122. Google At A Crossroads

0:11Hello and welcome to The Rest is Money with me Steph McGovern and with me Robert Peston. We've got a guest today who is someone who, it's fair to say, has been at the kind of forefront in terms of technology, business, sustainability, as the president of Google's Europe, Middle East and Africa operations. He's been doing that job for over a decade. He's been at Google for over 18 years now. So he's someone who knows, well, an awful lot, isn't he, about what is going on in the world in terms of things like AI, online safety, the future of kind of supercomputers and what they might be able to do.

0:49We've got loads to ask him basically, haven't we, Robert? I'm not sure an hour's going to be long enough. Well, we'll find out whether it is long enough, but certainly we are obsessed with this industrial revolution that we're living through, the artificial intelligence revolution, what kind of impact it will be having on our prosperity, on our mental health, on pretty much every aspect of society. he at Google is absolutely at the epicenter of these developments. But the other thing about him is he's about to leave Google after 18 years. And so this feels to us like a moment to talk to him about how companies like Google are changing all our lives.

1:38And what he's learned from 18 years, as some would argue certainly one of the most influential people in the digital space in the UK. So here it is, our interview with Matt Britton. Matt, thank you very much for joining us. Do you know, it's funny interviewing you because like where do you start? Google feels like it's part of everyone's life now. But I guess the thing I always wonder is what are your challenges at the moment? Like what are the things you're kind of worrying about at Google? What are the things that are, I suppose, potentially keeping you awake at night? Gosh, well, I mean, first, thanks for having me.

2:16I think it's been an incredible ride sort of looking at how technology has changed the world. And at Google, it always felt to me like we have an incredible responsibility. This mission is like organise the world's information, make it accessible and useful for everyone. And that's always felt very real to me. When it began, it was about search. And, you know, I started in 2007 and there were less than 20 % of people are online. And then we've had the mobile revolution. So we've got smartphones in our pockets. And now today, like 70 % of people on the planet are online. And so the moment we're in now is this huge acceleration, as you've talked on this podcast in depth about, you know, with AI coming in.

2:52And the promise of sort of putting the world's intelligence in everyone's hands is a really bold one. So we think a lot about how can we be bold to really harness the best benefits that can come from this technology, but also responsible. We've got to understand the risks and think about how we can build it in a way that is going to change people's lives for the best and challenge the misuse of technology. And that responsibility feels really real right now. We want to talk about the big social economic implications of AI more generally. But just in terms of a business transition, I mean, you've been at Google for 18 years.

3:29There have been massive changes. I mean, people like me think of Google as a young business, and compared to some, you are. But actually, you know, thinking about it now, you're quite a mature operation. I mean, more than 25 years old. One of the things that AI is challenging is your traditional search engine. I mean, I'm somebody, as is Steph, who uses generative AI a lot in our daily lives. A lot of what I'm now doing on generative AI, I would traditionally have done on your search engine, but I'm now, you know, if I'm completely honest, I'm not using Gemini on the whole. I'm using, you know, GPT-40.

4:09But you've got, you know, your own version of this. How worried are you that, just in a narrow sense, that this sort of very broad use AI programs are going to ultimately make your search engine, if not completely redundant, and at least send it into reverse in terms of use and advertising revenue and all of that? So I think there's always been huge competition for attention. And I think it's always felt to me like there's been huge competition in search. When I started, you know, we were dial-up desktop and people were finding things through Google search quite often. But over the years, you just think about the coming of apps.

4:47You think about, you know, if you're going on holiday, you've got a number of different travel apps you'll search. So we've always had sort of competition. and what we try to do is continue to invest in search and make it useful. And we've actually got AI built into search, our AI overviews, which are proving really popular. Yeah, they are useful. I mean, it pops up at the top. But I think, you know, some have said, you know, and they've got their own reasons for saying so, you know, search is going to be dead and everyone's going to go to these chatbots. And I think our view is different. You know, there are going to be different tools for different uses.

5:19We've got to work hard to make our tools as good as possible. So I could ask you just on Gemini, right? is it cannibalizing search for you? No, I think what we're seeing is that where people are seeing these AI overviews in search which is a sort of a summary of results and gives you a synthesis generally they are going to more sites as a result of it and they're searching more often it also seems to be the case that young people are finding this particularly attractive so it's early days but I think we're finding that's a good way of responsibly bringing our content into search one of the concerns we have right is that this generative AI, as you know, hallucinates, and people trust search.

5:57So we need to be very careful about not undermining that trust and making sure that people feel they can get useful results from us. And just very briefly on this, Gemini is a subscription model. Does that tell us something about the way that you're evolving? Is there any way of building your advertising model into, you call it a chatbot, the chatbot, as it were? We're at a particular moment where, to the world, this sort of generative AI revolution looks like sort of chatbots. And we haven't got that many other apps yet. And, you know, if I step right back, you guys did a great podcast earlier in the year about AI and sort of where it's going.

6:32And I think we're at a very early stage. It's a general purpose technology. And we're starting to see applications, one of which is chatbots. And you're finding it useful for lots of things, and people are. Generative AI in search is another way of doing it in a different fashion. And I think we're starting to learn the way there. But we've also got these incredible breakthroughs in other areas. So we've got something called Notebook LM. And if you haven't tried it, you should do. You can put in a load of files, ask questions about it. It can even generate a podcast, not as good as yours, but a pretty good podcast.

7:01And that's pretty terrifying. I wish you hadn't told the listeners that. Have a go, listeners. So I think we're just starting to see useful apps emerging. So don't think of generative AI as just chatbots. We're just at the beginning. No, no, sure. No, we don't. Has there been a bit of a, like, Do you think you guys have kind of rushed out too much with things like Gemini, though? Because there was obviously the controversy, wasn't there, around the images that were being generated by Gemini. So is that because you kind of suddenly thought, ah, chat GBT, we better get out there? What are we doing?

7:33It's a good question, Steph. I first sort of stepped back. I think, you know, famously, you know, we'd initiated a bunch of the research that led to the Transformers models and made this sort of new generation of generative AI possible. But I think the world was surprised and we were surprised by how users were keen to try out, you know, ChatGPT when it first launched. Given the scale we're at, we wanted to be a bit more thoughtful and responsible in the way that we brought our products forward. But I think it was pretty clear to users that we are experimenting. And, you know, when you go to a chatbot, not everything's going to be right.

8:04And where we've had issues, and you mentioned one of them, we've moved pretty quickly to fix it. But I think helping people to understand experimental technology, have a play, let's learn together. and use it. We used to always launch products sort of under beta, sort of saying it's not quite there, some things are going to break. And I think we're in that, you know, it's such a rapid period of expansion that users' experience of products is shaping how they emerge. And I'd say, you know, the pace of change now is as fast as I've ever seen it in my time working in technology. There are so many big issues that we'd like to talk about.

8:38Is this macroeconomic impact of this industrial revolution that we're living through as a result of AI. You called it a general purpose technology. I was just at a lecture actually that Reid Hoffman gave over the weekend and he said something that I think we both passionately believe, Steph and I, that this industrial revolution could be as powerful as the steam age. One of the characteristics of the steam age was despite the fact that over a period of decades economies grew as a result of that initial Industrial Revolution, for several decades, individual people got poorer. It's a period known in economic history as Engels' Poor's.

9:18And one of the things that I don't think we're debating enough is artificial intelligence massively increases productivity, no question. We can't take for granted that those productivity gains will be distributed to working people, because actually quite a lot of the history of this digital revolution over the last 25 years is that so much of the profit has accrued to the early investors, the founders, the owners of various sorts. You, like me, will agree you can't stop an industrial revolution, but how do you make sure that an industrial revolution like this works for the public good rather than simply increasing the wealth and power of a small number of people?

10:03I think it's a great question. One of the reasons I think the way you're looking at it is the right way to look at it. If you sort of step back, I would say this is a bit more like electricity in terms of a technology revolution. So we're at the stage where we can see we've got this powerful thing. We're not quite sure what it's going to be able to do. It's also the question you've posed before is how quickly will this revolution happen in terms of the way it changes our lives? So I think what we've seen in the last two years, demonstration, powerful technology, it's not quite ready for prime time, but it's starting to be productive.

10:33And what we can see is, you know, it's been estimated, I think, in the UK could be worth as much as 400 billion to the UK economy by 2030, could be worth 40 billion of savings in public sector, 20 percent efficiency for SMB. So there are some big, bold numbers out there. Then the question is, well, how's it going to change work? What are the sustainable jobs of the future? And I think it's hard to postulate. In these moments, I think it's quite easy to see the kinds of jobs where the productivity change is going to be big and the jobs could go away. You know, so, I mean, some of the early applications of AI, things like call center technology, in marketing, you can create many, many more versions very more quickly.

11:12So you can see some of those jobs changing quite rapidly. What's hardest to see is the new industries that are going to be invented and created and how quickly they're going to come. So if I take my own example, like many people in the UK, I was the first in my family to go to university. I graduated the year Berners-Lee wrote the memo that created the web. And I've spent over 20 years working in a job, in jobs that entirely depended on that invention. And so it's something that's really close to my heart, Robert. And when I took this job looking after Google in Africa, Middle East and Europe, I read a report from the EU.

11:46And the UK was part of the EU at the time. And it said a million jobs are going to go unfilled because we haven't got the digital skills. So this is the new jobs. And so we said, we've got to try and do something about that. So we we pledged to try to train a million Europeans in two years in digital skills that would help to fill those jobs. We were blown away by demand. We trained a million in the first year. What type of people are we talking to? Everyday people. Everyday people. And did they come to you or did you go and find them? So what we did was in each country we launched in partnership with others.

12:14We tried to have a programme that was tailored towards the needs of the country. So in Spain, it was youth unemployment. In Germany, it was Mittelstand businesses. In the UK, we got out and about across the country outside London, tours across the country. I think in the UK, we train now a million people in digital skills in our digital garage program. And so we were blown away by demand. It was typically average age was over 40. It was nearly 50-50 male-female. In fact, we now have all our programs at least 50-50 male-female, which is very different from sort of STEM education, as you understand.

12:43So I think what we've got to be ready for, to your question, Robert, is to be able to reskill people at speed through all life learning. I mean, one of the things that is very striking is we are seeing those sorts of programs in the UAE. We're seeing those sorts of programs in Singapore and Korea and Scandinavia. We're not seeing them in the UK. Why aren't we seeing them in the UK? So I think government, I mean, they're about to launch a white paper, I think, on AI shortly. And we've been trying, I've been having conversations with governments in my job across Europe and Africa about this. And you're right.

13:19I think, you know, government needs to do a bunch of different things to be ready for this revolution. You should think about it in two ways. I think there's a sort of AI producing economy. So the research, we've got brilliant research going on in our universities. I was in Dundee recently talking to the life sciences team there. They're using some of the AI technology to accelerate drug discovery. So we've got research and innovation. you've got the sort of data centers and the pipes that you need the infrastructure investment and then the ai using economy where public sector could really benefit strategic industries like health could really benefit but we need everyone to be able to use these technologies and one of the stats that worries me is you know the adoption of ai by smbs it's much higher in the us than it is in europe in the uk and so there's an education that's needed i think partnership with tech companies is is needed like i said i'm proud we've trained a million people in the uk there's lots of learning there i know that microsoft and others are doing similar things there's a moment for government to step in do you think the curriculum is good enough though because i'm at that point now my little girl's five and obviously just started school this year i actually watched her in a nativity yesterday so i'm not that kind of just got the emotion of that but she's doing computer lessons and i'm saying that with quotation fingers here because it is like and i know she's only five but she can use an ipad or whatever a tablet really well she she knows what she's doing But in school, they're doing really basic stuff that I probably did when I was at school, which is what a mouse is and how you use a mouse.

14:47And it's like, I mean, I haven't used a mouse for God knows how long anyway. And I'm not blaming the school at all, you know, and it's a typical state school. But like, it feels like, you know, I remember my IT lessons at school and I was at school when the Internet came out and everything else and when email started. and back then that the teachers and again i'm not blaming teachers but they weren't of the standard of which society was already at who are you blaming things i'm blaming the government i'm blaming the curriculum i think it's that needs to change like we've got kids coming out of school who who know how to use digital devices themselves because they've got them but in school it's not being harnessed in the right way i mean my kids are in their early 20s and i saw them being taught the same stuff in the same ways as my parents were taught, let alone me.

15:36And, you know, I do think that's a concern because in a world where you can get the world's information in a second in the palm of your hand, I think the skills that become important are critical thinking, collaboration, curiosity, communication. If people are going to live to some extraordinarily old age, you're probably going to have to change your career several times in the course of your life. And you've got to learn adaptability, the ability to learn. So when I was running Google UK 15 odd years ago, one of the things we were pushing was, can we get the school curriculum to include computer science and push towards STEM subjects?

16:10And the government at the time made a bit of a move and there was computing taught in schools. I think with this AI revolution, if you think about it, the parallel being electricity, we don't need to train everyone to work in power stations. We need everybody to be able to understand how to operate like a dishwasher and a typewriter and all the appliances that we enjoy today. So I think what schools need to be teaching is those skills I mentioned and, Robert, the kind of ability to retrain yourself. Adaptability, resilience, all those things. Artificial skills experience is about people actually really want to reskill in these moments.

16:41And so, you know, the people we've learned from, we've trained, you know, lots of people coming off parental leave. You know, Steph, you just mentioned that kind of moment. That's a really important moment for people to come back into the job market. People who've taken a redundancy package and gone, you know what? I love yoga. Maybe it's time for me to be a yoga instructor. And now, you know, in a few moments, I can actually build a campaign and put a business online. And I think the big thing that's changed in my time at Google is anyone with a smartphone and an idea can now market their services to 70 % of the planet.

17:11And that's one of the big opportunities for the UK. You know, we're a nation of digital shopkeepers and we're quite good at this stuff. And in this moment, you've got that huge scale. But in terms of competitive disadvantage, I just want to ask you about your own experience. we don't have enough supercomputing capacity in the UK. We don't have enough data processing capacity in the UK. Elon Musk in the past 18 months has created three, which he owns, supercomputers that are more powerful than any supercomputer in the UK. Just looking at some of the structural problems, one of the things I'm constantly told by people in your sector is if they want to build a data centre or they want to build a really powerful supercomputer, A, the planning restrictions are appalling, and B, they can't get connected to the electricity grid.

18:00So they just can't get the power. And this is one of the things I said, you know, you've got to invest in the AI innovation economy, which is the infrastructure and the research. We're building a data centre right now in the UK at Waltham Cross. It's a billion pound investment, I think, and it's taken time to get through planning for sure. And what about the energy needs of that? Yeah, so we're also like one of the world's biggest investors in renewables. So we're constantly bringing on, you know, wind power and solar power and so on, investing heavily in Europe in that. But then connecting it, because I'm told the real problem is you simply can't connect the renewables to the data center you want to power.

18:36So grid capacity is an issue. And I think, you know, there's lots been written about this moment where everyone's investing big time in chips, you know, the high end chips that allow generative AI to work. And hence Nvidia's incredible success. Yes, everybody needs to invest in data centers and in the energy infrastructure for sure. It's also a question, I think, for governments about sovereignty and strategy. So if you've got a competitive market for compute capacity with players like Amazon, Microsoft and Google investing heavily, most of the economy can use the services that we offer as long as the market's competitive to access those tools and skills.

19:12And of course, you need domestic capacity to do the things that you can do. One of the opportunities of Brexit actually is for the government in the UK to consider that data centres are part of the energy intensive use and give some advantages to people investing in data centres. What would they look like? So I think, you know, there are rules in Europe around energy intensive industries which allow some state sort of subsidy. But, you know, data centres don't come into that. But if you're not in Europe, you can advantage that. So there are some things that government can do to have a differential advantage.

19:41And are they working on that as far as you can see? We have to wait and see when the AI white paper is published, which I think is soon. But I think the bigger picture here is, you know, for governments to recognize the industrial revolution, as you call it, the sort of accelerated revolution that we're in, to think about how can that transform their economies and where are the strategic industries? But also where are the skills trainings that we just talked about that are needed so that we can be ready for the jobs of the future? And there's going to be a transformation of how we do our work in most jobs.

20:12Most jobs aren't going to go away, but we'll get more productive in how we do them. That's the stuff where everyday training is needed in the moment for people. Are we as a nation being ambitious enough? I'm struck that you'll be aware of the sort of Microsoft Stargate project, which is apparently$100 billion, including building their own nuclear power station. You know, it's great that some would say, I mean, some people with environmental concerns are going to be concerned about, you know, the data center in Waltham Cross. but it's a billion pounds versus a hundred billion dollars i mean there's a big gap there yeah so i think i mean the first thing on the on the energy side is one of the things closest to my heart is you know i've long been involved in in sustainability and we've been carbon neutral since i began at google and we're always focused on making our data centers run as close as possible on renewables all the time we're nearly 90 renewables in some of our most advanced data centers so it's not right to say that this is going to be a big sort of environmental issue But if the world doesn't push the renewables aspect of data centers as this big investment happens, then there's a challenge.

21:14The opportunity with all this investment is to push renewable energy investment at the same time and accelerate that transformation. To your core question, though, are we being ambitious enough in this country? That's what I'm asking. And I think that's what I would encourage this country's government and others to do. Be ambitious about the fields we can lead in. You know, as you think about Europe, you've got aging population. we've got incredible sort of state health sectors. Health's one of the biggest opportunities right now with AI. So we talked a bit about chatbots. The most exciting AI breakthroughs I see are things like AlphaFold3, which I was just saying our colleagues won the Nobel Prize for their work on AlphaFold3.

21:51This is a database of protein structures. Prior to their work, there were about 175 ,000 protein structures had been identified through sort of painstaking inspection work by PhD students who would take five years. for a PhD student to find a 3D structure of one protein. And AlphaFold3 was able to model the 3D structure of 200 million proteins in a matter of 24 months. So suddenly you've got an explosion of research. And what happens, I mentioned I was in Dundee and they've got a world-leading life sciences department. And I went around and talked to people working on Parkinson's and malaria, an array of diseases.

22:28And they were saying, you know, we did something in a day last week that would have taken five years previously. and already there's over 2.3 million researchers working with this database it's available free to researchers and a bit like generative ai can give you you know a first draft of a script in seconds this is taking you steps down the road towards drug discovery people are working on things like enzymes that eat plastics transformational and you say well how does the uk play into that well you know firstly we've got incredible research departments like that Secondly, we've got the NHS.

23:01And although data and the NHS is all locked in different places, it has got this incredible patient record understanding of a whole array of health issues. Yeah, we've talked a lot about that, haven't we? We don't capitalise on it. It's a huge opportunity. But how do we make sure... You don't get all the money. That Google doesn't get all the money, that the British people get the rewards. It's totally fair. So firstly, just to be clear on AlphaFold, we've given it for free to these researchers to research. on. And there's lots of opportunity for them, you know, commercialization over time. I think the opportunity in health is to ensure that people are using these tools and technologies.

23:40I was just going to say, I met with three clinicians a few months ago in Cambridge, and they were working, one was a cardiologist, one was an oncologist, and one was a geneticist. And they're all using AI in their work. And what they were telling me were there were sort of three challenges. The first is the NHS putting the data together in a way that can be useful for research. The second is that clinicians themselves are nervous, just like people are nervous in their jobs. You know, is this going to take away my job? Does it undermine my expertise? Once they started using the technology, they realized it was assistive.

24:11It's not taking away the job, it's actually giving them an assistant that's really helpful. And then the third challenge is the public feeling comfortable that these technologies are really going to help them. So there was some research that you will probably have seen that was done with some doctors at the end of last year actually using the last iteration not gpt4o but the last iteration of chat gpt and they basically gave some symptoms to a bunch of doctors some of those doctors didn't have access to chat gpt a bunch had access to chat gpt and then the third group was just chat gpt without any doctors and And one of the things that was a slightly shocking result was that the diagnostic accuracy was best for ChatGPT with no human impact.

25:01That even, you know, it was better than even the doctors being aided by ChatGPT as an advisory tool. How does that work, though? Because if the doctors were aided by ChatGPT, were they disagreeing? They were disagreeing. Right. So what happened was it's a bit like your Uber driver who says, I don't trust Google Maps or Waze. I know the best route. And it turns out the individual doesn't. We all have this instinct that we know better than the machine. And the other thing that was really interesting to me was, again, I don't know if you agree with him. But as I said, I had this lecture from Reid Hoffman.