. We were built on the internet. At that time, we couldn’t be called a mobile-native company by any means. So mobile was something that emerged, and we had to work hard to adapt it into our products. It was a disruptive moment. People now use apps directly. You can install apps on your phone, and so on. So there were many challenges. In AI, I feel this is our seventh year as an AI-first company. I think we are AI-native. Part of the reason is that almost all of Google’s teams intuitively understand the AI used in our products.

I remember every year... and we’ve pushed the boundaries of technology. In a way, we’re helping drive the platform shift. So I feel we are AI-native. We deeply understand what it means to advance technology and integrate it into our products... All these shifts are disruptive, but what I see is the scale and size of the opportunities AI brings, and I feel we’ve been working on it. We’ve invested so deeply in AI, and we have clear goals—not just to build AI into our products but also to make it available to the rest of the world. We planned for this from the beginning. So this makes me excited about this moment.

Niraj: You’ve said it’s been seven years as an AI-first company. I’ve seen you showcase large language models. I’ve also seen other generative AI technologies at past I/Os. You’ve been talking about this for a long time.

Sundar: In 2015, one of the biggest debates at Google I/O was that I wanted to showcase... we launched Photos. I wanted to show that these were powered by deep neural networks. There was intense debate during the keynote because we used a frog to demonstrate how the network identified it as a frog, and people were horrified. They said, 'Why are you showing the frog’s legs?' The first thing to understand was that you were breaking the frog down into its components. But I felt it was necessary to explain to the world that there’s something called deep neural networks, and this transformation is important—it will change everything. Anyway, this made me think about it. We’ve been discussing this for a long time.

Nile: I remember very clearly that you once had a conversation with Pluto.

Sundar: Yes.

Nile: No one could fully understand why you were talking to Pluto. Then you jump to now, like, "Oh, that was technology, that was a demo."

Sundar: We built LaMDA because we were developing AI-based conversations—it wasn’t accidental. Because we built Google Assistant, we realized the limitations of our approach. Ultimately, we had a vision for its direction, but it was an artificial system. So we knew we needed a deeper AI approach. So yes, by talking to Pluto, we effectively had a conversation inside LaMDA, but for safety reasons, we limited it to just Pluto.

Nile: Everyone can see that the platform shift happening wasn’t initiated by Google. To some extent, it was started by OpenAI and ChatGPT, and by Microsoft. This is because you were responsible, you were cautious. I think the launch of this platform shift might have been accidental. I don’t think OpenAI was working toward such a moment. What made Google react at this moment instead of proactively initiating the platform shift?

Sundar: I’d argue that part of what drove the platform shift was our work on Transformers. Much of the underlying technology was too. I think what changed the inflection point was user readiness. It’s like the moment you realize… because these technologies have flaws, they have gaps, but you realize you’re at a moment when people are ready to use them. They understand it and are adapting to it. At that moment, we realized it, and we started working on it.

I just think it took us some time to get it right. For us, it's important. I believe that, given how many people use our products and at critical moments, it's important to get this right. So, for me, that's what it is. If you go back to the origins of the internet. When the internet transformed, Google didn't even exist. So I think there's this notion that one of the most profound platform shifts happens on day one. I just don't agree with that.

Nira: As the head of all Google products, do you find the level of hallucinations or errors you see in ChatGPT-like products unacceptable?

Sundar: We have to figure out how to use it in the right context, right? For example, if you search for the dosage of Tylenol for a three-year-old, hallucinations in that scenario are unacceptable. But if you're just asking me to write a poem about a topic, getting it wrong is fine. What I mean by getting it right is handling these nuances correctly. In the context of search, we've made progress on hallucinations, validating what we've done in ranking work. It just takes time. That's what I'm saying. It's a research problem. We'll all make progress on hallucinations. I'm not saying it's unusable, just that we have to take the time to get it right.

Nira: But what I'm saying is that OpenAI is truly a disruptor here. Their product may not be as reliable as Google Search in answering questions, but for certain queries, it does better, is more fun to use, and represents a different paradigm. Users are ready, but then it makes mistakes, like left and right. One of my favorite examples is: people walk into libraries asking for books that don't exist because they got a list of books. I think that would be unacceptable in Google Search results.

Sundar: I was looking for some products in Bard, and it provided a URL to purchase them, but the URL didn’t exist. Right? So all these models have the same fundamental issues, but there are also many use cases we're excited about. I think both can be true.

Niraj: But do you see that classic disruption curve? Apologies for using this imperfect analogy—Google Search is the mainframe, AI is the PC. It's a classic disruption example. It can't do everything a mainframe can, but it's cheaper, more accessible, and perhaps more useful in some cases, though worse in many other aspects.

Sundar: No, I don’t see it that way because Google Search is evolving with what you see. Google Search won’t remain static. For many years, we didn’t go beyond the 10 blue links.

People would ask us, 'Why are you doing this?' We’d always say, 'This is what users want.' The debate sometimes centers on what answers users truly desire. So we always strive to serve users correctly. This is a moment when user expectations are shifting. We must adapt. We’re also working on Bard, now rolling it out widely, which gives us a sandbox to push the boundaries of possibilities in an unconstrained way.

Between Search, the new Search Generative Experience, and Bard—to me, it’s far from a zero-sum game. That’s how we see it today. People are using Search, experimenting with new things, and that’s why I’m excited to launch this new experience—I believe people will respond to it.

Niraj: A few months ago, I attended the launch of ChatGPT-powered Bing. There, I met Satya Nadella. I’m sure you know this—he said, 'I have immense respect for Sundar and his team, but I want Google to dance.' Then he added, 'I want people to know that Microsoft made them dance.' First, I’d like to know how you felt when you heard him say that. Second, do you think you’ve danced? Are you dancing now?

Sundar: Look, I’ve said I have great respect for Satya and his team. I think part of why he said that was to get me to answer this question.

Perhaps I would say this. We've been working on this new Search Generative Experience since last year. For me, it's important to separate the signal from the noise in these moments. The signal here is that there's a new way to make search better, a way to improve our user experience, but we have to get it right. For me, that's the North Star—the signal—and everything else is noise. So what's important to me is to execute it well, and that's what we've been focused on.

Niree: So let's talk about search today and where you want to take it: the Search Generative Experience. The search business is lucrative. The EU has spent 20 years trying to compete in search. There are browser and search votes on Android. Google still dominates, but it has declined over time. Have you ever done a search and ended up with some terrible SEO content? Has that happened to you?

Sundar: Yes, but it's been happening to me for over 20 years. So, to me, it's like Gmail and spam. Search has always been about finding high-quality content from others. Sometimes we feel like, okay, there's a direction where we're not doing well or falling behind, but then we work to fix it. That's how search has always been.

Niree: Do you have a dedicated team for this?

Sundar: We measure these things quantitatively, right? Our work on search quality is about... internally, we work hard to quantitatively measure user satisfaction with search. How do users perceive search? Over time, we've seen this. So in a way, when you complete work like BERT and MUM, all of this leads to the biggest improvements in search quality we've seen in a long time. So you're right. I'm not saying... I've encountered content forms where sometimes users say, "Look, I want more unique voices and perspectives." We've been working hard to achieve that. This is also part of how we'll improve search for specific use cases. But I think today in search, you have a good example. I mean, you redesigned The Verge website. I think it's been about a year.

Nire: Almost there. Yes.

Sundar: You didn't design it with what Google Search expects in mind.

Nire: We did. Our designers care about SEO.

Sundar: But that's the good part. I think I see StoryStream and the most popular feeds—I use it to see what's important. The Verge does this well, and I believe great work is still possible. The information ecosystem is vast, often underestimated. When I look at Facebook, Instagram, YouTube, TikTok, or news from The New York Times, The Wall Street Journal, and The Washington Post, even sports sections, it's richer than people realize. But effort is still needed to get things right.

Nire: But you can only be as good as the web.

Sundar: Always.

Nire: Ultimately, Google Search can only show you what's on the web.

Sundar: The richness of the internet is great. Yes.

Niraj: Yes. But if you're a new creator just trying to connect with an audience, you might end up on TikTok, Substack, or Instagram, maybe YouTube. However, these platforms aren't as visible to the average Google search user. So new, high-quality, and more interesting content might end up on platforms that Google search can't see. The web is being pushed toward Google's incentives. When was the last time you applied for a new credit card?

Sundar: It's been a while.

Niraj: It's a completely optimized experience. To some extent, it's almost no longer human-readable. I wonder, do you see search-generated experiences as an opportunity to change these incentives? Could this better motivate people to create for the web again?

Sundar: It's a fact of life. I think mobile is here to stay, video is here to stay, and as a result, there will be many different types of content. The web is no longer the center of everything as it once was. I think this has been the case for a while. That said, ironically, all these recently released products like Bard and ChatGPT are web-based. That's just how it is.

Niraj: I could spend 30 minutes talking about mobile app stores and why innovation happens on the web. But I feel like I don't have enough time.

Sundar: I work on Chrome. I've been following the web for a long time, but I believe the web doesn't belong to anyone. There's intrinsic value in that. Some aspects of the web are more powerful than most people realize. But I wouldn't underestimate AI. As AI becomes more multimodal, the distinctions between text, images, and video will blur over time. Right now, we feel these barriers. At Google, we've been working to address these issues. We've developed universal search, attempting to integrate all these formats. With AI, what happens? Maybe a young content creator produces something in video format, but ultimately, you might be able to use it in Google's context through other means. Obviously, all the details, including the business models, need to be figured out.

Niraj: But these platforms have to let you in, right? You can't search Instagram.

Sundar: But we can compete with YouTube.

Niraj: Yes.

Sundar: That's true. There may be other platforms too. We need to incentivize them to make content available, and that's what we're working on. I see it this way: Is user demand for information increasing or decreasing?

There are more sources of information than ever before. Somehow, over time, I've grown more optimistic because these issues were equally profound years ago. I remember people asking me about it. But looking at where we are today, I think, if anything, I'm using the web even more. I frequently visit The Verge website.

Neri: We are believers in the web. We are the last ones.

Sundar: Yes, I understand. But my usage still tells me... I visit websites directly every day, and every day they try to get me to download mobile apps. They ask me to agree to cookies and all that, but the web already does all this. I hope it will get better, but I'm optimistic.

Neri: As you answer more questions in the search-generated experience, you gave an example of what is essentially an automated buying guide, right? Like buying a bicycle. Then you asked a follow-up question: 'I want a red one.' It would give you the correct color. Do you think search engines will drive as much traffic as before, or will it decrease?

Sundar: This is an important part of our design goal. I think people come to Google with different purposes. Sometimes you just want an answer, like 'I'm going to New York tomorrow, is it raining?' You want the answer. But often, especially with Google, people come to explore and discover. I think that's true. People want to read reviews.

So, in our search-generated experience, we really don’t want to just be a place where you come to talk to a large language model. That’s why we initially separated the chatbot. In the search-generated experience, you’ll see many links. Click 'Expand.' We review and provide the sources supporting every response generated by the large language model.

So one of our design goals is to ensure people experience the richness of the web because I think creating a win-win structure is important for us. This is something we’ve put a lot of thought into, so I’m optimistic we’ll achieve it.

Neri: You're the CEO of Alphabet and Google. You made a major organizational decision. You have a company called DeepMind, which is part of Alphabet. You merged it into Google, combining it with Google Brain, Google's AI division. You selected a new leadership team. Please tell me about that decision in the context of "I want to build these products, and I need to change my organizational structure to achieve that goal."

Sundar: Clearly, for some time, I've felt fortunate that we have what could arguably be two of the top three research teams in the world. I mentioned this at I/O. I didn’t even list all the things they’ve accomplished. If you look at the 10 to 20 groundbreaking breakthroughs that have shaped where we are today, these two teams have contributed a significant portion.

But it was clear to us that as we started building more powerful models, one of the constraints was the computational resources we needed. So, we needed to bring them together. In some ways, I think the good thing is that the teams themselves recognized this. The two teams had already begun collaborating on the Gemini project (Google’s next-generation multimodal large model) before the merger. It was a great experience because it was almost like bringing two great teams together and seeing how it worked. I think my conversations with Demis (Hassabis) and Jeff (Dean) naturally led to that moment.

So, I believe this was the right time to do it because we’re also shifting more from research to large-scale, production-ready models, while ensuring it’s done safely and responsibly—which means investing heavily in testing and safety. I think the combination of all these factors made it the right moment. That’s what led to this decision.

Neri: This is the strategic aspect, right?

Sundar: Yes.

Nile: You have two teams. There is redundancy in their resource and infrastructure needs. But you actually decide: 'Alright, I have two leaders for two teams. I'll choose one. The two teams have different cultures. I want this culture. I want more output, less redundancy.' How do you make these decisions?

Sundar: You're right, it's always about... I think the most important thing is to be clear about what you want to accomplish. Once you do that, everything else follows. In this case, Jeff has obviously expressed for some time his desire to become more of a chief scientist. Jeff has indeed built some of the most important systems we use at Google today. At least for me, there's no doubt that he's the best engineer Google has ever had... He wants to spend more time on that.

Demis is an extraordinary team leader. From the first day I met him, he has been committed to building powerful AI systems. This is what he has always wanted to do. He's super excited. So, understanding the people you have, what makes sense, all of this comes from the first principles of what you want to accomplish, which leads to other decisions. So in a way, it's a set of clearer next steps.

Nile: Did you make a phone call? Did you have meetings? Did you have Gmail write you a message?

Sundar: No, no. We had many good meetings. James Manyika played an important role because, in the context of Gemini, we were going to bring these teams together anyway. Jeff naturally spends a lot of time on engineering work. So, it all made sense. Just a few conversations led to the right outcome.

Nile: I'd say merging these two teams signals a bigger change for Google and the entire tech industry—becoming smaller, more efficient, and less redundant. You and I have joked a lot in the past about Google's six messaging apps. Are you focusing on tightening up, being more execution-focused?

Sundar: Yes. I would say, I do think this is one of our strengths. We have 15 products that have reached this scale, with 6 products each having over 2 billion users—this is no accident. These are products we've been committed to for a long time. But obviously, I think we're now making more focused efforts in areas where we can be more flexible. In areas where you can be more agile, we've been very intentional about it. We've been doing things like merging the YouTube Music and Google Play Music teams, saying, 'No, you'll become one music team.' There are always moments like this...

Niraj: But is this Google's default approach—having multiple initiatives and then consolidating them later?

Sundar: Well, in some cases, think about Search, Maps, Photos, Gmail, Workspace, our focus on cloud computing, and the fact that we acquired YouTube in 2006 and how we've executed to make it what it is today. I think there are some high-profile areas, like messaging, but even there, if you look at our focus on Google Meet, Chat, and the RCS platform over the past few years, the truth is we've been building from the ground up. I'm confident... I mean, we announced that RCS now has over 800 million users.

Niraj: The audience erupted in applause yesterday.

Sundar: Yes.

Niraj: I think Dieter (Bohn) was the loudest cheerer in the room.

Sundar: I heard that. I really think I can distinguish Dieter's applause from others. I believe we are committed to deep focus. I mean, even AI is an example. We've been focused on AI for over a decade. In terms of AI, it was a deliberate decision because it's so important. We're pleased with the exploration from two different teams, as they have different strengths. DeepMind was an early believer in reinforcement learning, while Google was not. So, to me, this diversity is also important. But sometimes you say, 'It's time to change direction.' But I think these are the decisions you need to make.

Nirel: I have a few more questions, and then I have a concluding question before we continue for another hour. [Laughs] Google faces another challenge, right? If you consider this a platform shift, it might be the first one regulators truly understand, because it's clear what kind of workforce will be displaced. From what I've gathered, it's mainly lawyers. They can see that entire floors of white-collar workers—like those drafting C-plus emails about deals—could be reduced. They seem very focused on this risk. Then there's the broader AI risk we all talk about.

When Google first launched search, it was the underdog, right? It won many court cases while building the internet: the Google Books case, the Perfect 10 image search case, the Viacom case with YouTube. It was the underdog, but it clearly delivered immense value. Now, you're holding AI summits at the White House. I imagine you'll eventually be discussing AI in government capitals worldwide. Do you feel your position is different now compared to that scrappy underdog inventing the internet? You're the incumbent. Are you playing a different role?

Sundar: This question has two parts. First, simply put, in 20 years of technological automation, people have predicted the disappearance of all kinds of jobs. Cinemas were supposed to end, and—

Nirel: They were.

Sundar: Hmm. But movies are more thriving than ever.

Neri: There's a writers' strike, right? I mean, the labor costs for writers have dropped sharply, and they're now on strike.

Sundar: No, but there have been writers' strikes before, and these things will continue, right?

Neri: Yes.

Sundar: There will always be... Over the past 20 years of technological automation, job losses haven't fully... 20 years ago, when people accurately predicted what technological automation would bring, they very specifically declared that entire job categories would disappear. That hasn't fully happened. So I think AI might... because I believe the legal profession is much more than... You have the opportunity to learn more about being a lawyer. That's why I can't comment, because I don't know much about it. But something tells me that more people might become lawyers because the fundamental reasons for the existence of law and the legal system won't disappear, as these are human issues. So AI will, to some extent, make the industry better, possibly with some unintended consequences, but I'm almost willing to bet that 10 years from now, there might be more lawyers—I don't know.

So I'm not entirely sure how all this will unfold. I think we often assume that new professions are constantly being created. I don't mean to be too flippant... I do believe there will be significant upheaval in the societal labor market. Governments need to get involved. We need to adapt. Skills will be important. But I also think we shouldn't underestimate the benefits of some of these changes. It's complicated, I might say.

Regarding your second question, I think governments and legal systems will always face the same issues. There's a new technology that offers unprecedented benefits but also drawbacks. I think you're right. With AI, people are more willing than ever to think ahead, which comforts me because there are potential downsides to this technology. I feel we need to consider them. We need to anticipate them early.

But I do think the answers to these questions aren't always obvious to me. I believe that simply blocking AI in a straightforward manner isn't the right solution. It carries geopolitical implications. Over time, this becomes an increasingly complex issue. From our perspective, as a larger company, I do believe we have a responsibility to address this matter in a more accountable way. In certain areas, we will engage and attempt to find the right answers. As our research progresses, our approach will undoubtedly evolve.

Niraj: One thing I often think about is the series of cases I've mentioned: Google Books or Viacom on YouTube... You're disseminating more information than ever before. A group of media companies says, 'No, that's ours. You can't have it.' You have to fight hard to obtain it. The value to people was so great that Google could prevail. This is a different twist, right? Publishers, media professionals, Hollywood artists, and Drake worldwide are saying, 'Hey, that's mine. You took it and trained an AI.' Now, there are fake Drake songs on YouTube, and they'll try to stop you, right? Copyright lawsuits are already underway. Do you think, as an incumbent, you bear greater responsibility in this conversation than some startups that might be following Google's original playbook, saying, 'We'll ask for forgiveness, not permission'?

Sundar: I do believe we have greater responsibility. So, I think YouTube has done an excellent job with Content ID, providing a robust framework for content rights holders... I believe our responsibility is to ensure this new wave continues to benefit artists and the music industry. This is something we need to think deeply about as we navigate this issue.

Niraj: Do you think you'll have to share revenue with publishers and musicians? Because that's their biggest concern.

Sundar: Taking YouTube as an example, we've clearly done this directly, but I also think it's crucial at moments like these. We're not the only players. These are massive disruptions on the horizon. Our goal is to assist the music industry by collaborating with them and supporting them. This might mean giving artists choices and control over transformative works, allowing them to have a say and helping to find the right solutions.

Nira: Do you think you have to stay ahead of the law? The question here is, are you saying: 'There are many players, perhaps the government should do this, so when AI Drake sings, Drake gets paid.' Or do you believe you must stay ahead of the law to be a good partner?

Sundar: We all need to understand user trends and developments. So, when we say bold and responsible, I mean it. You'll see us being bold in certain situations, but it's all grounded in a responsible direction. We want to do things right.

Nira: We have to wrap up. So, let me ask the most comprehensive question. When Google turned 25, you took over from the founders, transforming it into Alphabet. You've done an outstanding and highly successful job at Google, especially in business growth. Here, we've discussed major decisions—restructuring, appointing new leaders, shifting people, changing culture for focus, navigating regulators, competitors inviting you to dance. All this requires a special kind of ambition and focus. So, personally, what drives you to lead the company through this moment?

Sundar: It comes from first principles, with clarity. I believe in our mission. For me, access to technology has profoundly impacted my life. So, my motivation has always been to bring information and computing to more people for societal benefit. From that, everything else I need to do becomes clear. In a way, starting from first principles simplifies things.

But this is an exciting time. Look, I've been preparing for this AI moment for over a decade. At Google, bringing in Geoff Hinton, building Google Brain, or acquiring DeepMind wasn't accidental. We made the necessary investments, developed TPUs, and announced them at I/O about six years ago. This is something I've long awaited. I'm thrilled it's a turning point. But to your earlier question—because we've been at this for 25 years—we know how crucial responsibility is from day one. That's why you heard about our early work on image watermarking and metadata at Google I/O.

Nira: I could have talked about metadata for an hour. You should be glad I didn't.

Sundar: Let's talk another day. So both parts are important. This is an exciting time.

Neri: Sundar, thank you very much for coming to Decoder. I look forward to talking again soon.