WEBVTT - Why AI-Driven Productivity Is a Decade Away in the UK 0:00:03.120 --> 0:00:18.600 Bloomberg Audio Studios, Podcasts, Radio News. 0:00:19.760 --> 0:00:22.560 Welcome to Maren Talk's Money, the podcast in which people 0:00:22.600 --> 0:00:24.960 who know the markets explain the markets. 0:00:25.120 --> 0:00:26.239 I'm Maren sum Zetweb. 0:00:26.480 --> 0:00:29.320 This week, we're focusing on artificial intelligence and the role 0:00:29.400 --> 0:00:32.280 we really hope it's going to play in boosting productivity. 0:00:32.640 --> 0:00:34.560 For a while, it seemed that the consensus was that 0:00:34.640 --> 0:00:38.720 AI was going to revolutionize everything. Last year, Goldman Economists 0:00:38.800 --> 0:00:42.520 estimated that AI would increase annual global GDP by seven 0:00:42.560 --> 0:00:47.080 percent over ten years. The IMF predicted that AI has 0:00:47.120 --> 0:00:50.960 the potential to reshape the global economy. Look at pretty 0:00:51.040 --> 0:00:54.120 much any report from any boog organization and you will 0:00:54.160 --> 0:00:59.480 see something very similar, the idea that this will change everything. 0:00:59.560 --> 0:01:02.360 Here in the the labor government clearly believe it too. 0:01:02.560 --> 0:01:05.240 They spent a good part of the recent Investment Summit 0:01:05.319 --> 0:01:08.920 positioning themselves as a champion for AI, called it an opportunity, 0:01:09.160 --> 0:01:12.560 saying that the country needs to run towards it. Not 0:01:13.040 --> 0:01:16.440 everyone is so convinced that AI is a silver bullet 0:01:16.480 --> 0:01:18.440 for all our productivity. Rose and I'm afraid that we 0:01:18.480 --> 0:01:20.880 are cynics on this podcast. We rather believe that too, 0:01:21.200 --> 0:01:24.840 and so does this week's guest Diane Coyle Professor. Dame 0:01:24.920 --> 0:01:27.520 Diane Coyle is the Bennett Professor of Public Policy at 0:01:27.560 --> 0:01:28.839 the University of Cambridge. 0:01:28.959 --> 0:01:30.440 The first book of hers I read. 0:01:30.280 --> 0:01:33.199 Was GDP, A brief but affectionate History, and I often 0:01:33.280 --> 0:01:36.959 refer to that even now. Her latest Worse Cogs and 0:01:37.080 --> 0:01:39.720 Monsters What Economics is and what it should be? That 0:01:39.840 --> 0:01:42.319 explored the challenges for economics in the context of the 0:01:42.360 --> 0:01:43.360 digital transformation. 0:01:43.600 --> 0:01:45.280 I particularly liked the title. 0:01:45.520 --> 0:01:47.440 She's got a new book coming out in April called 0:01:47.440 --> 0:01:49.760 The Measure of Progress, and maybe we'll get her back 0:01:49.800 --> 0:01:52.320 on to talk about that then. For now, her current 0:01:52.400 --> 0:01:55.640 research focus is on productivity and on economic measurements. 0:01:56.000 --> 0:01:56.960 Earlier this year she. 0:01:56.960 --> 0:01:59.720 Authored a very good article that caught our eye, Don't 0:01:59.760 --> 0:02:02.400 bank on AI being a quick fix. 0:02:02.120 --> 0:02:08.840 For elusive growth? Diane, welcome to Merin Talks Money. 0:02:09.000 --> 0:02:09.880 It's a real pleasure. 0:02:10.240 --> 0:02:11.960 And now one of the reasons that we asked you 0:02:12.000 --> 0:02:13.799 to come on, apart from all these wonderul books and 0:02:13.840 --> 0:02:16.200 writings that we've talked about, was an article that caught 0:02:16.200 --> 0:02:19.120 our eye earlier this year called Don't bank on AI 0:02:19.240 --> 0:02:22.960 being a quick fix for elusive growth? And rather like 0:02:23.040 --> 0:02:25.040 a lot of other people on this podcast, we've talked 0:02:25.080 --> 0:02:28.200 a lot about how everything's going to be absolutely fine 0:02:28.280 --> 0:02:31.760 as soon as productivity pops up, and some somehow, one 0:02:31.760 --> 0:02:34.320 way or another, technology and AI in particular, will do 0:02:34.400 --> 0:02:37.360 that for us. So we will go from productivity growth 0:02:37.400 --> 0:02:39.160 being as a miserable one or one and a half 0:02:39.200 --> 0:02:41.320 percent across the Western economy is up to two and 0:02:41.320 --> 0:02:46.320 a half maybe more, and all the problems will disappears 0:02:46.520 --> 0:02:47.040 going to happen? 0:02:47.120 --> 0:02:52.160 Is it? Well? Technologies do drive productivity growth over the 0:02:52.200 --> 0:02:55.880 longer term, but they tend to be much slower than 0:02:56.080 --> 0:03:00.079 people often imagine when there's a lot of new and 0:03:00.120 --> 0:03:04.080 exciting news about what's coming along. And the classic example 0:03:04.120 --> 0:03:07.480 from economic history is electricity, which is a nineteenth century 0:03:07.480 --> 0:03:11.040 technology fundamentally, but it didn't affect productivity growth until at 0:03:11.120 --> 0:03:14.360 least the nineteen thirties. And AI might be a little 0:03:14.360 --> 0:03:17.560 bit quicker, but I think they're still looking at a 0:03:17.600 --> 0:03:20.480 decade or so before it starts to have a measurable 0:03:20.520 --> 0:03:23.360 impact on productivity, and there are lots of reasons why 0:03:23.840 --> 0:03:27.359 that uptake is slow. So I get quite concerned about 0:03:28.120 --> 0:03:32.040 the silver bullet claims that you referred to, because it 0:03:32.120 --> 0:03:34.920 raises expectations and they get disappointed, and then you get 0:03:34.960 --> 0:03:39.360 a sort of backlash against technology. AI is an amazing technology. 0:03:39.400 --> 0:03:41.880 It will be able to do some extraordinary things for us, 0:03:42.240 --> 0:03:44.320 but it's not going to fix productivity. Is ye. 0:03:44.480 --> 0:03:47.240 Should we go back briefly to electricity, because that's very 0:03:47.240 --> 0:03:51.320 interesting and is that? Was that an infrastructure problem? Because 0:03:51.320 --> 0:03:55.200 obviously connecting everyone electricity is a huge infrastructure problem. Got 0:03:55.240 --> 0:03:58.600 to be expensive, it's difficult, and you can't immediately see 0:03:58.760 --> 0:04:01.080 the impact of it. So you get that presumably Jacob 0:04:01.160 --> 0:04:03.440 effect right where you've got to spend all the money 0:04:03.520 --> 0:04:06.560 up front and the productivity doesn't appear until everyone's connected, 0:04:06.920 --> 0:04:09.680 and then it comes. So that's why it takes many decades, 0:04:09.920 --> 0:04:10.960 did take many decades. 0:04:11.320 --> 0:04:13.840 There's several reasons for the delay. Infrastructure is certainly one 0:04:13.880 --> 0:04:16.320 of them. So with electricity, it was building out the networks. 0:04:17.160 --> 0:04:22.000 With AI, it's building the data centers and the computational power, 0:04:22.440 --> 0:04:26.760 but also actually getting the data in order because AI 0:04:26.839 --> 0:04:30.240 eats data, that's its fuel, and a lot of the 0:04:30.400 --> 0:04:34.760 data that we have isn't available, isn't interoperable, isn't in 0:04:34.760 --> 0:04:38.080 a good shape, isn't high quality, And so there's a 0:04:38.120 --> 0:04:41.440 big challenge in that sort of intangible infrastructure that we're 0:04:41.480 --> 0:04:45.240 going to need for AI. But there are other investments 0:04:45.279 --> 0:04:48.240 that are needed as well. In the electricity example, it 0:04:48.360 --> 0:04:51.720 was building new kinds of factories because you can have 0:04:52.279 --> 0:04:55.920 a dynamo on each machine, and the assembly line was 0:04:56.000 --> 0:05:01.320 part of what made the electricity revolution create productivity gains. 0:05:02.320 --> 0:05:06.920 But then also once you had to build new factories, 0:05:07.000 --> 0:05:09.799 you needed to build new transportation networks to get workers 0:05:09.800 --> 0:05:14.000 to them. And even then there's a reorganization of work 0:05:14.040 --> 0:05:16.960 that's needed, so the different kinds of jobs in factories 0:05:17.000 --> 0:05:21.080 and so on. So it's infrastructure directly involved, it's other 0:05:21.200 --> 0:05:24.520 kinds of infrastructure and reshaping the economy. And then it's 0:05:24.560 --> 0:05:26.960 what actually did people's jobs involved from what are the 0:05:26.960 --> 0:05:29.320 skills that needed for that? So all of those things 0:05:29.360 --> 0:05:31.239 need to come together, and that's why you get delayed. 0:05:32.440 --> 0:05:35.040 Let's go to the optimistic end first, So go back 0:05:35.120 --> 0:05:37.640 to that bit and talk about the main things that 0:05:37.680 --> 0:05:40.560 people expect AI to do for us. So if you 0:05:40.680 --> 0:05:45.640 divided into generative and predictive, it's really generative AI that 0:05:45.800 --> 0:05:48.320 is the thing that we really expected productivity to gain 0:05:48.800 --> 0:05:49.640 gains to come from. 0:05:49.680 --> 0:05:50.520 Should they come. 0:05:50.720 --> 0:05:55.640 Yes, and the models are still improving dramatically over time 0:05:56.760 --> 0:05:59.719 according to the benchmarks that the industry itself uses to 0:06:00.000 --> 0:06:02.960 tess that. So I've just come back from a week 0:06:03.279 --> 0:06:06.240 in Silicon Valley, and every time I've been there in 0:06:06.279 --> 0:06:08.839 the last couple of years, people have been saying, oh, well, 0:06:09.040 --> 0:06:11.760 the thing that the next generation of models will do 0:06:11.920 --> 0:06:15.440 for us is absolutely amazing, And then that has been true. 0:06:15.480 --> 0:06:18.520 So I think technical advances are still extraordinary. But then 0:06:18.800 --> 0:06:20.640 to think about the economic impact, you need to think 0:06:20.640 --> 0:06:24.800 about what will that actually do in specific activities in 0:06:24.800 --> 0:06:28.160 the economy. So we've got the language models, and they 0:06:28.279 --> 0:06:32.160 are clearly able to do many kinds of administative task, 0:06:32.240 --> 0:06:36.480 so they could create productivity that way. There are visualization models, 0:06:36.640 --> 0:06:41.000 so let's think about what tasks involve visual inspection, or 0:06:41.040 --> 0:06:46.200 where visualization would help improve productivity. There might be things 0:06:46.240 --> 0:06:49.000 about three D real time visualization that are going to 0:06:49.000 --> 0:06:52.760 become much more possible. So mapping from what the technology 0:06:52.800 --> 0:06:56.480 can do in general to what are the actual things 0:06:56.560 --> 0:06:58.400 in a workplace it might be able to do for 0:06:58.440 --> 0:06:59.960 people is part of the journey. 0:07:00.120 --> 0:07:02.760 I suppose what I'm really trying to ask is, as 0:07:02.960 --> 0:07:07.800 an ordinary worker, an ordinary person, what is it that 0:07:07.880 --> 0:07:08.800 we will see change? 0:07:08.800 --> 0:07:10.520 What will happen around us? I mean, we know. 0:07:10.480 --> 0:07:14.760 What's happened with digitalization and the rise of the smartphone, 0:07:14.800 --> 0:07:16.480 all those kinds of things. We can see the difference 0:07:16.480 --> 0:07:19.840 around us. We understand how that productivity changed, right. We 0:07:19.920 --> 0:07:22.440 understand how the smartphone changed how we work. We understand 0:07:22.440 --> 0:07:25.160 how our computers have changed things. We understand how cloud 0:07:25.160 --> 0:07:27.360 computing has changed the way we work and operate, etc. 0:07:27.920 --> 0:07:30.520 But as generative AI moves forward, what is it that 0:07:30.520 --> 0:07:32.520 we will we will see and we will feel what 0:07:32.560 --> 0:07:34.320 will be different for us? 0:07:34.600 --> 0:07:37.520 Let's think about it in two ways. What products are 0:07:37.520 --> 0:07:39.600 we going to experience as consumers that will make things 0:07:39.680 --> 0:07:43.720 much better? And for there, I'd be looking for time 0:07:43.800 --> 0:07:47.000 saving things, things that I don't like doing that AI 0:07:47.120 --> 0:07:49.520 and robots might be able to do for me. Hans 0:07:49.640 --> 0:07:52.000 Rostling a long time ago did a wonderful ted talk 0:07:52.000 --> 0:07:54.640 about the impact of the washing machine on people's lives 0:07:54.640 --> 0:07:57.760 and how extraordinary it's been. At the moment, AI can't 0:07:57.800 --> 0:08:01.960 do that because it's not embedded in robotics. It's abstract, 0:08:01.960 --> 0:08:06.280 it's not computers. Work is starting on that kind of 0:08:06.400 --> 0:08:10.160 embodied AI and down the line, if that can do 0:08:10.280 --> 0:08:13.560 my laundry for me with a super washing machine that 0:08:13.600 --> 0:08:15.600 will also fold and iron clothes. That's going to be 0:08:15.640 --> 0:08:20.280 something quite amazing in terms of daily life applications in medicine, 0:08:20.320 --> 0:08:27.080 so a lot of opportunities for medical scanning, monitoring, tailored 0:08:27.120 --> 0:08:31.200 personalized medications. Huge potential there to improve the multi of 0:08:31.200 --> 0:08:33.800 people's lives. So that's one way to think about it, 0:08:34.240 --> 0:08:35.640 and then the other way to think about it is 0:08:35.640 --> 0:08:38.559 all the process innovations it can help us to introduce 0:08:38.559 --> 0:08:41.400 at work that will save time do boring work much 0:08:41.440 --> 0:08:44.679 more cheaply, and then they're free up the time for 0:08:45.160 --> 0:08:47.880 workers to do things that are going to add more 0:08:47.960 --> 0:08:51.520 value for their customers. So if you think about the 0:08:51.559 --> 0:08:55.719 medical examples, well, it saved doctors time writing letters so 0:08:55.760 --> 0:08:58.199 that they can spend more time with their patients. If 0:08:58.240 --> 0:09:01.120 you work in manufacturing, is it going to speed up 0:09:01.120 --> 0:09:04.920 those processes or help you track waste or energy use 0:09:05.040 --> 0:09:08.360 in better ways than are currently available. And that's the 0:09:08.440 --> 0:09:12.280 kind of process innovation that allowed the just in time 0:09:12.400 --> 0:09:16.680 revolution and manufacturing productivity in the nineteen seventies and early eighties. 0:09:17.160 --> 0:09:19.559 So thinking about it in those two ways will give 0:09:19.600 --> 0:09:22.840 us some clues about where we can hope for ultimately 0:09:22.880 --> 0:09:24.079 productivity gains. 0:09:24.160 --> 0:09:30.120 My AI reverse the trying trend from a consumer's point 0:09:30.160 --> 0:09:32.440 of view of having pretty much all the work that 0:09:32.559 --> 0:09:35.400 used to be done by companies when you buy something, 0:09:35.520 --> 0:09:38.200 or book something, or rain something being outsourced to you, 0:09:39.040 --> 0:09:41.800 and the maddening difficulty that that caused with me, might 0:09:41.840 --> 0:09:43.920 say is one of the big downsides of the product 0:09:44.200 --> 0:09:45.640 of the technology revolutions. 0:09:45.679 --> 0:09:50.040 So far, it might. It depends on the incentives of 0:09:50.080 --> 0:09:51.959 the companies to do so, of course, but we've all 0:09:51.960 --> 0:09:54.640 paid that time tax of getting into voicemail hell and 0:09:54.640 --> 0:09:56.800 not being able to get talk to a human being. 0:09:56.800 --> 0:09:59.319 Or chatbot and talking to a chat but I'm begging 0:09:59.360 --> 0:10:00.880 it to let you talk to a person so you 0:10:00.920 --> 0:10:02.360 can get to the end of the process. 0:10:02.400 --> 0:10:03.280 But not happening. 0:10:03.440 --> 0:10:06.400 So one model might be what's happened in retailing, or 0:10:06.440 --> 0:10:08.599 is happening in retailing where you used to go to 0:10:08.640 --> 0:10:13.040 the shop and somebody would scan your groceries and possibly 0:10:13.080 --> 0:10:14.800 even put them in a bag for you back in 0:10:14.840 --> 0:10:18.320 the old days, and increasingly that labor has been outsourced 0:10:18.320 --> 0:10:20.319 to us, and so now we've got the automatic checkouts 0:10:20.320 --> 0:10:22.559 where you do your own scanning and packing and you've 0:10:22.559 --> 0:10:24.439 got to persuade the machine that you haven't put the 0:10:24.480 --> 0:10:26.959 item in the bag the wrong way. But now there 0:10:27.000 --> 0:10:29.640 are stores where you just take the stuff off the 0:10:29.640 --> 0:10:32.199 shelf and go away with it, and so that then 0:10:32.200 --> 0:10:36.160 would be labor saving. How we measured these in productivity 0:10:36.440 --> 0:10:40.200 obviously has a different effect because the automatic checkouts and 0:10:40.200 --> 0:10:43.560 the shops are going to increase the measured productivity of 0:10:44.559 --> 0:10:47.600 the retailers because it's not measuring your unpaid labor in 0:10:47.640 --> 0:10:51.120 doing that scanning and packing yourself. So we don't know 0:10:51.200 --> 0:10:54.400 yet whether that AI enabled or you walk into the 0:10:54.400 --> 0:10:56.640 store and take it out is going to last. Maybe not, 0:10:57.280 --> 0:10:58.920 but that would be the kind of journey we might 0:10:58.960 --> 0:11:02.000 hope for. Elsewhere, there are some call center experiments with 0:11:02.120 --> 0:11:06.240 AI where it actually seems to be improving the experience 0:11:06.280 --> 0:11:09.160 that consumers have because the AI can train call center 0:11:09.200 --> 0:11:12.040 workers and gives them better scripts so they become better 0:11:12.080 --> 0:11:14.240 at doing what they're doing. But I think you know 0:11:14.840 --> 0:11:16.800 that's what we need to see, that's what's going to 0:11:16.840 --> 0:11:17.240 help us. 0:11:17.600 --> 0:11:19.679 Interesting, let's go back to what you just said about 0:11:19.679 --> 0:11:23.040 the measurement of productivity, because it's a very interesting point 0:11:23.280 --> 0:11:27.360 that these making us do the work improve the productivity 0:11:27.360 --> 0:11:29.959 of the firm, and if you then take that work 0:11:30.160 --> 0:11:33.280 back in house using AI, it may make no difference 0:11:33.320 --> 0:11:37.839 at all to what looks like productivity. So the change 0:11:37.840 --> 0:11:40.000 has happened, but it's not caught in the numbers. And 0:11:40.040 --> 0:11:41.679 this is one of the things that you specialize in 0:11:41.880 --> 0:11:44.200 writing about how the numbers do not count to the 0:11:44.280 --> 0:11:47.319 change as digitalization and technology moves. 0:11:47.120 --> 0:11:52.880 Toward, the numbers really entirely miss the digital revolution in 0:11:53.000 --> 0:11:57.520 our experience as consumers and workers. So there are all 0:11:57.600 --> 0:12:00.440 kinds of phenomenon that have been enabled by the digital 0:12:01.120 --> 0:12:05.880 from global production networks that rely on communications and logistics 0:12:06.000 --> 0:12:09.960 that have been enabled by digital to the way that 0:12:10.040 --> 0:12:13.840 many manufacturing companies wrap services around their products it's called 0:12:13.840 --> 0:12:17.760 servitization in the literature, to all the platform models and 0:12:17.840 --> 0:12:21.319 these free goods, and how do we take account of 0:12:21.360 --> 0:12:23.520 the zero price that we pay for the free goods? 0:12:23.559 --> 0:12:26.440 But what are the other implications of all of that 0:12:27.920 --> 0:12:31.000 just not captured in the economic statistics. It's always been 0:12:31.040 --> 0:12:34.000 the case that it's really hard to measure the major 0:12:34.080 --> 0:12:37.520 impacts of big innovations in statistics. The statistics are good 0:12:37.520 --> 0:12:41.360 at small changes, but not big changes, And so economists 0:12:41.400 --> 0:12:44.280 would also look at things like improved life expectancy or 0:12:46.120 --> 0:12:50.400 the quality of life as well to supplement GDP figures. 0:12:50.400 --> 0:12:54.400 But at the moment, this extraordinary change in how we 0:12:54.640 --> 0:12:57.040 lead our life since the smartphone arrived in two thousand 0:12:57.080 --> 0:13:00.199 and seven is pretty much invisible. And that's a lot 0:13:00.200 --> 0:13:02.520 of what my next book is going to be about. 0:13:02.520 --> 0:13:05.240 Okay, And what are the solution as to that? How 0:13:05.240 --> 0:13:05.920 do we pick it up? 0:13:05.920 --> 0:13:08.120 How do we change the way we measure I mean, 0:13:08.280 --> 0:13:10.520 in a very minor way. We get very irritated on 0:13:10.559 --> 0:13:13.559 this podcast a lot by GDP measurements because we're maddened 0:13:13.559 --> 0:13:17.400 by people constantly talking about GDP instead of GDPP ahead. 0:13:17.920 --> 0:13:20.000 But that's a very minor problem in the context of 0:13:20.000 --> 0:13:22.920 what you're talking about. So how do you reframe the 0:13:22.960 --> 0:13:24.480 measurement of economic progress? 0:13:25.040 --> 0:13:27.640 Well, a lot of it is about collecting new data 0:13:27.679 --> 0:13:29.679 and new statistics, and we've got to use the technology 0:13:29.679 --> 0:13:33.640 itself to help us to do that. It is slowly starting. 0:13:33.800 --> 0:13:39.000 So on things like global production chains, the impact of 0:13:39.240 --> 0:13:43.200 a tariff on imports is now going to be different 0:13:43.240 --> 0:13:45.360 than it would have been in the nineteen sixties because 0:13:45.400 --> 0:13:47.400 to manufacture anything, you've got to import a lot of 0:13:47.400 --> 0:13:50.800 the components, but the data are starting to capture what's 0:13:50.880 --> 0:13:53.280 called value added in trade, how much do you need 0:13:53.320 --> 0:13:57.160 to import to manufacture particular kind of export. So that's 0:13:57.200 --> 0:14:00.160 slowly starting. But there's all kinds of new days to 0:14:01.080 --> 0:14:06.360 collecting credit card data, using administrative data like taxes, using 0:14:06.400 --> 0:14:11.440 mobile phone data, webscraping, and the research into how to 0:14:11.520 --> 0:14:14.200 develop statistics using these new methods is kind of in 0:14:14.200 --> 0:14:16.400 its early days. So I think we're probably in for 0:14:16.480 --> 0:14:19.800 a I don't know, twenty year gap before we've got 0:14:19.840 --> 0:14:22.840 a settled set of statistics on the digital economy. 0:14:23.200 --> 0:14:27.960 Okay, So if we were in optimistic mode, might we 0:14:28.000 --> 0:14:31.920 say that, in fact, living standards are improving much faster 0:14:32.000 --> 0:14:34.000 than our current statistics tell us they are. 0:14:34.320 --> 0:14:36.880 I think they're improving in some ways and for some people, 0:14:37.120 --> 0:14:40.920 faster than statistics tell us, But in other ways they 0:14:41.000 --> 0:14:44.280 are less good at telling us what's going on. The 0:14:44.280 --> 0:14:47.040 obvious example people always give is that statistics don't tell 0:14:47.080 --> 0:14:51.160 us anything about the environmental costs of recent economic growth, 0:14:51.920 --> 0:14:56.040 which are going to hit productivity because places will get 0:14:56.080 --> 0:14:59.600 flooded much more than they used to be. Agricultural productivity 0:14:59.640 --> 0:15:02.840 will decl line because soil quality and biodiversity is declining, 0:15:03.480 --> 0:15:06.120 so they will have real economic consequences, none of which 0:15:06.120 --> 0:15:10.480 are measured. So there are pluses and minuses, and we 0:15:10.520 --> 0:15:13.520 need really a broader framework to think about all of 0:15:13.560 --> 0:15:18.160 these things that are outside the traditional standardized manufacturing economy 0:15:18.200 --> 0:15:20.840 for which these statistics were devised in the nineteen forties. 0:15:21.120 --> 0:15:22.680 Okay, that wasn't the answer I was looking for. 0:15:22.760 --> 0:15:24.160 I was looking for you to tell me that the 0:15:24.600 --> 0:15:27.880 statistics were hiding progress, as in effect, you're telling me 0:15:27.880 --> 0:15:29.479 they're hiding a type of decline. 0:15:29.560 --> 0:15:33.560 They're hiding both the hiding decline and progress. And if 0:15:33.560 --> 0:15:38.320 you think even about things like medical innovations, something that's 0:15:38.360 --> 0:15:41.440 now pretty basic, like the ability to do a really 0:15:41.560 --> 0:15:46.840 straightforward cataract operational hip replacement operation, which has become a 0:15:46.920 --> 0:15:50.800 much more standardized and straightforward process because of advances and technology. 0:15:51.000 --> 0:15:53.760 They're not big, shiny advances like generative AI models, but 0:15:53.760 --> 0:15:56.960 they're really important improvements in the quality of people's lives. 0:15:57.240 --> 0:15:59.960 Well, let's move from that, just briefly, because no one's 0:16:00.120 --> 0:16:02.400 well on this for too long, to the NHS and 0:16:02.440 --> 0:16:06.360 the possibility the possibility that there are ways that AI 0:16:06.520 --> 0:16:10.480 might be able to help us with the NHS in 0:16:10.520 --> 0:16:13.080 a way that constant large lugs of cash do not. 0:16:13.960 --> 0:16:18.000 Yes, So it's obviously one of the hopes for AI 0:16:18.160 --> 0:16:22.400 that it will enable big improvements in public service productivity 0:16:22.440 --> 0:16:26.000 and particularly in the NHS, And we already talked a 0:16:26.040 --> 0:16:30.400 little bit about some of the potential there. There are 0:16:30.480 --> 0:16:34.440 already quite well established gains to be made from having 0:16:34.480 --> 0:16:40.840 AI screen test results and provide that information to doctors 0:16:40.960 --> 0:16:45.160 or making decisions, a lot of potential for administrative simplicity. 0:16:45.680 --> 0:16:48.160 But to me, it's a question of organizational than it 0:16:48.240 --> 0:16:52.120 is of technology. It boggles my mind that it isn't 0:16:52.160 --> 0:16:56.560 already the case that the NHS has not mandated all 0:16:56.600 --> 0:16:59.880 of the hospital trust and GP surgeries to use interoper 0:17:01.000 --> 0:17:05.320 technology and standardized forms of data. So even getting to 0:17:05.440 --> 0:17:09.080 the basics of how easy is it going to be 0:17:09.160 --> 0:17:12.919 to adopt AI, we're not there yet. And it is 0:17:13.000 --> 0:17:16.439 clear that there's been underin investment in very basic technology. 0:17:17.640 --> 0:17:20.280 My husband was in hospital last year and we noticed 0:17:20.280 --> 0:17:22.120 that the nurses had to spend all of their time 0:17:22.160 --> 0:17:25.800 with their backs to patients, typing things into computers with 0:17:25.920 --> 0:17:28.960 these really user unfriendly kinds of interfaces that were very 0:17:29.000 --> 0:17:32.240 out of date. So is there potential for AI to 0:17:32.280 --> 0:17:35.000 fix that, sure, but that's going to need some investment 0:17:36.119 --> 0:17:39.840 of money, but also time to restructure those information flows. 0:17:39.880 --> 0:17:44.880 Within the NHS. There's an issue about data and data sharing, 0:17:45.280 --> 0:17:49.679 and there is real conservatism and risk aversion across the 0:17:49.760 --> 0:17:53.440 NHS and linking up people's data. Recent announcements about the 0:17:53.520 --> 0:17:57.440 NHS the NHS app bringing that together for individual patients. 0:17:57.440 --> 0:18:00.879 That's really welcome, but I think that fundamental lack of 0:18:00.920 --> 0:18:04.159 trust in the possibility of data sharing. How secure will 0:18:04.200 --> 0:18:06.320 it be? Who's going to profit from that? Will it 0:18:06.320 --> 0:18:09.640 be private sector companies or will you? Will patients see 0:18:09.720 --> 0:18:13.600 some benefit? That's a really big hurdle to overcome yea. 0:18:13.600 --> 0:18:15.840 And not helped by the way that public trust in 0:18:15.880 --> 0:18:19.040 the NHS has fallen very dramatically over four or five years. Yeah, 0:18:19.240 --> 0:18:21.520 last four or five years, So how would one then 0:18:21.560 --> 0:18:22.280 trust them further? 0:18:22.800 --> 0:18:26.760 Absolutely? And then there are questions about organization and financing. 0:18:27.119 --> 0:18:29.480 So there's lots of innovation in health tech. People are 0:18:29.520 --> 0:18:32.560 coming up with some amazing new products, but they've got 0:18:32.560 --> 0:18:35.000 to be paid for. So what's the business model that 0:18:35.080 --> 0:18:37.639 will enable the NHS to invest in those kinds of 0:18:38.160 --> 0:18:41.080 new innovations that might save them a ton of money, 0:18:41.440 --> 0:18:45.080 but their budgeting structures don't easily allow that, and then 0:18:45.200 --> 0:18:47.600 who's got the authority to make decisions? So AI is 0:18:47.640 --> 0:18:51.680 an information technology. It uses data and it creates useful 0:18:51.760 --> 0:18:55.200 information which will allow people to make decisions that can 0:18:55.200 --> 0:19:00.359 improve productivity. But that's going to be decentralized across the 0:19:00.359 --> 0:19:04.199 public services and including in health. They're quite hierarchical for 0:19:04.280 --> 0:19:09.080 reasons of accountability and also training. So how can the 0:19:09.080 --> 0:19:13.240 skills needed and the authority needed to take decisions be delegated. 0:19:13.680 --> 0:19:17.240 So there's no point giving nurses amazing information through AI 0:19:17.280 --> 0:19:19.959 products if they can't then do something with it. If 0:19:20.000 --> 0:19:22.760 they still have to call up doctor to get them 0:19:22.760 --> 0:19:25.000 to take the decision, they're still going to be bottlenecked there. 0:19:25.240 --> 0:19:28.760 So there's a whole range of barriers alongside this huge potential, 0:19:29.240 --> 0:19:32.320 and obviously the state of the NHS and the public 0:19:32.359 --> 0:19:34.000 confidence and it means this is one that we've got 0:19:34.000 --> 0:19:35.160 to crack as sin as possible. 0:19:36.040 --> 0:19:40.040 I imagine briefly, when I was introducing the current government 0:19:40.080 --> 0:19:42.520 talking a lot about wanting the UK to be an 0:19:42.560 --> 0:19:45.680 AI leader and how important that is and how they 0:19:45.720 --> 0:19:48.320 can help the state can help drive that. What do 0:19:48.320 --> 0:19:51.200 you think the state's role is here in driving investment 0:19:51.320 --> 0:19:53.920 into AI or improving the sector in the UK. 0:19:54.280 --> 0:19:57.040 So multiple roles. Actually, we do have a really good 0:19:58.600 --> 0:20:01.920 AI sector at the research frontier and in parts of AI, 0:20:02.200 --> 0:20:04.320 so the government has funded a lot of that research 0:20:04.320 --> 0:20:07.200 and should carry on doing so. Except it's got broad 0:20:07.240 --> 0:20:10.520 benefits and want to enable more startups. There's an issue 0:20:10.560 --> 0:20:14.320 about how do those companies grow. Many startups in tech 0:20:14.440 --> 0:20:16.200 still hope that they would be brought out by a 0:20:16.240 --> 0:20:18.680 big American company, and we need to be able to 0:20:18.720 --> 0:20:23.040 grow our own and understand which niches we have real 0:20:23.440 --> 0:20:26.639 advantage in and can export in. So that's again something 0:20:26.680 --> 0:20:30.199 that government needs to help with because public markets are 0:20:30.200 --> 0:20:32.879 making it very difficult for companies to grow across the board. 0:20:33.920 --> 0:20:37.280 But then there's a whole area of things about coordinating 0:20:37.320 --> 0:20:41.560 activities de risking investments. Is, for example, using public procurement 0:20:41.600 --> 0:20:46.200 in the NHS to encourage innovation and say there will 0:20:46.200 --> 0:20:49.720 be a market for new kinds of AI products. That 0:20:49.840 --> 0:20:52.359 it doesn't have to specify which ones are going to 0:20:52.359 --> 0:20:56.159 be brought or exactly which technological approaches needed, but the 0:20:56.200 --> 0:21:00.160 fact of it's called an advanced market commitment. In my world, 0:21:00.240 --> 0:21:01.960 and the fact that there will be a market through 0:21:02.000 --> 0:21:05.560 public procurement can also really help encourage us better technology 0:21:06.280 --> 0:21:10.439 and you know, public services can drive the take up 0:21:10.480 --> 0:21:13.639 of technology, get things to scale, make it cheaper for 0:21:13.720 --> 0:21:18.320 other users, demonstrate effectiveness. So this is an area where 0:21:18.320 --> 0:21:21.640 I think government action can make a big difference as 0:21:21.680 --> 0:21:24.240 long as it's realistic. So I think if the government 0:21:24.280 --> 0:21:26.280 to say we have this silver bullet and it's going 0:21:26.320 --> 0:21:29.240 to change things overnight, that that's very dangerous and they'll 0:21:29.240 --> 0:21:33.240 be backlash. But if it's got a realistic approach in 0:21:34.160 --> 0:21:36.840 you know, strategic industrial policy kind of framework, that could 0:21:36.880 --> 0:21:37.720 be really powerful. 0:21:38.240 --> 0:21:43.040 Okay, interesting, and what about the concerns around competition and 0:21:43.160 --> 0:21:46.280 the idea that what with the you know, the amount 0:21:46.359 --> 0:21:47.960 of investment, the amount of data. 0:21:48.040 --> 0:21:49.160 Except with that AI. 0:21:49.040 --> 0:21:52.160