Public Trust and Future Productivity

The race for AI supremacy has two contenders The US and China Where does that leave everyone else Specifically, where does it leave a middle power like the UK, which isn't an AI superpower, but does have big AI ambitions Can the UK really be an AI maker and not just an AI user. This is the Economics show with Samaya Kaines I'm joined today by the UK's Minister for AI and Online Safety, Kanishka Naran Can I go hello Gsreat to be out here Great. o, so on this show, we always start with a silly question. And so I'm going to ask for your candid assessment of where the UK sits in the global AI race, right? And so this very serious scale. ten is we are totally on track to catch up with China and the US. And one is we have barely left the parking lot Right. I think I think I'm going to do the classic political thing of saying we're on seven and a half. We're doing well. There's a lot of space to get to ten out of ten but I feel very proud of where we are as a ting point Okay, wait, but ten out of ten, you think we could be on track to catch up with the US and China on AI? I think it depends on sort of what the objective measure here is. If it is, can we be masters of our destiny and shaping parts of AI that are critical in light of our values? I feel like there's a partath to doing that. I don't think we're there today I think that last ten to fifteen years the president has not positioned us in the best place for it, but the fundamental ingredients are there for us to make sure that over the next five to ten years, we're in a radically better place, getting close to ten out of ten, if not there Okay, great. well, I want to get back to that. but first I suppose I should address the elephant in the room, right? We are speaking after Kistama announced his resignation. We're going to get a new Prime Minister this summer, probably Andy Burnham Obviously, we're facing some uncertainty right now. and presumably that includes the UK goovernment's focus on AI, right? Do you think that this is worrying investors? and what's your message to them I don't think's worrying investors because what investors have seen, I think, especially in my patch on AI is a very consistent focus right across this government and a pretty enduring set of underlying values in the Labour Party that have driven it. We are doing everything to make sure that prosperity from AI, both in terms of the compute this country needs, manufacturing, in terms of chips, applications, so that we can cut waiting times in the NHS and give professionals a lot more time to do things they care about at work and beyond. All of those things are enduring priority areas for the Labour Party they are for the UK government And then the second thing to say is that the business of government, the tricky bits and the positive bits are all continuing promptly as well. So I feel confident that whatever comes next in terms of leadership. I think the underlying enduring values and the demands of the British public will continue to be the focus of us. Before we go on, I want to talk definitions, right? Because when people talk about AI, they talk about the AI stack, but often the discussion can flip between different parts of it. And so I just want to clarify first. So the stack refers to these multiple layers involved in making AI or the AI supply chain And at the bottom of the stack, you have the physical infrastructure, underpinning AI, the chips, filling the data centers, all powered by huge amounts of energy. Then nexte in the stack you've got the development and the operation of the AI models And then at the top of the stack, you have the applications, right? So these are the businesses transforming AI into products that are useful for businesses and consumers. And then you need institutions, right? So security measures, some sort of governance.? So that's the stack Qion is What are the priorities across that stack? Or is it the idea that actually the UK will excel in every single bit of it We think we've got strengths that allow us to be pretty surgical in different parts of the stack. But all we're trying to do is to be very precise in which parts of each of those layers the UK has strengths in where we can build both economic upside but also a sense of strategic leverage internationally on the sort of fundamental underlying level of the data center, but also the chips that are a big part of what goes into data centers So far, a lot of the focus on chips has been on training the large language models and NvIidia GPU's have been the main game in town We think that the next generation of focus will be on inference in application, not just training the models. And in that, the chips you need for those sort of specialized applications are very different. We think that's going to be an area of strength for Britain. We want the best firms to be built here We want those jobs, which are actually very jobs intensive kind of areas, manufacturing capacities. We want those to buil in Britain In terms of models, we don't have a frontier language model company in this country headquartered here. Obviously, Demind has been a remarkable presence, but not the parent entity incorporated here. But we think there are lots of non language areas of intelligence computer vision materials discovery, drug discovery. even new generations of next paradigms of language where we think Britain can bet on where the future is going, not where the past has been. And so at the model layer, we think we've got some strengths. And then in terms of applications, we want to be very targeted at where we think Alication strengths can move the needle in terms of productivity for us. We think those are professional and financial services. We think cybersecurity is really critical in that context. We think drug discovery and life sciences are important. And we think ultimately that Britain has always been not just the home of innovation the home of responsible innovation. And so trust, safety and assurance, some of the governance aspects you talked about, I think are important areas of strength. So some parts of each of those layers, but in a much more specific way in light to where our strengths have been and where the future is going I want to go backwards through that stack, right? So starting at that top layer of the AI applications. So I think there's a sense that actually this is where the UK is particularly good, right? We have lots of very strong companies. I think there's a question about whether you know AI can be combined with areas of research we're already excelling you know, so quantum defense, healthcare, Is there one specific one where you feel like actually you know we've really got the chance to be world leading So in terms of sort of I guess fundamental model companies, the areas where we have particular strengths are Comuter vision, in the first instance focus on self driving cars. So Wave, for example, Oxer being another the long term vision for those companies is to start with computer vision applied to driving and in the long term move to industrial robotics. And so that's going to be a huge area of strength for us. Another area of strength for us is going to be drug discovery, the feuture of life sciences and academia presence of very unique data sets and a history of pretty deep integration between fundamental drug discovery work and its clinical application All of those position Britain very well. And so companies like Isomorphic Labs, for example, run by Denis Aabz and Max coming out of Deep mind has been a company that we have backed in government. I think those companies are going to be a big part of the future as well And then the third area that I'd highlight is a very novel area which is companies looking at tackling things like fundamental climate change challenges in terms of materials that are more sustainable. processes in materials discovery and materials engagement that are quicker, cheaper, more efficient, and also more sustainable. And so we have amazing firms there bringing together some of our best academic strengths in chemistry. with AI to discover new materials applied to whether that's aerospace or construction, other areas where there are significant emissions challenges. So I think those three computer vision applied to self driving in the first instance of materials discovery applied to tackling climate change challenges and then drug discovery tackling some of the core healthcare and life sciences challenges. I think those are the most important application areas that Britain has strengthed. Yeah, so I suppose there is this narrative that the UK is particularly good at this application layer when it comes to AI. But there's also a concern, right about sovereignty. because You know, if what we're doing is using the AI and applying it well, that potentially leaves us vulnerable to another actor whose name rhymes with America shutting off our access to the AI stack, you know the AI supply chain overall, right? And so there was obviously this event a few weeks ago that freaked everyone out when the US restricted access to anthropics frontier models, Myithils and Fabablet to non US citizens How do you think about the kind of vulnerability that the UK faces there Well I the diagnosis has to be that we are not in a terrific position from a strategic leverage point of view. The frontier models are developed either in the United States or slightly behind them by a few months in China And so Britain does not have companies when it comes to frontier language models And we don't have the core chip architecture, the Nvidia chips as well as the memory and packaging. So SK HinX in South Korea, for example, gives them an advantage. TSMC in Taiwan gives them an advantage. AML in the Netherlands giv them an advantage. And so broadly speaking, we have some strengths in that supply chain, but none of those you would point to ArM does a lot of the chip design, but none of those you'd really point to and say Britain is currently competing sufficiently with the United States and China so that we can have continued access to critical inputs. And so the nature of the problem is very stark, and I think it's a serious question for the future of our economic and national security. And so what do you do about it? And there are two ways I think about this Broadly what you're trying to do is have with a pun, chips on the table so that you can access others as critical inputs. And you can do that either by having chips on the table in the AI stack Or you can also think about it as a broader international negotiation bundle. We might not just offer AI things, we might offer other things that are important to say the United States or to China to be able to secure access. On AI, what we're trying to do is effectively say, we think it's really critical, we build more leverage. And that's why we are focused on things that give us leverage We think our chip companies will, if they succeed, be in a position where it's very hard to replace them in some cases, and that gives you leverage. We think some of the frontier models that we're developing outside of language, in materials discovery, in new drugs discovery, in world models and vision models We think those will give us leverage. But those are not overnight things and they take a little bit of time And the third thing is we have built pretty remarkable state capacity. We have built the AI Security Institute, which is the only lab in the world that gets pre deployment access to all the core models. And so by understanding the risks, sharing it in a trusted way with core allies across the five ey'es We think we're able to build trust that also gives us leverage in the AI stack as well. And so my focus across the AI stack both on hardware on models and at the heart of it on governance, assurance on safety on state capacity is to try and build things that serve British interests domestically as well as internationally Can I just go back to that Mythos Fable event, right? I mean, were you surprised by that? and did you try to do anything to get our access back Well, I think I think the honest answer is of course, the abruptness of it caught us by surprise. But in terms of what we've done since then, we've continued to engage with the US government to make sure that we are supporting them as much as possible in building an understanding of that model in particular And then alongside that, we're continuing to make sure that where access is available to frontier models, the UK government is well placed. But the big thing I would point out is that by being able to understand the risks associated with those models early We were able to make sure that for core national infrastructure National cybersecurity Center understood those risks ahead of pretty much most countries in the world and were able to mitigate some of them. And so I feel like Britain has been in a very differentiated position. Of course, we want to long term be in a position where we have even greater strategic leverage and can maintain access to critical inputs that we need And thinking about this sovereignty issue, to what extent do you think the UK should be banding with other middle powers, right? I mean, the UK is not the only country in this position of, you know, having AI weaponized against it potentially Should we be working with, say the EU, Canada, Japan, South Korea, even if it risks US displeasure? I wouldn't place the sort of trade offs between these choices of global partnership as darkly as that That's because the fundamental reality of where we are is that The capability of frontier closed models is so significant compared to lagging models And those models are entirely located in companies in the United States and to an extent China And so if you start with that empirical reality and the history of American UK relationships I think it is inarguable that you want to make sure that you have a close partnership with the United States to make sure that those models are available to core users in the UK, but also that British capability in assuring them and assuring their safety is shared with the United States as well. So I think that's a pretty fundamental feature of just the empirical reality of where the capability sits today and it's likely to sit even on the most optimistic horizon for a number of months ahead alongside that. I think there are clear areas in which we can partner with our allies across what you call the middle powerower Alliance and I've had conversations. I was in Australia a few weeks ago. I've had conversations with lots of European partners. The Indians hosted the II summit earlier in the year as well and their series of conversations particularly focused on adoption where we have shared interests, particularly focused on aspects of regulation when it comes to children's safety in particular where we have shared interests. And then there are some shared interests also in building future leverage collectively as well. And so the way I describe it is sort of certainly in terms of time horizon, I think in the here and now We have a great starting position with the United States. We want to make sure that we continue to deepen that and then over a slightly longer horizon we're trying to do a lot more. W Medal powers on both mitigating the risks and capturing the opportunities Can you be any more specific about where you hope to gain that leverage in conjunction with those other middle powers know when you think about regulation, for example, recently, I was in Australia, as I mentioned, they banned social media for the sixteen'. They're looking at aspects of how you sort of support children's safety when it comes to dangerous AI chatpot engagement When you're able to move a bit more collectively on those questions when you have relatively limited market sizes, you can move the needle much more in terms of product design and where the companies are as a result of doing so together. that doesn't give you leverage when it comes to a you know, hostile geopolitical actor, right? That's that's kind of coordination on safety But it's not saying, oh, you know, if you threaten to cut us off, we'll do this So I think leverage is functionally used on two fronts. One, it's used to maintain access to inputs for a longer period of time, but two, it's used to meet your public policy objectives more widely as well. And one of those is safety as well. And so if you don't have leverage by the context of your weight or your companies you need to create it so that you can make sure that the technologies being used in this country are used for purposes that are appropriate for the British public. And so we've exercised leverage collectively to keep the British public safe. You're right, that that in itself doesn't give you leverage in terms of making sure you have access to inputs for the future. And so on that front, one of the things that we've tried to do is make sure that we're continuing to speak with European partners on thinking about financing of companies coming out of different countries in Europe I think there's a lot more for us to do. so many of our venture funds cut across the UK and other parts of Europe. and we want to make sure that as we think about scaling capital in particular, that we are supporting AI, anies in the UK and Europe with the weight of broader capital markets provision across the UK and the EU as well. And so that's been one clear area I think the second area that we're starting to look at is to say, hey, we think there are partners. I had conversations with the Dutch. They have great strengths with ASML, which provides some very important machinery that goes into advanced packaging for chips as well How can we partner together on parts of the chip supply chain so that in the next generation of chips in particular for application and inference, that we can build supply chain in our shared countries, which collectively give us leverage as well Okay, interesting. I do want to get to chips in a minute, but I think reading between the lines, British financing, joining with the Europeans is the idea to create a kind of European competitor to the large language model companies, European or British open AI. We're not at the moment pursuing an active project to create a frontier LLM with European partners. One, I think the companies In that mix mistal, for example, Chere is another, but also companies creating open source and open weight models like reflection. And we have some companies trying to do that as well as a lab that we launched yesterday. And so we're making bets across different sort of model types None of them are sort of really frontier capable models at the moment. I think the level of capital expenditure required not just upfront, but on an ongoing basis is so material that I think that is not a project that we're pursuing at the moment. And the second thing to say is that that isn't the only way in which I think we can secure leverage in the stack And in fact, there is a pass here where you get more and more adoption benefits from smaller niche application specific models. And I think that's another area where we can do quite a lot. So just from a kind of capital efficiency point of view, that is not an area that we're currently actively pursuing Okay, let's talk about chips. So earlier, you were sounding pretty optimistic about the UK's capacity to be a leader in the chIips world And really you're saying that we can catch up. So how do we do that? What needs to be done to really know fulfill that ambition to be a Chips leader Chips are very capital intensive. So the first thing you do need to make sure is that you have sufficient capital supply in the country daily from sources that are relatively close to home if not at home so that those companies continue to access capital here and don't feel the need to move elsewhere just to access capital. The second part is that you also need sufficient skills, chips design and electronical engineering is a very specific area. It is an area that we have strengths in but the supply of that skill has not been sufficiently Elastic it hasn't moved as much as it ought to have given the demand in the world and from firms. And so we want to make sure that we're accelerating how quickly young people in this country feel excited about that, get into those educational parts and then as a result are able to build careers. So skills alongside capital is an important thing Demand is really important becausecause we can build the best chips in the world for AI But if all the demand for it is from labs and model companies in the United States thenen ultimately there is a bit of a pull to be near your customer. And so what can we do in the UK and nearby? to make sure that customers, both frontier labs, private customers, but importantly, the public sector is also a major player in this. How do we make sure that we're anchoring more and more demand and customers here so that companies don't feel like they have to only sell in one area and therefore be in that area. Of course we want companies to be export oriented, but there is just a sort of fundamental table states level of domestic demand that gets them to a decent enough scale early on to stay here. So those are three important areas we can talk about any of those That last one sounds like a trade restriction, right? So you're effectively saying, okay, we're going to give preferential treatment to British made Are the governments happy about that? Two things to say on this. One, it's sort of very early stage innovation focused support. And so what we announced was an advanced market commitment for novel chips right And so this is not us saying we're not going to buy mature Nvidia GPUs, we are buying lots of Nvidia GPU's. What this is about saying is that we think there's a next generation of innovation. We want to pull that innovation forward and anchor it in the UK. So we have set technical risk milestones as you hit those, you unlock incremental government revenue and you can go and raise five to ten X that in the equity markets. And so generally speaking, what we're trying to do in Britain is, I think structurally a lot more sort of using the market. to do industrial policy intelligently rather than sort of wholesale, put up the barriers and stop trade happening M Are there any other tools where actually the UK is having to catch up with other countries just being much more aggressive with their industrial policy? Caniby, I think the one area that we hear is the scale of manufacturing subsidy just sort of recred grant support directly from the state. The fiscal space in this country is in a very different place to say countries which benefit from a pretty happy degree of dominance of their currency. And so I think the sort of space to be able to do those sorts of direct grants has been different And so we have to be much more intelligent about how the state is partnering with capital markets to be able to support those companies rather than just directly giving gant money Okay, so cash is a constraint and will continue to be a constraint, presumably given the UK's fiscal position. Can I ask about other constraints though, to the UK's AI buildout? becausecause there was the recent case of open AI pusing a massive investment into the UK, citing energy costs. More needs to happen so that we're attracting those kinds of investments Yeah, I think I've been on record saying previously that that announcement came alongside a pause of opening eyes Norwegian announcement, as well as a pause in a major American announcement they made and happened to also coincide with their IPO planning, which they had to sort of pretty nervously fix their financing for. People can read into the motivations on that particular announcement, what is available in the public domain. But let me talk about what the build out requires. We need to build out at least sort of table stakes level of compute capacity in this country to me it's really core sovereign applications in areas like national security, healthcare, areas where data residency is important. And it's really critical that we do that. And what is needed for that is in the short term here and now The primary input is going to be pace of grid connection People talk a lot about energy pricing, and of course, it's an empirical fact that industrial energy prices in this country are not where they need to be to support our industrial applications and we're trying to do a lot on that. But in the short term, the thing that actually most matters is there's a desperate race for an eighteen to twenty four month investment horizon And actually, it's not as price sensitive as people make it out on an energy point of view. It's much more timeline sensitive than price sensitive And so the biggest thing that we've focused on is reforms to try and meet that objective. Yes, in the medium term, energy pricing is a question. And then in the longer term, the question becomes overall scale of generation and in particular renewables generation as well. And in that context, again, Britain is actually in a pretty good place We announced a major data center with SMRs in North Wales, We announced a majorataenter with solar and battery generation in Scotland. We've got a major data center project in Cullam alongside at the home of a series of research and nuclear fusion Is there any chance of getting that open AI investment rebooted? What would that take Well, I think we continue to speak to openen AI. The first thing to say is that a number of those underlying projects continue because of course, open AI is one customer for data center capacity, but investors in those data centers feel that there is sufficient demand from other customers still within the country. And so OpenAI might have decided to look at other projects, but the projects themselves continue and we want to make sure we're the best place for a bunch of other customers, whether that's hyperscalers, whether that's frontier labs, whether it's other application firms as well looking to buy up capacity. Okay, so it doesn't sound like openp AI is interested for now Opening Eye is totally looking at a bunch of different projects. I think they've expanded their office. I think London is now their biggest office outside of the Bay Area, including for their research team. and so they're very welcome. I've had positive conversations with them, but we don't have a commitment on the replacement yet Okay, so just thinking about the data center rollout, right? As you say, the grid connections is a huge constraint how much of The is that it's hard to build anything in the UK. And how much is this sort of AI specific? Are you seeing any signs of a kind of backlash as we are seeing some of in the US? Well I think a lot of it is actually general rather than specific and it's really about the pace with which you get through the planning process, the pace with which you get through the permitting process, the pace with which you get transmission capacity opening up for these projects. And those are things that are shared by other projects as well. What we have done is a series of forforms where C picked out AI growth zone areas of data centers, get accelerated grit capacity allocated to them. We effectively surge in resourcing into those local authorities to accelerate planning. We have a central concierge team holding people's hands and trying to get them through the process. And so we're trying to do a lot too fix all of that. But I think we haven't seen the nature of backlash that you're describing for a couple of reasons. One just empirically, the United States has something like seventy five percent of global data center capacity today. They have built out a radically huge amount of data center capacity In Britain, we have, I think, at best, something like two gigawatts of compute capacity altogether, not even AI optimized compared to the United States is sort of fifty plus. And so Generally speaking, Britain doesn't have that many data centers. and so it would be pretty empirically detached to say that we have a major data center energy or water problem in this country because we actually don't have very many of them But the second thing which is really important is that the way we have tried to do these projects is that we have tried to do it in a way which meets the local communities's needs. In North Wales, for example, for ten years, people have been promising a small modular reactor investment, never happened As a result of what we did, the data center is now part of the customer. which means that investment in that small modular reactor can go forward and the community's aspirations for nuclear generation can be met alongside AI We're not trying to blow the lights out and the scale of our buildout, unlike, say the United States. What we are trying to do is say we require a level which is appropriate for national sovereignty. We want to do that in a way that furthers our clean energy goals alongside our AI goals. And especially if you build in areas of ennergy curtailment, Scotland, the Northeast Cumbria, we think the win win between AI and clean energy is very significant Okay, before we go to a break, I just have one more question about this open AI investment. So one of their stated concerns reportedly was uncertainty over the UK's copyright regime, right? And so they said that they were worried they may not be able to use copyrighted works to train their models Is there a choice between leading in AI and leading in the creative industries where the UK is also pretty strong. I think we've pretty clearly set out in the public conversation on this that we don't want to sell our creative sector down just to get an advantage in AI But let me step back and reflect on the kind of wider debate. There is pretty material uncertainty in the copyright regime in pretty much every jurisdiction Yes, it's true that in some jurisdictions, companies have been able to train the model today and then litigated in the court And so effectively there's uncertainty there, but that uncertainty is played out in the venue of the court. In Britain, we don't allow training of that sort under the copyright law at the moment. And so yes, firms might experience uncertainty that they can't do this thing. I think actually they experience certainty. they just experience a negative certainty. But that is no different to me fundamentally then the level of uncertainty they experience in pretty much every jurisdiction And that's because the underlying question has not been resolved in pretty much any geography in a durable way, which is How does the economics of AI when trained on the output of many writers, artists, musicians What is a fair economic distribution over time of that in a way that supports many countries' creative aspirations? And so far, I don't think anyone has j answered that question. So I'm not disputing that there is uncertainty on this question. I think that uncertainty is globally shared rather than differentially affecting the United Kingdom I think we should go to a break now. And when we get back I'm going to be asking about how we're doing at consuming AI. Are we ready for the AI wave We are back from the break Big question, is the UK workforce ready? for the AI related disruption we're going to see. and Do we need to be thinking about our social safety net? L packed in there. I think here's everything thing about it. I think the public policy conversation and the public conversation does not reflect the significance of this question at the moment. I think both in terms of the micro evidence, the model's capability today, the macro fear that people might have about the future I think we haven't really measured up to that and I feel desperately that this is one of the most important questions for economic policy and frankly, a kind of labor market policy that we need to think about much more. And so that's the starting position. I think the workforce is actually slightly better than us on this question One of the things has an interesting pattern here as it is elsewhere is that you ask People in a firm how many of you use AI chatbots in your day to day work and you get answer You then ask the firm, how many formal licenses for AI chatbots have you purchased? And you get an answer that's often something like twenty percentage points lower In other words, individuals in our workforce are using these tools, feel that it's augmenting a number of instances of work in their day to day lives, But the pace at which organizations and politics has caught up with that felt experience of individuals has lagged behind very materially And so I think there's a lot more to do You asked about social security, the web frrame my thought on this is that I think we have to be very alert and more agile in how we think about social security support in the future. but I don't currently see a material enough impact that's different from other things going on the labor market to argue for a wholesale change. There's some sort of signs of AI adoption, but no signs really of large scale AI displacement. In many ways, actually there's some signs of AI augmentation That said, we know from the model's capability pace with which they moveved from one application to another application. they've gone from language to solving some fundamental maths questions to being a big part of scientific application, being used in contact centers, being used in marketing, being The pace with which they're moving across domains is so significant that we have to take seriously the idea that those advances combined with advances in more agentic AI that those combined can make a bigger dent on jobs in the future. And as a result, I think we have to prepare as a big motivation for why we set up this AI economics issute. with Simon Johnson, so that Britain has some of the earliest, quickest, best understanding of labour market impacts so that we can react very promptly as well. Okay, but suppose there is mass unemployment in consultants, lawyers, accountants, the kinds of highly paid services that Britain specializes in could that reaction look like? What if anything would need to change British lawyers just to take that example, are brilliant at also doing a bunch of other things outside of law. And so One, we're not seeing a big impact in terms of lawyers losing their jobs. So it's very important to be clear about that. In some ways, actually we're seeing firms use AI to grow their headcount, become more productive and gain more market share. But secondly, even if I kind of partially give away that assumption that you see job losses for lawyers, what we're finding is that lawyers who leave law firms, for example, are actually really good at them also being product managers in AI firms in services. We're finding that AI firms need more and more people to help support the implementation of AI in firms. and that is a role that is really about client engagement, relationship building, communication, human understanding of technical things applied in a real world firm context. And so there are lots of alternative job options available in particular to a set of professions that you mentioned, lawyers being one of them That means that the quality and pace of our active labor market support, job search, job matching, progression support, that can be radically better and we might expect some increase in the demand for that. and we have to do a good job of both in the private and public sectors making sure that measures up. And so then having contested this assumption in the first instance, if I sort of seed it, the question is, what you do if you see very large scale impacts of the sortce No I think We can't rule that hypothetical thing out in the long term. and so we do have to think about what are ways in which we might explore public policy support both in terms of active labour market policy. I think the conversation too quickly goes to financial support directly. I'm not ruling that out as a prospect we have to look at. We are doing a series of economic work and th think about scenarios and clearly the fiscal and social seecurity conversation is a part of that. But I think the most important part of it has to be the quality of active labor market support. What can we do to use AI bring frontier technology to the dignity of people seeking new jobs and trying to progress in the jobs that are durable Okay, so AI kills the job but then helps people find the new job so it's fine. AI kills jobs, creates jobs, and you just have to make sure you get people through that process as smoothly as you can The public seems quite skeptical about AI. so concerns about jobs, information, creativity, has AI Minister Are you worried about those negative narratives And the top line is I'm very worried it becausecause the framing is should we have more of this or should we have less of this And I worry because that framing to me is very inaccurate. The framing is Do we have it here? And do we shape it in the way we want to get the app upside and limit the downside or are we bystanders in this process where it's going to happen in an inevitable way from elsewhere, from Tch billionaires sitting in Silicon Valley or from China and we are exposed to it because this is a very significant question for our economy. Already, you know better than me, that the highest source of capital investment in countries that are actually growing is from AI infrastructure, as is the case for jobs. If you look at national security, the wars in Ukraine and Iran have shown that you can't just set out drones and AI capability. In fact, you can only be a participant and player if you build that capacity And right across pretty much most domains of interest to the British public, I think that's the case. So what do we do about that debate? I think one We have to try and build trust in the first instance by calling things out when they are clearly bad. We have to say that is not the right thing by British values. and we have to call it out, we have to stamp it out in regulation That's what we've done we've banned social media for under sixteen,' bannedI romantic chatbots for young people. I think we're the most advanced when it comes to mitigating the downside risks of AI online safety The second is, yes, add nuance to that debate. We talk about AI and label it as such when we talk about the negative things. But when we consider the opportunities of it we often talk about things that are not labeled AI. So we say, isn't it really easy in my constituency to navigate rural villages on Google Maps Well, that's pretty powered by AI. We talk about the material advances in vaccine discovery and we don't recognize AI's contribution to increasingly being the primary driver of vaccine discovery. And so I think alongside building trust by regulating where things are going wrong, we have to make sure that we add nuance so that the British public has a robust debate on it, but armed with the facts of what is being served by AI and what is being damaged by AI as well. And the third thing, which is really critical is I think the national security and sovereignty part of this. We talk a lot about the size of defense funding in this country but not the scope of it. And to me, increasingly, we have to have a much more rigorous debate about how we're building the capability on cybersecurity on drones and counter drone defense systems, all of them underpinned by AI, which is now the central determinant of both our domestic and national security. So not a very ambitious agenda then. Well before I let you go, I have one final question about Your overall optimism for the potential of AI to improve growth productivity, right? Because you've got again, this is the huge debate among economists, right Is it going to do diggly squat? orr is it going to know lead to the singularity where AI is improving our researchers and we get infinite growth and everything's brilliant. we all get a free puppy So on a scale of know one to ten and being that singularity. Where do you think we are? What do you think is the potential of this tool You know, on the sort of It's very skeptical sort of Aonogue scale through to singularity bros in Silicon Valley. I think I'm probably on like a six, six and a half, if we kind of get the rates of adoption that we expect, I think The potential here is larger than most technological shifts that we have seen historically because the scope and scale and speed of this adoption feels very different to historic technologies. And so I'm not saying we're kind of seeing blow out double digit annual rate of productivity uplift, but I think it can be materially more than just the sort of eking out up to two percent and the historic patterns that we've seen from technology. And that said, I'm also very mindful that that alone is not going to create fulfilling sort of lives for everyone in this country. And so that is an important objective, it iss a critical, I think primary objective, but we have to make sure that we get all the other things right from a regulation and governance point of view to make sure that benefit is shared widely, enjoyed enjoyed fulfillingly And do you think the UK is already on the road to that path? Do you think we're already seeing some effect as AI in our productivity performance I think so. I actually think it's still pretty slow. And so I actually if you sort of say to me, hey, if you see a productivity improvement today is that coming from me, I think we're actually still eking out a lot of the upside from kind of core digital transformation, teechnology is being applied and taken up in workforces But we are seeing in some firms and divisions, a pretty significant productivity outlift. So Yes, we're seeing sort of early signs of it, but nowhere near sort of macro relevant relevant scale Well, that is a great note to end things on. Keneshka, thank you so much for joining me This is great. Thankks very much That is all for this week You have been listening to the Economics show with Samaya Kainnes If you enjoyed the show, then I would be so happy if you could give us five stars on your podcast app and write a nice comment I will read it This episode was produced by Michia Franl Dubal with original music from Brene Turner and sound design from Seaan Nagarty Andrew Giuljiiaz was the broadcast engineer. Our executive producer is Manuela Saragosa, and Flan Phillips is the FT's head of audio I'm Samaya Kainnes Thanks for listening