Meet The Man Who Runs Google Search — With Prabhakar Raghavan

Prabhakar Raghavan is the Google senior vice president who oversees Search, Maps, Ads, Commerce, Payments, and Google Assistant. He joins Big Technology Podcast to discuss the evolution of these products, how they're integrating cutting-edge artificial intelligence, and why he's steering Google tech products toward more human-style interactions with users. Raghavan joins fresh off an appearance at Google's Search On launch event and breaks down the new features announced Wednesday. Learn more about your ad choices. Visit megaphone.fm/adchoices

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Alex Kantrowitz?

  This episode is brought to you by Genesis. What does it actually take to put Agentic AI to work across an entire enterprise? I recently spoke with Genesis chairman and CEO Tony Bates about why orchestration is emerging as a competitive advantage in customer experience and what companies need to get right in the AI era. Then I sat down with Adam Mitchell, head of enterprise business solutions at Voya Financial, to go inside a large-scale Agentic CX transformation, including the business decisions, trade-offs, and lessons along the way. You can watch both conversations now on my YouTube channel today. It's never been more challenging to get frontier compute. How do you know where to look, who to trust, or how much to pay?

• 0:43Alex Kantrowitz?

  Akash is an open cloud that provides easy access, flexibility, and peace of mind. That's how companies like Razer were able to cut their inference cost by 15 times. Visit akash.network/alex for a free consultation. That's akash.network/alex for a free consultation.

• 1:03Speaker 01

  LinkedIn Presents.

• 1:19Alex Kantrowitz?

  Welcome to the Big Technology Podcast, a show for cool-headed nuanced conversation of the tech world and beyond. We are joined today by a really special guest. He runs, and let me know if you can give anything else inside Google that is not under his umbrella, but Search, Ads, Commerce, Maps, Payments, and Google Assistant. I'm talking about none other than Prabhakar Raghavan, Google's Senior Vice President for all of the above. Prabhakar, welcome to the show.

• 1:48Speaker 02

  Hey, Alex. Thanks for having me on your show. And it's a super exciting time with our Search On event with so much to share. I'd love to share more about it today.

• 1:59Alex Kantrowitz?

  Also, I hope we hear some details that you have presented today. Just for listeners edification, we're getting Prabhakar right, basically hot offstage at Search On, which is just taking place. And then this pod goes basically as the folks get offstage. So it's a big event that Google uses to talk about search. So over the next 45 minutes or so, we're going to talk about You know, is there actually any innovation in search? What's happening with maps? And then some of the themes that we like to talk about more on the show. Broader market outcomes and you know, maybe we get some hints about where everything else is going. Because Prabhakar has so much visibility into a lot of the worlds of tech.

• 2:43Alex Kantrowitz?

  So let's talk about search to start.

• 2:46Speaker 02

  Okay.

• 2:46Alex Kantrowitz?

  That sounds good.

• 2:47Speaker 02

  Great.

• 2:47Alex Kantrowitz?

  Okay. So you just did a search event and the natural question that I ask is, why are we doing an event for search? Because you know, search, of course, there's been innovation in search, maybe the algorithms get better. But at the end of the day, the way that we're searching today is very similar to the way that we started when Larry and Sergey, you know, founded Google, where you write what you need to find into a search bar. And then you get these links. Okay, maybe a little bit more information. But the biggest innovation I heard about search is now that people are searching on TikTok when they're trying to find stuff. So from your vantage point, where is the motion? Where's the movement in search right now?

• 3:27Speaker 02

  I think the biggest motion, I'm glad you appealed to our founders to begin with. So they laid down the mission to make information universally accessible and useful. But the way we pursue that mission has been exactly like you said. We've kept users in a box. It's a text box. And we've conditioned users over the last two decades to express themselves really precisely. And when they do, we give them the perfect answer. The world has moved on in a couple of respects. Number one, increasingly, there's user intents that don't have the single right answer that has to go on top.

• 4:06Speaker 02

  So that's one way things have changed.

• 4:07Alex Kantrowitz?

  Can you give an example of that before we move on to the more fundamental issue?

• 4:11Speaker 02

  Let me give you two examples, since you asked for one. The first is if you ask the query, yellow dress, your intent is to find yellow dresses. There isn't the world's best yellow dress followed by the second best yellow dress and so on. What we should be offering you is a palette of choices and then iteratively adapting to your taste.

• 4:32Speaker 02

  This means we're going to have much more visually rich, immersive experiences where you browse through a large number of results. It's no longer about what we used to call precision at one, which is, you know, you have to nail the right answer on top. So that's one example. Second example, dinner seating arrangements at a wedding. There is no right answer for that, but there's a plethora of opinions and a lot of good information out there from people's, so a lot of knowledge in people's heads that is captured in discussions, forums, et cetera. And we need to do a much better job and have been improving our focus on elevating the highest quality of that content. And so that's two examples of going beyond, you know, the single right answer.

• 5:15Alex Kantrowitz?

  Yeah, it's surprising to me that you want people to spend more time in search typically. Like the idea was get them to the most relevant website right away. Is that no longer the case?

• 5:25Speaker 02

  You know, the examples I gave you aren't particularly about keeping them on or off. If it's wedding seating arrangements, you're off to finding the best ideas, but you will go back and forth. So it's not about time spent, right? But let me get to the, the second, I think, deeper point.

• 5:42Alex Kantrowitz?

  Yeah.

• 5:42Speaker 02

  Underlying your first question, right? Which is over the last two decades, we, we've trained users to that if they express their query perfectly, we'll nail the answer, right? But that's a box we have to get out of. And over the last two decades, network and storage costs have come down, which has the implication that it's easier than ever to create and disseminate rich content, whether it's videos, images, et cetera, right? What that means is people should be able to express their interests much more multimodally and get responses that are much more multimodal.

• 6:23Speaker 02

  One immediate consequence or consequence we've developed over the last few years technically, but this is what you'll hear a lot about from our search on announcements, which you've just heard. The camera is your new keyboard. The camera is in your keyboard, meaning instead of forcing the user to type the intent in a particular way, you can just hold up your camera and capture this, express this intent that is so hard to put into words. And with that, you're off to the races. Being able to append to that text, voice, et cetera, you can express yourself much more richly. And that's what we're after.

• 7:01Alex Kantrowitz?

  How does that work when you're searching? So just thinking about the natural way of doing things.

• 7:07Speaker 02

  Yeah.

• 7:07Alex Kantrowitz?

  I like never will like point a camera at something and be like, you know, this is usually my, if I'm at, if I'm using a camera, I'm at my destination. Search I use as a point of trying to figure out, you know, I'm at the, the earlier stages of discovery.

• 7:23Speaker 02

  Yeah. So you're walking along the street, right? Let me give you two examples. Just, you're just out there in the street, right? Uh, you walk by a store. Normally you window shop, now you can point your camera and summon up exactly the thing that you see in the window and say, oh, that's what that is, right? Learn more about it. Right? Uh, so that's one use case. The second use case, which you just also announced, uh, search on, imagine holding up that camera in a busy street in Tokyo and saying, I'm looking for a coffee shop that's not too busy right now. That is the capability we're bringing with the live view and maps, right? So the camera now becomes an input modality that

• 8:06Speaker 02

  is far more powerful than your destination, right? It's already been underway, Alex, for the last, I would say, five years with Lens usage having grown to over eight billion queries a month with us, right? So it's not a new.

• 8:19Alex Kantrowitz?

  How does that compare to, to, to type queries? So eight billion queries on Google Lens?

• 8:24Speaker 02

  Still small, but growing.

• 8:26Alex Kantrowitz?

  Yeah.

• 8:26Speaker 02

  It's, yeah, but growing many fold faster, right? So that the point is especially with younger users, these newer modalities are the ones that are taking off because these are users who weren't trained to search the old fashioned way like you and I.

• 8:40Alex Kantrowitz?

  Yeah, yeah, I'm definitely in the old fashioned search way. So you'll excuse me if my bias is to the old way. It does take me some time to figure out the new way. But let's go back, let's keep on this line of thinking. So talking about, you know, discovery versus you're already there. Okay, now we're into maps, right? It doesn't seem to me like when you're searching a map, so maps obviously a big part of Google search now, even though people might think of it as a map program, it's a search program. But when I'm trying to find a coffee shop, if I type in coffee shop, I can see all the little red, you know, what do they call them? Pointers, placers?

• 9:21Speaker 02

  Pins, whatever.

• 9:23Alex Kantrowitz?

  With the coffee shops, the star rating, that's pretty efficient. And in one view, I'm already basically looking at my options.

• 9:31Speaker 02

  Yep.

• 9:32Alex Kantrowitz?

  When I, and I, you know, Google's been talking about use the camera to search, you know, use lens, eight billion is not, eight billion queries a day? Not bad.

• 9:42Speaker 02

  Yeah.

• 9:43Alex Kantrowitz?

  How long, how often is it?

• 9:44Speaker 02

  Eight billion a month.

• 9:45Alex Kantrowitz?

  A month? Not bad. But I think with most users, especially we're talking most of the globe, not on fast internet speeds, not on fast phones. So it ends up being this much clunkier experience where, you know, I could have typed, it takes me maybe, you know, five to 10 seconds to query something on Maps as it is today. If I want to load the camera, then try to get it connected, have these, you know, augmented reality features catch, that takes so much more time. So I understand that Google, you know, is taking advantage of the fact that we have higher processing speeds, you know, more data, all that stuff. But when the rubber meets the roads, I wonder if that's, you know, what's actually happening for the average user.

• 10:26Speaker 02

  Let me give you a couple of data points, two data points to sort of set that straight, right?

• 10:33Alex Kantrowitz?

  Yeah, crush me on this one. Yeah.

• 10:35Speaker 02

  In many locales with new internet users, right? Text is not the dominant form of interaction, right? In India, one in three queries is voice already, okay? Which means we have to get the audio voice rec and regurgitate results. So people have already moved beyond text in those locales. That's because of what we started with. These are users who were never trained to do the traditional Google two key word search. Number one. Number two, I'd argue that the comment about, you know, networks are slow and so on.

• 11:14Alex Kantrowitz?

  Yeah.

• 11:15Speaker 02

  Was true maybe five to 10 years ago, but in a lot of the newer markets that kind of skipped a generation of telecoms. Uh, they've gone to such high speed internet, like if you've seen what Alliance has done in India and Africa and so on. Uh, that, so I don't think it's bandwidth is no longer a constraint. This is the point I was making. Networking has gotten cheaper, faster than computation. So computation is really still the thing that holds us. And that's why you see us and others putting more computation on the phone.

• 11:47Alex Kantrowitz?

  Mm hmm.

• 11:48Speaker 02

  So this overhead that you're thinking of like, well, I have to fire up a camera, insert IR, you know, the kit. Uh, all of that is pretty instantaneous right now, even on what you would consider a low end phone in an emerging market. Last thing I'll add is, remember, we're talking about users. I said they're not used to the traditional Google search. They also have never seen a paper map. So why, why would you make them look at a paper map and search one? And you can just hold this thing up. It's almost like you're looking at the world naturally and intuitively. And make it work the way we think as humans.

• 12:24Alex Kantrowitz?

  So let me take one more swing at, at standing up and making the case for the paper map.

• 12:30Alex Kantrowitz?

  And, uh, all right. So here, here it is. The paper map allows me to search four, five, six, seven, 10 blocks away.

• 12:37Speaker 02

  Yeah.

• 12:38Alex Kantrowitz?

  You know, maybe a half mile if I'm thinking about walking. The augmented reality, point your phone somewhere, you know, it's generally like what's, you have a really hard time building a UI where you can see what, you know, I'm at a coffee shop that's four and a half stars, but if I were to walk 10 minutes, I'm going to get to the 4.9 star. And with a paper map, that's easy to see with, with the augmented reality stuff, not so much. So, so how do you respond to that, um, piece of advocacy I'm doing for the old, old school style?

• 13:08Speaker 02

  Uh, so let me give you a response that's actually built into our, uh, AR directions product.

• 13:16Speaker 02

  Right. If you come out of the subway in London, Tokyo, New York even, where you are, right, my, my biggest challenge visiting these places is I am headed for that restaurant ten blocks from the subway station. I don't know which way to go. I hold up the phone and it shows me the first step, go this way first and that, that. So it's not like it has to be direct line of sight for me. But the ten block away restaurant is represented to me in a different way than the paper map, right? And I understand that if you've grown up with a paper map, it might be somewhat intuitive, but it's almost as easy to hold this thing up and say, well, start by going up there and then we'll tell you what's next, right? So it is

• 13:56Speaker 02

  a different, you're right that it's a different UI. Uh, you're right, it takes more imaginative work from a design perspective. I'm not convinced that from a cognitive load in the user, it's worse. I actually think people get acclimatized to it pretty much.

• 14:13Alex Kantrowitz?

  All right, let's talk about form then, because when I hear your talk about, hey, Google, instead of necessarily being, you know, two D is really overlaid over the three D world, I think to myself immediately, how does Google move to its next iteration without being an augmented reality company? And it's interesting because you have Snap, clearly an augmented reality company. That's what they say on their branding. Um, Meta, it's used to be Facebook, now it's Metaverse. That includes AR. We have Apple, we know Apple is working on these AR glasses. Are they ever going to ship them? That's another question. Maybe you have some intel on that. We can ask you about that. Um, Google's tried, right? We know, you know, Google Glass

• 14:54Alex Kantrowitz?

  has tried and failed, but it seems to me like it'd be in Google's interest to take another very big swing very soon. What do you think about that?

• 15:04Speaker 02

  So, uh, you began by saying these are all augmented reality companies. I, I think our first and foremost objective is to be an information company, to do whatever it takes to get the user and facilitate their interaction with you as well as possible. If that, in the current technology, requires augmented reality.

• 15:24Speaker 02

  Great, we're starting with the phone. And as headsets move from being clunky, you know, one pound things that hang off your face to something more usable, right? The nice thing about glass, Alex, if you remember, is it didn't feel like an appendage, right? It was just there.

• 15:41Alex Kantrowitz?

  Maybe.

• 15:42Speaker 02

  Well, maybe you say, but, right? In comparison with some of the things you're seeing, right? But necessarily that limited the amount of compute and even networking that was available from those devices. As miniaturization proceeds, we're going to see those trends get to the point where these appended devices are actually much more lightweight and usable. And we're all in, you know, looking at that hard. That doesn't make us an AR company. We will let the core remain an information company.

• 16:14Alex Kantrowitz?

  Okay. So you're gonna, I mean, you've already previewed these glasses. How important of initiative are they inside Google? AR glasses?

• 16:25Speaker 02

  You know, look, we, we're conducting research in a lot of areas and.

• 16:31Alex Kantrowitz?

  Let me put it another way. Do you want, do you want Maps and Search to be a application on, let's say, Apple's device? Or do you want it to be something contained within a Google ecosystem? Because you did do Android, right? Android, the mobile operating system has paid off really well for, for Google. And the, the native Gmail, the native Android, I mean, the native Maps, the native Assistant, you know, some of these products, very core to your portfolio, benefit a lot from having the big app and the operating system. When it comes to augmented reality, do you want that type of thing? Are you, are you content being an app on, let's say, the Apple glasses?

• 17:08Speaker 02

  Well, I'll note that all of the apps you mentioned, including Gmail and Search and Maps function, I hope you'll agree, well on iOS. So it's the goal of any app that's a, aims to serve all users should be platform agnostic and serve iOS users as well as Android users. The intent is not to say, okay, we're going to do better on the one platform or the other, right? So you'll see us continue to do, you know, do our best to be, to function as well on any platform.

• 17:44Alex Kantrowitz?

  Okay, I see what you're doing. There's, you know, some thought, but you can't give away the whole augmented reality roadmap in one conversation. But there's a reason why Google did Android, right? Of course, it's fine to operate on the iPhone. But you know, there's something to be said for having your own operating system. Some of your competitors, Meta, for instance, is finding that out the hard way. You'd agree, right?

• 18:09Speaker 02

  I'm not commenting on what Meta's strategies. I wouldn't profess to know what they want to do.

• 18:14Alex Kantrowitz?

  But for Google, there's a reason why you have an operating system.

• 18:18Speaker 02

  I think to the extent it facilitates delivering the best end-to-end experience, that is the ultimate goal. If we can actually make it work really well for the user, great. And it goes back to ultimately, we are an information company, and that's what we want to service, right? It's not like we have to be an operating system company or an AR company. We are an information company, and that's our bread and butter that we hope to keep doing really well.

• 18:47Alex Kantrowitz?

  Interesting. You know, when we talk about, it seems like I'm hearing a theme here, right? When you talk about the way that people are going to search, the way people might use maps, is that they're interacting with search in a more human-type-focused way, right?

• 19:06Speaker 02

  We call it natural and intuitive.

• 19:08Alex Kantrowitz?

  Yeah, exactly. Okay. Well, yeah, same idea, right? You're, instead of typing in, you know, one or two words, you actually maybe type in a sentence. Instead of looking on a paper map, you show maps the world around you, and then it helps you make sense of it.

• 19:27Alex Kantrowitz?

  On that note, I'm curious what you think about the innovations that we're seeing with Lambda. You know, Lambda is a favorite topic on this show. We had Blake Lamont.

• 19:39Speaker 02

  So I hear. Now, that is a bit of a leap, but let's...

• 19:42Alex Kantrowitz?

  I'm not asking you any sentient questions, because I know you don't believe it's sentient, but... The idea that a chatbot like Landa could end up being, you know, maybe not a person, but the way that you start interacting with Google Search, search that knows you, search that can help you as a friend might. That's pretty interesting. So obviously, we've heard a lot about Google Assistant, another product that you oversee. Does Google become more chat-based or even more? You mentioned that one-third of users in India query with voice. You know, does Google become like super smart Clippy down the road, where you have just a conversation with it?

• 20:27Speaker 02

  You know, I love the way you're leading up to it. The notion of a super smart Clippy, I'm not so sure about.

• 20:34Alex Kantrowitz?

  You know what I'm saying, like a back and forth as opposed to that knows you, that understands your life.

• 20:40Speaker 02

  Yeah, look.

• 20:41Alex Kantrowitz?

  Your context.

• 20:42Speaker 02

  You're absolutely right that having an assistant that understands you and using that understanding to condition your experience is a super interesting idea. But if you look at the current state of large language models, there are at least two difficult challenges to overcome, right? The first we've seen before in a number of earlier times, which we call safety, right? How do you know that the bot, the language model, doesn't drift into territory that's toxic, potentially racist, harms the user, et cetera?

• 21:22Speaker 02

  And it's a very subtle question because harm doesn't just come from one utterance, right? And so this is a deep question for us as computer scientists, together with the colleagues in the social sciences to try and understand, right? And we don't have enough understanding, I would say. So that safety is one thing, right? And we have to get this right. The second part is what we call factuality. You know, we are relied upon billions of times a day to make sure that we stick to the facts as far as we can, right? Now, it's known that large language models can sometimes hallucinate without...

• 22:04Speaker 02

  Necessarily giving any clue that they are hallucinating. So if you came in and asked, what's the height of the Eiffel Tower? Today, we give you the answer. Now, let's say a large language model said, Alex, the height of the Eiffel Tower is three feet, six inches. Right? Now, if you didn't know better, how could you check the factuality? Right? So in other words, that that problem of using a large language model to answer factual questions and ensuring the veracity is is a hard one. So these are the two really hard challenges we'll have to overcome before anything of this sort can be productized.

• 22:40Alex Kantrowitz?

  That makes sense. So it might be a way down the road.

• 22:43Speaker 02

  Yes.

• 22:44Alex Kantrowitz?

  What did you think when one of your colleagues said that your technology was a person?

• 22:50Speaker 02

  No, I I'm thinking, first of all, I see no reason to believe that, right? And again, as a computer scientist, I go back to Turing's original test, which is can you put a machine and a human behind a screen? And fool the person on the other side without them knowing which one is which, right? And in many straightforward interactions, like what is the height of the Eiffel Tower? You could probably fool them, right? But I would say we don't even have a full understanding of what that question means. You know, the the S word, the sentient word, right? We're a few years from resolving that, right? But if you ask me today, I would say there's no basis for believing that there is sentience

• 23:34Speaker 02

  and in a language model.

• 23:37Alex Kantrowitz?

  Right, but also as a researcher coming from a research background, Google doesn't hire dumb people, right? So.

• 23:43Speaker 02

  We try our best.

• 23:46Alex Kantrowitz?

  I've met many, many of your colleagues. I I can confirm that. So just knowing that it actually convinced someone that it was a person. If I'm from like coming from a research background, my reaction might might have just been like, God damn, our our stuff is good. You had some of that?

• 24:02Speaker 02

  Definitely, but remember that, you know, I started with arguably a simpler problem, which is the Turing test. And today, that's not been convincingly passed. You can pass it in slivers, right? But it's not like anybody's claiming AGI, artificial general intelligence, to the point where you can do an end-to-end Turing test. And the questions you're now asking, I think, are an even further step beyond, which is why I think it's going to be some years before we can resolve that. I can understand the viewpoint that as well, we've shown facets of intelligence or sentience. I cannot make a definitive assertion that would be built as intelligent or sentient. I don't think so.

• 24:40Alex Kantrowitz?

  Yeah, I don't, I'm not, I don't think you, you know, I guess that's not what I'm getting at. I kind of think that the interesting and the cool thing is the advance. You know, because we know the chatbots, try to change a flight through a chatbot, you throw your phone out the window. Maybe good for Android's business, but not very good for AI.

• 24:58Speaker 02

  That's, that's actually an institute observation, right? Which is, the progress and, and the, you know, the flashes of brilliance these models show are actually astounding, right? I was a junior researcher at IBM when they did Deep Blue that beat chess, you know, Kasparov chess, right? And at that time, I remember, okay, chess, sure, go, never, not in my lifetime. Okay. It's completely wrong, right? It took about 20 years from then to, to get to go on DeepMind, right? But along the way, I heard about the famous protein solving folding problem. And I'm like, okay, we have no clue how to solve the N-body problem in protein folding. You're never going to do this, right? And guess what? Now they do AlphaFold.

• 25:43Speaker 02

  So in each of these cases, I think there's astounding advances, far beyond what we predicted. And the same is true of language understanding as well, in these language models.

• 25:53Alex Kantrowitz?

  Right.

• 25:55Alex Kantrowitz?

  Prabhakar Ragavan is with us. He's the Google SGP in charge of search, ads, commerce, maps, payments, Google Assistant. What's left of Google? I don't know. We've talked about a bunch. We talked about we talked about search. We talked about maps, a little bit about assistant. When we come back after this, we're going to talk about the state of the market. And then we're going to get to a little bit more of the products that Prabhakar oversees. Back right after this. This episode is brought to you by App Point. Everyone's racing to roll out AI right now. Co-pilots, chatbots, agents doing real work. But here's the part nobody loves talking about. All that AI runs on your data and most teams have no single way to see it, secure it, and prove it's under control. That's exactly what App Point does.

• 26:40Alex Kantrowitz?

  For 25 years, they've been the trusted layer beneath the world's most demanding data, now extended across your entire AI estate, your data, your cloud, and the agents acting on your behalf. It's how more than 28,000 organizations deploy AI with confidence. So innovation scales without scaling risk. App Point, the unifying trust layer for AI. Head to apppoint.com to see how enterprises deploy AI with confidence. Learn more at avpt.co/bigtechnologypodcast. That's avpt.co/bigtechnologypodcast. Violent threats against executives are growing at an alarming rate.

• 27:23Alex Kantrowitz?

  When a company experiences backlash, executives are the first people blamed. When their address and other personal details are sitting online, this backlash can lead right to their front door. The team at Ironwell knows this better than anyone. They've protected some of the most targeted executives and individuals on the planet for almost two decades. As a wide-glove enterprise security service, they protect your people with continuous personal data removal, proactive prevention tools, and human-led emergency support. So when someone goes looking for your executives, Ironwell ensures they hit a dead end. So here's what to do. Go to ironwell.com/bigtechnology, fill in the quick form, and request your free risk assessment. Their team will show you just how exposed your executives are and how to lock it down before a threat reaches their front door.

• 28:08Alex Kantrowitz?

  That's ironwell.com/bigtechnology. Learn how to stop online threats before they become real-world attacks. The link's in the show notes.

• 28:25Alex Kantrowitz?

  And we're back on the second half of Big Technology Podcast. We're joined by Prabakar Agavant, Google's Senior Vice President for Search Ads, Commerce Maps, Payments and Google Assistant. When you're in the elevator, do you try to someone to someone say, hey, what do you do at Google? You can't list all those by the time the floor rings, right?

• 28:43Speaker 02

  I say I'm a scientist working at Google.

• 28:46Alex Kantrowitz?

  Yeah, that's much better, much better. Cause then you start to end up in a conversation like this one. And once in a while, I imagine those are fun, but not all the time.

• 28:55Speaker 02

  It's fun, Alex.

• 28:58Alex Kantrowitz?

  Fun for me, too. A week ago, I wrote about some of the pullback happening, especially when it comes to big tech spending.

• 29:08Alex Kantrowitz?

  Sundar, your CEO has talked about 20 being trying to be 20% more efficient. We've heard conversations coming out of meetings where people are asking about perks and he's having to double down on the fact that like, listen, the cool thing about being at Google is you had to work on new projects, not that you, you know, I don't know, I guess use the most cliche example, play ping pong or go down a slide.

• 29:30Speaker 02

  By the way, there's people, you know, from this floor going down on slides right now as we speak.

• 29:35Alex Kantrowitz?

  That's good to know. I'm glad we can at least confirm slide health is excellent at Google. It's nice that people are back in the office. So, you know, I imagine that your projects, especially search, right? Your products, search, maps, ads, these are the type of things that are core, that are going to get, you know, lots of investment. The thing that I wonder about is what happens to those, you know, weird projects, things that used to be common at Google, maybe still are in some ways, like Gmail, kind of builds on the side, you know, are they still going to get funding in a time where, where there's a response to the market? I'm just going to read you the headline of my, my story last week. I'm curious if you think, what you think.

• 30:16Alex Kantrowitz?

  So headline is big tech enters a new era of scaled back ambitions as the stock market contracts. Is that fair or do you dispute that?

• 30:26Speaker 02

  I cannot speak for big tech, whoever that is, but I can speak for Google.

• 30:29Alex Kantrowitz?

  Yeah, let's do that.

• 30:31Speaker 02

  I don't think we should be scaling back any ambition, right? At our core, there are two things I believe we've always done well. Number one.

• 30:41Speaker 02

  Move the technological needle, independent of non-need. Like if you see us stamping out large language models every year, it's not because we know that, okay, this next language model is going to do this exact thing in search and so on, right? There was BERT, there's MOM, Lambda, ARM, and you've seen that sequence, right? I expect that progress to continue unabated and it'll run a little ahead of product reality. The second thing, having worked at many research labs and then come here, right, Google is pretty speedy in moving this stuff into product, right? So the BERT motion into product, for instance, is pretty much over. MOM is having a pretty healthy run now.

• 31:24Speaker 02

  And all of these, if you count the years.

• 31:27Alex Kantrowitz?

  What's that?

• 31:28Speaker 02

  MOM is, geez, multimodal user models or something like that, right? Sorry.

• 31:36Alex Kantrowitz?

  Okay.

• 31:37Speaker 02

  I mean, I just think of it as MOM.

• 31:39Alex Kantrowitz?

  Yeah.

• 31:40Speaker 02

  Good, but it's a large model that combines text, images, videos and across many languages. That's MOM.

• 31:48Alex Kantrowitz?

  Hmm.

• 31:48Speaker 02

  And that is now starting to make its way into products, right?

• 31:51Alex Kantrowitz?

  And how's that happening? Yeah.

• 31:53Speaker 02

  It's showing up, for instance, in some of the search on announcements we just did around review content, right, and enhancing the fidelity of those. So that motion has taken about two years. Okay. And we need to keep priming the supply of amazing advanced technology, new language models, new lens technology and so on, to keep the products vigorous, vibrant and the best we can for our users. So I expect that to continue unabated. I don't think we should scale back ambitions in any way, right? You used the word, if I recall, weird, Alex?

• 32:34Alex Kantrowitz?

  It's one of my, yeah, favorite words on the show.

• 32:37Speaker 02

  Okay. You know, I don't know what counts as weird. The way I might interpret that, and you tell me if this makes sense, is anything, any technology that doesn't have a non-purpose of path to market.

• 32:52Speaker 02

  And if you take that, I would assert that not only are you hearing about it, you know, at I O here and beyond at our research conferences, I fully anticipate you'll hear more advances in the time to come, right? When I see some of the early work coming out of research and DeepMind, I'm super excited about what's coming up.

• 33:13Alex Kantrowitz?

  You have any good examples that you can share?

• 33:15Speaker 02

  I think my colleagues in research and DeepMind should be the ones to share that. It's their babies.

• 33:22Alex Kantrowitz?

  Okay, so what you're saying is not, it's not public versus?

• 33:25Speaker 02

  It's a lot of this is not public and I expect them to keep prying the pump.

• 33:30Alex Kantrowitz?

  Yeah, because I mean, I do remember when I was doing research for my book, talking about, you know, the, the large bets that just really had no immediate payoff. Things like natural language processing research that Google was doing, that eventually made its way into the assistant. And if we do end up moving to this more conversational interface with technology versus the, you know, taps and types like that I like, that stuff, that stuff is really important.

• 33:58Speaker 02

  Exactly. So important. And, you know, we cannot have a hiatus in that.

• 34:04Alex Kantrowitz?

  So how do you then become, I mean, are you, yeah, how do you then become more efficient without risking cutting that? And I'll just go one step deeper. How do you keep the people that are most eager to work on the stuff that doesn't have immediate results if everything that's going on in the company, the rhetoric from the top, the market is saying, profit now?

• 34:27Speaker 02

  I don't recall us saying profit now, but I think

• 34:30Alex Kantrowitz?

  More efficient.

• 34:31Speaker 02

  The, the two things you asked, how do you get more efficient? And how do you keep the best talent? Actually, I have the same answer curiously, right?

• 34:37Alex Kantrowitz?

  Okay.

• 34:38Speaker 02

  And the reason is when I talk to some of our best people, the biggest complaint I get is what they call bureaucracy. Bureaucracy is a big word, right? But it's too many layers of decision making, you know, they want to move fast. And sometimes like, well, I have to ask this person and check Joe over there and Mary over here and blah, blah, blah, right? And I think there's necessary gains to be had to cut through some of that. We've become a large company and there's, we can definitely harvest some of the deficiencies simply by making, streamlining a lot of our decision processes. And

• 35:20Speaker 02

  Sundar certainly was pointing at that and kicked off a bunch of internal sprints to actually try and clean that up. And uh, now just from, well, the, the great thing is Googlers care, right? We got thousands of ideas from, you know, Googlers and here's something that's staring you in the face, please do it, right? And we're triaging that and getting through that. I definitely anticipate we'll get a lot of that. Now, the second part of your question, if we are successful at cleaning up this craft, then it will be the case that the most talented employees will be the ones who will want to remain and not be daunted by the bureaucracy they perceive.

• 36:06Alex Kantrowitz?

  Do you think Google's moving slower today than when you joined?

• 36:10Speaker 02

  You know, it's very hard to say. It's, I've been here a little over 10 years and uh, the, the, the two things that survive, right? Number one, continuing to explore ideas, research that seem far ahead of their time. And you know, I gave you examples even of how I myself have been surprised by the speed at which some of this has emerged and actually moving that stuff into, into applications, right? Because if you thought about reinforcement learning and some of that stuff, right? Yeah, it was good to when go, when it go. But then you start to see this amazing application like protein folding that comes out of some

• 36:55Speaker 02

  of that, right? It is mind blowing in the impact it can still have ahead.

• 37:01Alex Kantrowitz?

  So can, can you unpack that for a minute? Um, winning at Go, so we, a lot of people know that DeepMind's technology is able to win this very difficult board game, AlphaGo. And that, that technology was used to unpack proteins?

• 37:15Speaker 02

  So, uh, the protein folding problem basically says, here's a molecule, it's very big molecule, it's typically an organic molecule that's sitting somewhere in our bodies, right? Uh, what is the three dimensional shape it falls into? Because that is key to drug discovery and, and, uh, you know, triaging and understanding disease, disease vectors, right? And solving that classically, using methods from physics and numerical computation was solving something called an N-body problem, which we really didn't know how to do at scale, right? And somewhat stunningly, some of the same insights and machine learning advances

• 37:53Speaker 02

  that helped beat humans in the game of Go could now be repurposed from AlphaGo to AlphaFold to fold proteins. And now hundreds of thousands of proteins have now been successfully folded, which is like a quantum leap, right? And we put that out for availability to researchers, life sciences researchers world over. And I was recently at a conference where they said every major lab in understanding diseases and finding cures is making some use of AlphaFold. And that is a very proud moment for us. And so the point I'm making is we want to continue

• 38:35Speaker 02

  pushing the frontiers because there's almost an inevitability with the applications that follow, right? And that's what I think that spirit persists at Google from 10 years ago. And I don't think we're going to lose that anytime soon. I think we're still very committed to those kinds of advances.

• 38:53Alex Kantrowitz?

  Hmm. Okay. Interesting answer to the moving slower question, but we'll take it. The, I gotta ask you about ads.

• 39:01Speaker 02

  Okay.

• 39:03Alex Kantrowitz?

  There's an advertising slowdown going on right now as the economy gets tougher. People have started to, you know, slow down their ad spend, and especially on digital advertising where you can shut it off immediately. That's where we're starting to see the biggest problems. We're seeing action from the Federal Reserve that leads us to believe that, you know, we're going to go into a deeper hole in the economy right now. As someone who's working on that product and sits on top of ads at Google, how do you feel about all this? And is the slowdown as bad as people are saying?

• 39:33Speaker 02

  So first of all, there's one thing I've learned about microeconomics is it's uncertain.

• 39:41Alex Kantrowitz?

  Hmm.

• 39:42Speaker 02

  Right? It's always uncertain. And some of the easy lines and connections you draw don't necessarily pan out exactly as expected, right? So far.

• 39:53Speaker 02

  First example, like when, what is it now, uh, two and a half years ago, COVID hit. An immediate conclusion, not us, me, Google, but the world over, was, oh, my God, the world economy is going to grind to a halt, right? And then you saw these pictures of ships stranded in ports and so on, right?

• 40:12Alex Kantrowitz?

  Yeah. Senators selling their stock portfolios.

• 40:17Speaker 02

  So I hear, I haven't followed the senators closely, right? So, uh, so that didn't quite play out with the, the way we anticipated, right? The economy didn't collapse immediately, right? So the, the thing I'd say is, from our perspective, if we can be, continue to be really good at providing value to advertisers, measurably so, right? While preserving user quality, that's the best thing we can do, whether in a great economy or whatever this is turning into, right? This uncertainty.

• 40:56Speaker 02

  And so I, I, I don't try and draw a line from what the Fed is doing to what our products have to do. Our products always have to be best to breed, and that's it.

• 41:08Alex Kantrowitz?

  And, and at Google, like, I get, as much as you can say, like, does this slowdown feel real and prolonged? Or you think it's overblown?

• 41:18Speaker 02

  Uh, I, you know, now you're turning me into a macroeconomics PhD, which I totally am not.

• 41:27Speaker 02

  So, uh, so you're asking me to read the future, and I don't think I'm going to try to read the future.

• 41:32Alex Kantrowitz?

  Okay, that sounds good.

• 41:35Alex Kantrowitz?

  One of your, one of your competitors, Apple, is blocking tracking in iOS or enabling users to block track in language that strongly suggests they should block it. What do you think about that? Is that a good step for privacy? Or is that, you know, something that you'd rather not deal with? My understanding is it's hitting some Google products as well. Is that, is that a good example for, for the rest of the economy? Or what do you think about that?

• 42:00Speaker 02

  So, so let me unpack a couple of layers of that. So first of all, I think the objective of protecting user privacy and trust is, is absolutely the right one, whether for them or for the whole industry, right?

• 42:15Speaker 02

  I'm not of the belief that...

• 42:19Speaker 02

  Privacy and advertising are at odds with each other. And I think it's quite feasible to deliver high quality advertising while preserving user privacy. But it does take a couple of things. One, uh, the best way to provide the trust is transparency and control to users, right? Transparency in what is done with their data and control with what they'll allow and disallow. The next thing, you know, I spoke of how privacy and advertising are not at odds with each other. I think you're familiar with the privacy sandbox, which has been our proposed approach uh, to avoid tracking individual users across the web, right?

• 43:00Speaker 02

  Uh, we are obviously in tests, we're working closely with partners and so far things are going reasonably well. The idea is to take a bunch of users who are in a cohort and treat them as one. So within that cohort, a user is anonymous, right? And say this person is a food aficionado, we'll show them food ads or whatever, right? The, the, the important thing, Alex, is to ensure we don't lose sight of what it takes to really provide privacy and trust. Because if you put up some pop-ups and say, well, you know, we will, shouldn't track you, et cetera, but then people still feel like something weird is happening, uh, then that deeply undercuts trust.

• 43:42Speaker 02

  So the example I'll give you is my wife searched for Roomba vacuum cleaners. And then for the next few days, whenever I was at home, I got ads for Roomba vacuum cleaners. Okay? And the only way conceivable way that could happen is if somebody was fingerprinting a device in the house, right? Now, that is a pretty bad practice. Uh, we do not tolerate it. We do not use fingerprinting for ad targeting or measurement. And you need to shut down those things. Uh, and that's a much deeper technical issue and ecosystem issue than some, you know,

• 44:24Speaker 02

  simpler solution like, hey, people shouldn't track, right? Uh, and we're working on solutions, but uh, that, that's what it'll take to garner user trust for privacy safe advertising.

• 44:37Alex Kantrowitz?

  Did you buy the Roomba?

• 44:39Speaker 02

  Yeah, we did buy the Roomba.

• 44:40Alex Kantrowitz?

  Yeah, I got one on Amazon Prime data this year and I will admit I'm a Roomba-holic. I just send that thing every day.

• 44:48Speaker 02

  Yeah, practically every day.

• 44:50Alex Kantrowitz?

  Yeah.

• 44:53Alex Kantrowitz?

  And the fact that Apple's making those changes and doubling its ad stuff, is that like quick get your antennas up?

• 45:01Speaker 02

  I mean, Apple should do what Apple should do. I think, you know, I go back to how do we deliver privacy-safe advertising in a way that our users will trust us? And do everything to raise their trust, including looking at some of these surreptitious attack vectors that are out there.

• 45:19Alex Kantrowitz?

  Are you satisfied with where Google is and do you think users genuinely trust Google?

• 45:25Speaker 02

  Yeah, my answer to that is we always should do better, right? You know, nothing short of a hundred percent trust is good enough. So when you say, are you satisfied? That's the bar I hold my team to.

• 45:39Alex Kantrowitz?

  And do you think people trust Google?

• 45:43Speaker 02

  I believe people trust Google for a lot of things. The quality of our information is unmatched and user studies show a lot of trust.

• 45:53Alex Kantrowitz?

  All right, we have like 60 seconds left. What are you optimistic about? What are you most optimistic about?

• 46:00Speaker 02

  I have to go back to the technology that's brewing across the community, the computer science community. And I don't want to be parochial about just Google, but a lot of it at Google. And the tremendous potential it has for uplifting human lives. Because I really think information is empowering in a way that very few things are. And the fact that even though we've been at this for close to 25 years, there is so much more we can do. And, you know, I concede, Alex, that you like the paper map better or people like map better. But there are so many new internet users out there, young people, people coming online for the first time,

• 46:43Speaker 02

  who are looking for entirely new experiences and technology can facilitate those now. I think that's tremendously exciting.

• 46:50Alex Kantrowitz?

  Yeah. And look, at the end of this conversation, I'll say, I'm willing to try out the non-paper map. And I hope it's on glasses one day. Brabaker, thank you for joining. This was super fun.

• 47:01Speaker 02

  Thanks, Alex. This was super fun for me, too.

• 47:03Alex Kantrowitz?

  Yeah, we should do it again sometime soon.

• 47:05Speaker 02

  Take care.

• 47:06Alex Kantrowitz?

  All right. See you later.

• 47:07Speaker 02

  Bye.

• 47:08Alex Kantrowitz?

  And that'll do it for us here on Big Technology Podcast. Thank you, Brabaker, for joining. Always good to have someone making the decisions inside these big tech companies, giving us some of their time, helping us think about the way that they make decisions. Thanks to all of you for listening. Thank you, Nate Quatney, for doing the editing. Thank you, LinkedIn, for having me as part of your podcast network. We will see you next week, where my interview with Francis Haugen, a person who brought forward all these revelations about Facebook in the Wall Street Journal last year, that conversation, which took place at Unfinished Live in New York City is going to go live. It was really fun. I'm excited to bring it to you on the feed. And we have another special guest coming. All right. So if this is your first time listening, please hit subscribe. If you're a longtime listener and enjoying the show, five star ratings on Apple or Spotify go a long way. So we'd love one of those as well. And we'll see you next week. So thanks

• 47:53Alex Kantrowitz?

  again for listening, and we'll see you next time on Big Technology Podcast.

• 48:05Speaker 03

  Now, more than ever, we need someone on our side. But Rob Whitman? Whitman's votes drove up costs, cut Medicaid, and risked 350,000 Virginians' health care to pay for tax breaks for himself and his donors. He's not on our side. Shannon Taylor will fight corruption, cut taxes, and bring down costs. This November, vote for Shannon Taylor. She's for us. Paid for by HMP, the House Majority PAC dot com. Not authorized by any candidate or candidate's committee.