300. Does Burnham get AI?

0:00How important is it for us to have AI sovereignty and what does that mean in reality? That's what we're talking about today. Do entrepreneurs feel less welcome in the UK now than they did a few years ago? Plus, does the government really understand that in order to create and support the next great British businesses, you need to lose some money along the way? But Matt Clifford is a British tech investor who co-founded the start-up programme Entrepreneurs First. He is also a man with a lot of influence when it comes to UK AI policy, having advised former Prime Minister Kia Starmer and one of the people to set up the first government body dedicated to AI safety.

0:41So he's a man who knows a lot about what's going on in the AI world. This episode is brought to you by Octopus Energy. Greg Jackson, the CEO, is with us now. So Greg, I've got a question for you. Is it true that by 2040 in China there will be no petrol stations because everyone will be driving electric cars? Yeah, crazy as it seems, the Chinese state oil company, the one in charge of its petrol stations, is planning on having none by 2040. If you visit China today, what's really striking is the roads are quiet. I mean, all of the two-wheeled vehicles that used to pump out pollution and noise are now electric.

1:20Pretty much all the taxis, buses. Increasingly, more than half the private cars that are sold are now electric. And even this year, so far, 25 % of their trucks are electric. So that electrification means that they're going to be less affected by global oil shocks and they won't have any petrol stations. Greg, thank you very much. Now on with today's episode. When you need to build up your team to handle the growing chaos at work, use Indeed Sponsored Jobs. It gives your job post the boost it needs to be seen and helps reach people with the right skills, certifications and more. Spend less time searching and more time actually interviewing candidates who check all your boxes.

2:01Listeners of this show will get a$75 sponsored job credit at Indeed.com slash podcast. That's Indeed.com slash podcast. Terms and conditions apply. Need a hiring hero? This is a job for Indeed Sponsored Jobs. This episode is brought to you by Facebook. So you were scrolling on Marketplace, and there it was, the bike you'd been searching for. You sent a message, and it turned out the seller was super chatty, kind of funny, and an avid cyclist. The next thing you know, you're in a cycling crew. Well, a community cycling group. The thing about Facebook, you might find more than what you're looking for.

2:38From a browse to a bike ride, this summer, find more on Facebook.

2:54Matt, really great to see you. I really wanted to start with where we are now, and in particular the dismantling of the Department for Science, Innovation and Technology. When it was first announced, I have to say, I think both of us were a little bit anxious because we were worried that this meant that the new prime minister was downplaying technology and in particular the artificial intelligence agenda. What is your take on what is happening, why they have dismantled it and whether we should be as anxious as we were initially? Yeah, I think, as you said, I was very concerned with rumors about the move started.

3:45But maybe if I start with some of the positives that have come out since and then sort of almost talk a little bit about what I think the tech community would want to see to feel that actually things were in a good place. So, you know, the good news is, I think, although, you know, I think it is a real shame to see DSIT dismantled, and I think its achievements in just three years are, I think, pretty extraordinary. You know, I think it is totally fair, as, you know, various people in the new government have pointed out, that actually, in some ways, AI in particular is being elevated now. now. So we have for the first time the AI minister attending cabinet.

4:25We have an AI task force in cabinet office jointly reporting to the prime minister and the cabinet secretary, led by a senior civil servant who was doing the senior AI role in DCIT. So in terms of the teams that were built in DCIT, which I think were very, very high caliber, both from internal Whitehall talent and external experts, they're being maintained. They're being moved to different places, but they are being maintained. I think the question that people will still have is just, what do we actually see now in terms of strategy and delivery? So I'm a little bit biased because I wrote the strategy for the previous prime minister, the AI Opportunities Action Plan.

5:08I think it's a good strategy. I think that said, I'd be the first to say it's two years since I started working on that. Lots has changed in AI. In AI years, two years is forever. And so, you know, I would like to see a recognition that actually in that time, the technology has advanced a huge amount. The politics has changed a lot. The geopolitics has changed a lot. And so if anything, I think we need to see more emphasis, more resource, more prime minister time on this topic. And really, the proof will be in the pudding whether we get that. Can I ask them, because when this announcement came out about the new AI task force, and as you rightly said, you know, they've got the AI minister, Kanishka, who we've had on the show, and chaired by Patrick Vallance, who we've also had on the show as well.

5:48And within that announcement was this statement that for the first time, AI will be positioned at the very centre of government. So what is it that they're going to do now that they weren't doing before? Because surely, like in the strategy you wrote for the previous prime minister and everything else, AI should have been at the centre of that. So what's going to change? I think one way to think about why this model could work, And, you know, probably if you were going to make the strongest case for the changes that the incoming government has made, you would say on areas of AI policy that were effectively entirely within DSIT's gift, it's performed really well.

6:29On areas of policy that are cross-cutting and require coordination across departments, it's been much more challenging. And I say that with some scar tissue having, you know, kind of been in number 10 and found coordination with, you know, there's a ton of planning implications of AI. There's a ton of energy implications of AI. You know, these are these are things where operating as a department, you know, kind of in the periphery is really hard. And where if this task force genuinely has the will of the prime minister and the cabinet secretary, then it can actually, I hope, make some of the trade offs that are very hard to make in a department.

7:01It seems to me these task forces are a brilliant innovation if they have a very clear set of goals. And in particular, if they have one goal in particular, could a task force like this be as powerful as the vaccine task force? Well, I would say there are both analogies and disanalogies, if that's a word there. The disanalogy, as you say, is there's not this one clear thing. I think you could create a one clear thing. It could be we would like to see an extra five gigawatts of domestic AI compute capacity built in this country by 2030. That would be a very different task force, I think, from the one that's planned.

7:36Where the analogies are helpful, I think, and it's huge credit to Kate Bingham and her team in the vaccine task force that they were able to make a success of this. The ability they had to cut through interdepartmental disputes, go straight to the PM and get a decision, their ability to procure things differently using different rules from the way we might normally do procurements, their ability to hire differently. These sort of capabilities, these freedoms, I think we've seen first with the Vaccine Task Force, then with the AI Security Institute in DC. I think these freedoms actually do allow you to do things differently and faster.

8:15But as you say, the thing that the government will have to decide now and be very clear on is what is the overriding goal of this task force? Because on the point about procurement, I mean, we've had various people talk to us about how difficult it is. And, you know, we're talking businesses in the AI arena who say it's really impossible to cut through the system and be taken seriously against the, you know, American businesses or these other businesses that are not putting money in the economy in the UK. So do you think that is genuinely going to change then? It's very hard to know. Again, to Robert's point, I think until it's very clear exactly what the overriding goal of the task force is, it'll be hard to say.

8:56I have seen encouraging things in the last few days, not least from the new chancellor, John Healy, talking about procurement and putting more emphasis on the broader benefits of buying British. Now, I have to say, there's a part of this that may not be a good thing. I don't think we need to buy British on every good. That's probably a way to spend too much money, get poor value for money, and actually end up in a worse place for the taxpayer. Where I think there is a case for buying British is in areas where sovereignty is actually of huge strategic importance to the UK. So actually, the truth is that if we're buying, maybe I'm going to make some enemies saying this, if we're buying stationery, I don't care whether we buy British or not.

9:45I don't think it's a huge strategic asset to have powerful British paper companies. However, I think it's very clear that we're now operating in a geopolitical environment where having valuable and powerful British AI companies is very important. We've seen that very clearly in recent weeks and months as the US administration has started to see AI much more as a national security issue, willing to export, control the most powerful AI models. And I think you start to see, well, there probably is a case, even if it's not the model layer, you know, the cords, the chat GPTs, where we would want to see, we want great British chip companies, we want great British drone companies, robotics companies, and actually using the procurement power of government to support those companies, if they're good, is a really powerful lever.

10:33I went in to see Palantir recently and was, you know, really actually impressed with the power of what they offer the British public sector. But we've already heard from Andy Burnham and people close to him that he is one of those politicians who is anxious about the extent to which Palantir is deployed through the NHS and other public services. So I suppose I'm interested in whether you think that the British government is right to be anxious about over-dependence on one American company, Palantir, when it comes to, for example, being able to process NHS data, for example, in a much more streamlined way.

11:29And if they're right to be anxious, how what is the best way to reduce UK dependence on a company like Palantir? Are there rivals out there that are British that might be in a position to offer something as effective as Palantir within the near future? If there isn't, how could a British government encourage the development of a British equivalent? So, yeah, I think this is really important because it touches on, I think, one of the big confusions in the debate about sovereignty, which is that when we say sovereignty, we often mean one of four or five different things. The sort of thing you're talking about is, do we have control over our data?

12:16You know, could we get it out if we needed to? You know, do we, is the data processed under our laws and under our norms? Actually, my view, which may not be a very popular one, is that on this level, things like Palantir are not really a threat to British sovereignty. You know, actually, Palantir holds and processes data under our laws. Under our laws, sorry. I mean, there might well be people that would like to see a British company doing that. But, you know, ultimately, I don't think there's a significant risk to the UK's ability to operate. Because actually, you know, the software that's being run is being run here.

12:53It's been run, you know, kind of locally. And we have a lot of control over that. So obviously, I understand your point about sovereignty. But even on your own set of concerns, surely it matters that one company, an American company, Palantir, has this effect of monopoly on joining out incredibly important data within the British public sector. This is a hugely valuable business. Why should it all go to an American? Why should all the profit go to an American company? Well, so I think there's two ways of thinking about this, which is, you know, is there something, supposing we decided it was really important to have a UK version of Palantir.

13:32And again, I'm not totally convinced that's where I would place my chips, but let's say that we did decide that. I worry quite a lot about solutions to technological sovereignty that are very top-down, government-led, you know, picking a winner. I think the track record of that working is very poor. For decades, various parts of Europe have tried to do this with things like search engines with basically zero success. Now, there are some people that would say, look, we got Airboss. Airboss was effectively a top-down thing. I just think there are so many differences between that kind of top-down engineering project and software.

14:14But I think what we could do, and what I do worry about a lot, and what I hope this government understands, is that the way to get, not just in what Palantir does, but in any part of the technological supply chain, the way to get British competitors is to make the UK a great place to start and scale businesses. Now, in some ways it is. As you may know, I've spent most of the last 15 years building and funding and helping people build technology companies. And the UK is a great place to start companies. But there are real risks. We've spent a long time as a tech community in this country building an ecosystem that founders, entrepreneurs want to be.

14:52I would point to two things that could put that at risk. One is that I think a lot of the rhetoric over the last few years around tax in particular, but really more of a sort of move music about our people who are building businesses, people that we want to have here. Are they people that we value, admire, want to build up? That, I think, has gone in the wrong direction. And actually, a lot of entrepreneurs, sadly, have left the country feeling that they're not welcome here. And that's a problem. There's a second problem, which is well known, and I think you've talked about on the show before, which is we do still lack scale-up capital relative to the US.

15:31And so there is still a problem that British founders get to a certain scale in the UK, raise capital in the US, and end up moving there. For me, a lot of that is downstream of our pension system. Our pensions still don't invest enough in private assets, in risk assets relative to their US peers. Successive governments have talked the talk on this, but we're yet to see a real difference. And I think if we can't make that change happen, that's the biggest reason that you don't have a UK Palantir, is that probably the UK Palantir ends up moving to the US to raise the capital they need to scale. So can I ask, because we talk to entrepreneurs a lot, and sometimes they're saying they don't feel welcome in the UK, there's not the finance in there, or the support that they need to keep them there.

16:15Do you think that's fair that they're not as welcome here as they once were? So when it comes to AI, I think there are three kinds of sovereignty that I do think we should worry about. The first is value capture. Do we have the companies, the AI companies in the UK that mean we're capturing value here, generating jobs and tax receipts here? That's key. The second is about do we control critical inputs to our economy as the economy becomes more and more driven by AI? are we protected against other countries deciding to shut off access to key inputs? If not, then we could find ourselves in a very difficult situation.

16:54And the third is, as the global governance of this technology evolves, do we have a seat at the table? Do we have a right for our voice to be heard in making critical decisions about how this technology is developed and governed? We saw very recently, over the last two months, the US administration putting export controls on the most powerful AI models built by Anthropic. And what that means is that, in theory, they said no non-US citizen can use these models. So today, maybe that's a pretty marginal thing. But you can imagine in a world where actually the economy is really dependent on powerful AI models, if a US administration decided to do that.

17:35and we were cut off overnight, that could have really huge consequences for the British economy. So there's kind of sovereignty there of like, do we control critical inputs to our economy? Maybe that sounds far-fetched today, but to me, the echoes of what happened in 2022 when suddenly, because of the invasion of Ukraine, we ended up without access to the energy inputs that we expected, there was a huge economic crisis. And then the third kind, which is a bit different, is, do we have a seat at the table? This is a technology where the governance of it is going to really matter. It's a technology where values are embedded in the technology where we might want to choose our own flavor of what kind of AI we get.

18:18And we're kind of used in the UK to having a seat at the table. We're on the UN Security Council. We're in the G7. One thing I would say repeatedly to ministers when I was working in government was, if they were creating the AI Security Council today, it's not obvious that the UK would have a right to be on it by virtue of the companies that we have. So there's another kind of sovereignty, which is how do we earn that seat at the table? Matt, there's still a lot to get into on all of this AI policy stuff, but sit tight for a couple of minutes while we go through a quick break.

18:50This episode is brought to you by Accenture. Now, when your advertising operations fall out of sync, campaigns slow down, insights get buried and opportunities get missed. That's why Spotify and Accenture are working together to reinvent the rhythm of ad sales, using automation, analytics, and smarter workflows to simplify campaign delivery and access better data across the business. The result, less time spent on operations, more time connecting brands with the moments and fandoms that matter most. To learn more, check out Accenture.com slash Spotify dash UK. This episode is brought to you by Google Chrome.

19:29You think you know a browser, but Gemini and Chrome? That's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks. Gemini and Chrome is here for it. Ready to make anything online make sense? There's no place like Chrome. Check responses set up required. Compatibility and availability varies 18+. Hey, Chicago. Class it up with Crocs. You know back to school is coming in fast. So why wait to find your new fave footwear? Step into a local Crocs store and step into your new look.

20:06Try it. Style it. Make it yours. Because the right pair doesn't just show up. It shows off. First day fits, handled. Walk out ready for whatever's next. Visit your nearest Crocs store today.

20:26Can I ask you, Matt, a bit about the kind of infrastructure side around AI? Because that's part of the problem is the speed at which we're not doing things in terms of data centres. And obviously that's a big deal up here in the North East because of what's happening potentially in Blythe and around Teesside. What does it take to get things built? So first, I would just say, because I think a lot of people say, like, why do we even need this? You know, like, why does it matter whether we have data centres in the UK? Can't we just, you know, it's all electrons traveling very, very, very fast. Can't we just use data centers in the US?

20:59And this, again, is a sovereignty question. You know, there's a lot of people who said before the invasion of Ukraine, oh, you know, energy is just a commodity. You buy it on the open market. Who cares? And yet we saw in the UK in 2022 the cost of not having domestic production, of not having energy sovereignty in the UK. my worry is that we could enter an era in the UK where the same is true of AI. And so I think it's very important that we have AI infrastructure in our own borders, that we control, that operates under our laws. Even if it's international companies that are building it and own it, I think that's really important.

21:41As you've pointed out there, Steph, it's not trivial to get these things built. In fact, it's really, really hard. Now, there's a number of issues. One is planning. Obviously, these are huge construction projects. But the biggest one by far is energy. And there are at least two distinct problems in energy when it comes to building data centers in the UK. One is just cost. As you know, we have the highest wholesale electricity prices in Europe. if you were to compare running a frontier data center in the UK versus, say, in France, it would cost you£400 million a year more in electricity costs alone in the UK to run it here.

22:22So that's significant even at the scale that we're talking about. And then the other energy issue is actually just our grid. So getting a grid connection takes a very, very long time in this country. There is a huge queue to be added to the grid. I remember when I was writing the AI action plan for the government, being pretty shocked when people who wanted to invest billions or even tens of billions of pounds in building AI infrastructure in the UK were telling me that they were being given dates for a grid connection in 2039 onwards, which in AI years is forever. Oh, my God. And so, you know, it's a real challenge is in an industry that right now is extremely geared on speed, speed to power.

23:07How quickly can you get a day at centre online? At the moment, we're not in a position to be competitive on that. And so really the sites where it's viable in the UK are those that have existing grid connections ready to go. But that's when you get into, you know, understandably political wrangling on what the right use of that grid connection is. So then how do you get past that? This is where you need a lot of careful diplomacy and deal making. I mean, my view, my personal view would be the amount of capital that is at stake in these deals is so large that there are absolutely ways to create win-wins.

23:45The size of the pie is very large. And so, you know, what I would like to see is for UK national and local government to explore ways to actually share that pie with stakeholders locally. So you can imagine a world where we say, yes, we will accelerate a grid connection on this site because we think it's strategically important for the UK. If, for example, the people building it, the beneficiaries of it, put aside some share of the value of the project for local investment, for creating jobs and training and investment locally. I think the sizes of these projects are big enough that it's very easy to do that in theory in a way that's a win-win.

24:28And so I think we need to be a bit more savvy about the politics of this and say, how do we actually make building this infrastructure a win for all the different parties who are involved? You chair ARIA, which is this huge fund, which provides funding for very, very early stage research into a whole range of really fascinating sort of technological and scientific areas. but most of which could well end up with a sort of serious commercial application. I suppose there are sort of a number of issues here. One of them is that as somebody who has been immersed in essentially the pressures that are on government for decades, you know, as a journalist.

25:29You know, if I were in your position, I would be permanently terrified that at some point, some chancellor is going to look at what are really significant hundreds of millions of pounds going to ARIA and just think, wow, they can't really prove that there are sort of short-term returns for Britain from this. If I were in your position, I'd be terrified that at any particular moment, some chancellor looking to save some money and allocate it to a short-term populist gimmick would just close you down. Do you lose sleep at night about this? I'm so chronically tired, Robert, with two small boys that I never lose sleep at night over anything.

26:17I mean, look, you're absolutely right. What a remarkable thing Parliament did in 2022 in agreeing in a cross-party way, you know, this was voted, every major party voted in favour of ARIA, to create something that very explicitly would not produce short-term returns. And not only did Parliament know that at the time, it was actually baked into the legislation. One of the things it says in the ARIA Act is that ARIA can't be shot down for 10 years. Obviously, they could repeal that, but there was an understanding, a cross-party understanding, that yes, actually, if you want to see the huge long-term impacts of fundamental breakthroughs in science and technology, then patience is what's needed.