WSJ News Exclusive | Google CEO on AI, the Future of Search, Efficiency and Battling Microsoft

Sundar Pichai: This is such an exciting moment because there are a lot of ideas we’ve had in terms of how we can help our users, but you didn’t quite have a powerful technology capability to actually realize those ideas. I think we’re moving fast, and when I look at our road map for the next few months, we’ll be bringing out a lot of these things like we’ve done in the last few weeks. Workspace has announced features both in Gmail and Google Docs, which are beginning to roll out. We’ve announced Bard. There’s a lot more to come. 

WSJ: What is the key to getting people to move fast on this, and do you still think there is room to move faster?

Mr. Pichai: You always want to think about how you can do things as fast as possible. It’s important to get it right. I think people are naturally energized by the moment. We’ve been incorporating features into our products, but the capabilities have gotten a lot more powerful and amazing, so I think there’s a lot more we can do. We’re also working through a moment to do this well at scale and to make sure you can do it efficiently from a compute standpoint. There are trade-offs involved. 

WSJ: You said you want the company to be 20% more efficient. Have you realized that goal?

Mr. Pichai: We are trying to accomplish that across many different ways. We’re literally looking at every aspect of what we do, and as we said on our last earnings call, we’re thinking about how to re-engineer our cost base in a durable way. We are definitely being focused on creating durable savings. We are pleased with the progress, but there’s more work left to do.

WSJ: How are you measuring the efficiency gains?

Mr. Pichai: It depends on the area. You can look at utilization of your buildings, or you can look at efficiency of your machines, how they’re getting utilized at a company level. You’re able to look at it in the context of your operating expenses and margins over time. We also want to make sure we are pursuing all the long-term opportunities we have ahead of us with the investment it needs. It’s a balance getting both of them right, but we’re doing both.

WSJ: Would you say there’s some room to go on getting to the 20% target you laid out?

Mr. Pichai: I think so. Some of these changes will take time. Some of it is just ongoing work. We have utilization targets, and we have quarter-on-quarter improvements in terms of our infrastructure, and sometimes we have made the longer-term [goals] to get there, because it needs some new technical work to be able to realize that.

WSJ: I’ve heard from some employees that their division has a goal to cut operating costs by 20% this year. Is that happening companywide?

Mr. Pichai: In different areas we have different subgoals, but at a company level, do we have goals to durably improve our margins and drive long-term savings? Yes, but not specifically worded that way.

WSJ: Are you thinking about a second round of layoffs?

Mr. Pichai: We’re very, very focused on this set of opportunities we have, and I think there’s a lot of work left. There’s also an important inflection point with AI. Where we can, we are definitely prioritizing and moving people to our most important areas, so that is ongoing work.

WSJ: Under what scenario would that be a consideration again?

Mr. Pichai: We are comfortable with our approach. We have a clear view of what we need to work toward, both in terms of innovating and making sure we are able to build the things we need to, as well as making sure we are being more efficient as a company. 

WSJ: There has been a lot of prognostication around how search might change because of large language models [LLMs] and even the rise of chat-based interfaces. Do you see link-based search as the dominant way people access information on the internet a decade from now?

Mr. Pichai: I think the experience will evolve substantively over the next decade. We have to meet users in terms of what they are looking for. It’s always tough to predict all the ways in which the future manifestations of this will play out, but I think it’s important to understand what users are trying to accomplish and work back from that at any given moment. 

WSJ: What are you hearing from your users about what they want from Google?

Mr. Pichai: Through recent developments, we can now think about serving them in a much broader way. Even, for example, through Bard we are now able to handle a lot of creative and collaborative use cases, so those are all exciting new directions for search as well. The problem space, the opportunity space, if anything is bigger than before, and so that excites us. 

WSJ: What role do you think AI chatbots in particular will play in search? You’ve talked about them as a companion, but do you think people will be retrieving information through these AI chatbot-based interfaces?

Mr. Pichai: Both. I think we’ve been using LLMs to improve search quality and the search experience, but I think we will bring natively the modern LLM capabilities in search. We are working to make sure it works well for users—they have a high bar, and we want to meet that bar. But yes, will people be able to ask questions to Google and engage with LLMs in the context of search? Absolutely. 

WSJ: Do you see it evolving into sort of a conversation? 

Mr. Pichai: We are testing a variety of approaches so I don’t want to comment on the future iterations of it yet, but yes at a high level people come to Google to ask follow-up questions. We’ll be able to give them more powerful tools to be able to do that.

WSJ: When you say the opportunity space is expanding, what do you mean exactly?

Mr. Pichai: There are newer types of queries which you can ask search, which you may not have thought about asking before. Would you have considered asking Google, help me write a poem? I’m sure we saw queries like that, but maybe in some of those things now we can do a much better job than what we have done before. There are queries in the past where the concept of a single right answer doesn’t make much sense. I think LLMs do great in those scenarios. 

WSJ: What kind of commercial potential do you see for LLMs and LLM search?

Mr. Pichai: It’s tough to carve it out that way. You may be interested in ideas for how to celebrate a birthday, and at some point aspects of it become more commercial. We don’t come into it with the view of, we want to give you this commercial journey. It’s important to get the order right.

WSJ: With what you’ve released so far with Bard, Google has been adamant that it isn’t a search product. Why not?

Mr. Pichai: We’ve put it as a companion to Google search. There are times if you want to search, we make it convenient for you to search. But over time, users will use these products, and we’re learning from it.

WSJ: What was behind the decision to hook up Bard to the web? 

Mr. Pichai: We are a company which grew up on the internet. Where there are elements of factuality by grounding it in search, I think that helps improve the experience, so I think it’s been a natural direction for us.

WSJ: Can you expand on elements of factuality?

Mr. Pichai: If you ask me, can you help write an essay about Thomas Jefferson, hopefully what we say about Thomas Jefferson is grounded in actual events and things which actually happened. Factuality there matters more. 

WSJ: Factuality around current events—is that something you’re thinking about with Bard?

Mr. Pichai: Definitely, yes. It’s something we want to make sure we’re getting right.

WSJ: Do you see any tension that Bard is this box that is hooked up to the internet that provides you answers and Google is this box that is hooked up to the internet that provides you answers?

Mr. Pichai: In the early days, I don’t see it as a tension. I use both products. Would there be queries which are overlapping? Absolutely. There are things we can do in Bard which may be difficult to do in search, or vice versa. I don’t view this as a constraint. I view this as a great opportunity to innovate. 

WSJ: How much of a threat do you view Microsoft right now in this space?

Mr. Pichai: We’ve always competed with Microsoft across many of our areas. We have operated in a very competitive space, and we’ve always said, you have to earn your position all the time. Be it mobile or AI, you always have these moments in which you have to make sure you are innovating and staying a step ahead and delivering what you set out to do. I view this as no different.

WSJ: You don’t think it’s different that now they’re saying explicitly we’re coming after Google and we’re pioneering this new chat-based search interface?

Mr. Pichai: They’ve been working on Bing for a while, and they’ve always been focused on improving their product, and we have been. We’re all going to incorporate the latest advances. Now we get a chance to give users direct access to LLMs in a way which is new. All this, to me, is a natural journey of how we are all going to iterate and make progress, and we are in very, very early days.

WSJ: Why did you not release LaMDA [Language Model for Dialogue Applications], the technology behind Bard, earlier?

Mr. Pichai: When you have new technology, iterating it and coming up with a good product-market fit through which you can launch these products is what all of us are working toward. We were iterating to ship something, and maybe timelines changed, given the moment in the industry. It’s been incredible to see user excitement around adoption of these technologies, and some of that is a pleasant surprise. We are taking that into account and meeting the moment.

WSJ: You were looking for product-market fit—were you not seeing that before the past few months, when ChatGPT became this overnight success?

Mr. Pichai: I’m sure OpenAI debated when—I think they had the product inside for a while, too, before they decided to ship it. Maybe their opportunity window was a bit different as a startup compared to a company like Google. We have to stay focused on how well we can use this technology and build it into our products, which we are planning on doing.

WSJ: Do you think you’ve missed out on anything by not being first to market with Bard?

Mr. Pichai: Throughout our history, there are many areas where we haven’t been the first to market something. We didn’t develop the first search engine. We didn’t develop the first browser or the first email product or the first mapping product, and so on. There are times when being first matters a lot. There are some times it doesn’t matter.

WSJ: What is next for Bard? 

Mr. Pichai: You will see us constantly evolve the product. There are better capabilities, better models, better tools we can connect it to, and so I think there’s a lot ahead.

WSJ: What kinds of capabilities are you adding?

Mr. Pichai: Coding is yet to come, but it will happen in the next few weeks. I don’t want to comment on the future road map here, but you will continue to see us make progress there.

WSJ: Do you plan to open-source any AI models like Pathways Language Model?

Mr. Pichai: We’ve open-sourced models before, but we’ll have to evaluate as we go ahead. I think it has an important role to play, and so I think we’ll look at it as an ecosystem. 

WSJ: You released the PaLM application programming interface. A lot of other people are releasing APIs [application programming interfaces] on top of their large models. Do you think this sort of access will center on a few large tech companies, as a lot of the current internet has, or will it be more diverse?

Mr. Pichai: My sense is it will be a lot more diverse than people think. Over time, you will have access to open-source models. You’ll be able to run models on-device. Companies will be able to build their own models, as well as people who use models through large cloud providers. I think you’ll have a whole diverse range of options.

WSJ: Traditionally Google Brain and DeepMind have been fairly separate. Are you rethinking that?

Mr. Pichai: There have been papers published between the two companies. They both work on joint projects today across the companies. I view that as something which has been organically happening. If anything, I expect a lot more, stronger collaboration because some of these efforts will be more compute-intensive, so it’ll make sense to do it at a certain scale together. Being able to bring the best together, I think, will help us lead to better breakthroughs.

WSJ: What does Google search look like in 10 years? Does it look more like the original Google—the 10 blue links—or does it look more like Bard?

Mr. Pichai: It may look like neither, or it will have elements of all that. I think we’ll be able to help users in much deeper ways. If I take a 10-year outlook, this will all be more ambiently available to users in radically different ways than we use them today. I think you’ll be able to do all of this in a much more personalized way, which means by nature we’ll be able to impact users in a deeper, meaningful way.

WSJ: Do you think TikTok should have a future in the U.S.? Should it be banned?

Mr. Pichai: These are important questions for the U.S. government to answer. Their app is on Google Brain. To us, they’re a partner we work with, and we make sure where there are shared experiences, we do that well. But these are questions not appropriate for me to answer.

WSJ: And you’re happy having them on the Play Store?

Mr. Pichai: We treat them like every other developer. They have to comply with our policies, and we make sure they comply with all that. We are protecting our users, so both of them are important to us.

Write to Miles Kruppa at [email protected]