Liz Reid, Vice President, Search, Google Interviewer: Jeremy Kahn, Fortune
Category
🤖
TechTranscript
00:00All right, Liz, thanks for being here with us.
00:02Thank you for having me.
00:03Great.
00:03It's a great room.
00:04Yeah, fantastic.
00:05I can't see you at the back, but otherwise great.
00:08I want to warn everyone, I'm going
00:08to go to questions pretty early in this session,
00:10so please think of your questions.
00:12But first, I want to ask you, Liz,
00:14there's been a lot of reporting out there in the last few weeks
00:17that younger people, in particular,
00:19are starting to Google things less.
00:20They're not going to traditional Google search.
00:23They're using some kind of AI chatbot,
00:25or they're using these new AI-enabled search engines,
00:27things like Perplexity.
00:30How are you viewing this trend at Google,
00:31and how do you want to try to combat that potential loss
00:34of market share?
00:36Yeah, so there's always been different options with search,
00:39and we welcome that.
00:40I think what we're really focused on with search
00:43is the user, and really helping that user find information.
00:47The truth is that there are not just billions of questions
00:49that people come to Google for.
00:51There's billions of questions people don't even ask,
00:53and they don't ask because it's too hard.
00:54They don't ask because they're not confident
00:56they can get that answer.
00:57And so we really think about, in search,
01:00how do we make it so that you can really ask
01:01anything you're curious about,
01:03and that it really feels effortless,
01:04that it's so easy to do that you don't think twice
01:06before you do that.
01:07And we think there's a huge opportunity ahead on that.
01:09Interesting.
01:10You mentioned people coming to search
01:13because they need answers,
01:13and they want quality information,
01:15and there's a lot of concern right now
01:17about what's happening to the information ecosystem
01:19on the web.
01:20There was already a problem even before generative AI
01:23exploded about search engine optimization,
01:26and whether that was degrading the quality of information
01:28that was available on the internet.
01:30Then there's a doubling of that concern with generative AI,
01:34and the fear that's producing a lot of AI,
01:36quote-unquote, slop that's out there.
01:38And how are you seeking to address this problem at Google?
01:43Is that a concern now that you're,
01:45okay, we have a great search tool,
01:47but it's gotta be able to surface information
01:48that's of value to people?
01:50So ensuring that we surface high-quality information
01:53is definitely paramount,
01:54and it's a huge reason that people have trusted search.
01:58This challenge has existed
01:59since the beginning of search, right?
02:02People have constantly created amazing content for the web,
02:05and they've created poor-quality content.
02:07In many ways, it's a cat-and-mouse game
02:09with those who wish to create spam across.
02:12I would say AI-generated content
02:14isn't by itself intrinsically bad, right?
02:16You can use AI to really help people write,
02:19and create, and produce amazing information.
02:22But there's also this challenge of scaled content abuse,
02:25and this is something we take very seriously.
02:27It's something we've continued to work on for many years,
02:29such that the level of spam in search is very low.
02:32And we'll stay focused on it,
02:33because it's absolutely paramount.
02:35Great.
02:37You've now rolled out these AI interviews,
02:39sorry, AI overviews pretty widely.
02:42That's created a lot of concern on publishers,
02:44the part of people and other businesses
02:45that really depend on search traffic coming from Google
02:49to their own websites,
02:50where they seek to monetize that,
02:51often through advertising.
02:53What are you saying to publishers
02:55about where this technology is heading?
02:57Because there's a lot of concern in the industry
02:59about where the effect of this is gonna be.
03:02Yeah, so I think we have a long history in search
03:05and having a value exchange with the ecosystem,
03:07and we've sent billions of clicks for a long time.
03:10But one of the things
03:11that there's sort of this common misconception
03:13is that people either want answers or they want links.
03:16That's not actually what we find from research.
03:18People wanna get started,
03:20but they actually really wanna hear from other people.
03:22They're interested in their perspectives.
03:24They're interested in their understanding of the space.
03:27And so we've really tried to build AI overviews
03:29and continue to iterate them
03:31on a way that enables people to get started,
03:33but then continue to connect to the web.
03:35And we see that that's one of the things people love
03:37about AI overviews,
03:39is this ability to have that hybrid,
03:40to understand and then to continue on in your journey.
03:44And I think that's only gonna continue.
03:46We see that people, especially younger users,
03:48really wanna hear from the authentic voices.
03:51They wanna hear from the people they trust.
03:53And that's part of what makes search special,
03:55connecting to the web.
03:56And so that's at the heart of how we think about
03:58in the AI transformation,
03:59still enabling that mission of connecting people
04:02to the best of the web,
04:03and how can actually AI make that even better
04:05in allowing people to connect to truly the best of the web.
04:08Are people still clicking links with AI overviews?
04:10We do see people click both within the AI overviews
04:14as well as on the rest of the page.
04:15And in fact, when we see AI overviews,
04:18we see that the clicks that go out to the web
04:21are of higher quality.
04:22People spend more time on the sites they go
04:24because these are really giving the information
04:26that they want.
04:27Interesting.
04:28All right, questions from the audience.
04:30Who's got a question?
04:31Right here, this gentleman has a question.
04:32That hand went up very quickly.
04:35Let's wait till the mic gets to you.
04:36If you can let us know who you are.
04:38Yeah, Josh Legassic with Insignium.
04:41I didn't even know that was a thing.
04:43It just started popping up and I love using.
04:45I'm glad to hear that.
04:46That's great to hear.
04:47Agree with the clicking back and forth.
04:49My question is, is it learning from me
04:51and feeding back answers to me?
04:54Or is it learning from everyone
04:55and feeding back answers to me?
04:56Or is there something else that's happening?
04:59So I think there's different levels of your question.
05:02There's sort of, what is the level of contextualization
05:04that we think about in these spaces?
05:06And there's some understanding of like,
05:08how do we interpret the query?
05:10That can be more contextual.
05:12But in general, we use throughout search,
05:14not just in AI overviews, but broadly.
05:16We try and learn from information in a very privacy,
05:19safe way, giving people controls.
05:21But we can learn a lot.
05:23This is how people learn.
05:24These are the best sites to go to, right?
05:26Is often based on user feedback
05:28in addition to our high quality signals.
05:30So we try to do it in a way that's very clear,
05:32that gives people agency,
05:33but that we can actually improve
05:35based on the usage of the product.
05:37All right, other questions from the audience?
05:40Oh, how about this woman back here?
05:42There we go.
05:43Wait for the panel.
05:45Hi, I'm Julia Martinson with ServiceNow.
05:48I mean, Google search has obviously been
05:50the best for a long time in terms of search
05:53and also providing the right results,
05:56you know, ads kind of maybe, you know,
05:57for some more cumbersome, for some more enjoyable.
06:01I'm wondering with seeing how LLMs went also
06:04into more domain specific, small language models.
06:07I wonder if maybe, you know, a roadmap question,
06:10would it be ever, could we ever see
06:12a domain specific Google search basically,
06:14where you're also kind of going down that level
06:17and making it more relevant
06:19for that specific domain you're in?
06:21I think we've continued to look at different domains
06:25in different ways.
06:26I mean, if you come to search today,
06:27we have a mode that enables you to do images, right?
06:30If you really wanna go deep in images,
06:32we have the ability to go deeper on shopping
06:35or on local or in others.
06:36And I think that will continue.
06:39There'll be cases where what really people want
06:41is a specialized understanding of information.
06:45I also think with AI,
06:46the ability to just personalize the responses
06:51evolves in ways that we don't traditionally think.
06:53A lot of people previously with personalization,
06:55with ranking, think of it as like,
06:56okay, I'm gonna re-rank just the sites, okay?
06:59But actually there's a lot of different ways
07:02in which personalization can be useful, okay?
07:05It starts with basics like some people love video results
07:08and some people really do not, okay?
07:10Can you understand that?
07:12What's the level of understanding
07:13of which you want a response is?
07:15Are you an expert in that domain and we should assume so?
07:18Or are you sort of trying to help your five-year-old
07:21with a concept that you don't actually know very much
07:23and explain it to her and how can you do that?
07:25Are you on the middle of a long journey
07:27or on a shorter journey?
07:29Do you like to scan information quickly?
07:31Are you an audio learner?
07:32And then we should make it easier
07:34to understand information in that form.
07:36And so I think besides getting better within domains,
07:40there's just a huge opportunity to start to understand
07:43what does it really mean to make information effortless?
07:46If you go back to the mission,
07:47most people, if they repeat the Google mission,
07:50kind of stop halfway through,
07:51they go organize the world's information
07:53and that's kind of where they stop.
07:54But there's this part at the end
07:55about universally accessible and useful.
07:58If we show you information that's organized
08:00but it is not useful to you,
08:02it's not really that useful, right, overall.
08:04And so I think there's an opportunity with AI
08:07to rethink what it means to give information
08:09in a way that's truly useful to you,
08:11that truly feels effortless to consume and interact with.
08:14There's a lot of people really like the AI overviews
08:16and a lot of businesses use Google Search
08:19either on their websites
08:20to allow customers to search for things
08:21or they use it internally on their own information.
08:23Is there the ability to kind of take
08:25the AI overview kind of technology
08:27and have that work on your own site
08:29or on your own information?
08:31Yeah, so there's a bunch of different offerings
08:33that Google Cloud has.
08:36And I will confess that I'm not the expert in Google Cloud
08:38so you may want to get additional information on this.
08:42But at a high level, we both have services
08:44that allow you to have your own data
08:47basically ground a response in the data on your site.
08:50There's also a service that allows grounding with Search
08:53so that you can get the best of sort of Search's understanding
08:55of what's high quality information in your product.
08:57And so Cloud offers a variety of different surfaces
09:00so that we can actually take some of the key underlying LLMs
09:04and understanding about how to get high quality information
09:07and make it available for others.
09:08I know there was at least one more audience question.
09:10This gentleman right there, let's get him in
09:13and then maybe I'll have time for one more of my own.
09:16Hi, my name is Matt Sandler.
09:17I'm involved in running Gen AI at McDonald's.
09:20Curious, Google's famously brought Search together
09:22in like one bar that you can do everything in.
09:25And yet now you're sort of splintering it again.
09:27You've got your Gemini product, you've got your Gemini app.
09:29Where do you see this going?
09:30Do you see it sort of coming back to one interface
09:32or do you see sort of the chat interface
09:34and the Search interface?
09:36That's a great question.
09:36I swear I didn't set you up for that
09:38because that was the other question I was gonna ask
09:39is are the products gonna blur together?
09:41You have Gemini as a separate product from Search.
09:44Are those gonna be kind of recombined?
09:45I don't think we honestly know what the future has.
09:47We're pretty early in the journey.
09:49I do think that one of the things I think about a lot
09:52is how do you meet users where they are?
09:54And so if they come to Search,
09:56can we give them the information in the form they want?
09:58If they choose to use Gemini, can they do that?
10:01If you actually go back in like long enough ago,
10:04people ask this question about YouTube
10:05and Google's video search.
10:06They ask this question about Google Maps
10:08and local search on Search.
10:10Like, wait, where should I search for a restaurant?
10:12On Search or Maps?
10:13Should we just merge Search and Maps together?
10:15Really what you wanna do is make it easy,
10:17like whatever is easiest for the user to come forth.
10:20The other thing that I would add in that space
10:23is that I think you've mentioned the keyword box, okay?
10:28Actually, Search is evolving
10:30since the keyword box is only a component, right?
10:32We introduce voice, we introduce,
10:34hopefully all of you know what Lens is.
10:36If not, you should come find me afterwards.
10:38That ability to search for what you see.
10:40And I think the space of information
10:42is going much more multimodality in its input.
10:45And the reason is because we're humans
10:47that existed a long, long time before technology.
10:50And the way that we understand the world
10:51and we think about the world
10:53is based on a lot of different senses, right?
10:55It's based on what we see and what we hear
10:56and how we talk across.
10:58And so if you wanna make information effortless,
11:00we're gonna continue in Search to think about
11:02what is the most natural way you would express?
11:04If you had no constraints,
11:05if you assumed tech was not a barrier
11:07and you just asked your friend as easy as possible,
11:09how can you have that experience with Search?
11:11So the tech really comes to you
11:13instead of you having to translate your need to the tech.
11:17Liz, you've got this, Google has, as everyone knows,
11:19this Department of Justice case that's out there.
11:22And there's just been this remedy phase.
11:23And some of the suggested remedies
11:25involve the separation of Search from other products.
11:28Are you guys actually sort of planning internally
11:30for if you had to implement one of those remedies,
11:32what that would look like?
11:33So I don't have more, really,
11:35that I can comment on regards to this.
11:36What I will say about Search
11:38is that we have consistently found that users choose Search
11:40because they really find it the most helpful.
11:43And I think we're gonna continue to really focus
11:45from my point of view on how do we help users.
11:48Got it.
11:48There's a lot of talk about sort of agentic AI
11:51and people maybe using other agents.
11:53Increasingly, you might have people coming to Google Search
11:55that are not actually human.
11:56You have an agent coming.
11:57How are you thinking about that,
11:58the idea that the user may not actually be a person?
12:01I think, in general, our goal at the end of the day
12:04is to help the user as people interface
12:08or as agents interface with Search in different ways.
12:12I think we will always try and evolve it
12:13to the question about like, okay,
12:15how do you help the agent such the agent helps the user?
12:18But at the end of the day,
12:19we're really focused on helping people
12:21and we'll continue to innovate in that space.
12:23Right.
12:24And can you give us any sense of,
12:25Robap, are we gonna see more multimodality?
12:28What are we gonna see from Search going forward?
12:30I think you will definitely see more multimodality.
12:32I think, in general, you will continue to see this evolution
12:36such that it feels like accessing information
12:38feels just much easier.
12:40It feels more effortless.
12:41It feels like the question
12:42that you actually have in your mind
12:45is the question you get to ask.
12:46You don't have to break it down.
12:47You don't have to translate it.
12:48You don't have to worry about what language you speak.
12:50You don't have to worry about how you learn best.
12:52And that Search just becomes much more accessible
12:54in a new and exciting way.
12:55Great, fantastic.
12:56That's all the time we have for today,
12:58but thank you to Liz and thank you all for listening.