Latest / Merryn Talks Money / Why AI-Driven Productivity Is a Decade Away in the UK

Transcript

Why AI-Driven Productivity Is a Decade Away in the UK · Merryn Talks Money

• 0:03
  Speaker 1: Bloomberg Audio Studios, Podcasts, Radio News.
• 0:19
  Speaker 2: Welcome to Maren Talk's Money, the podcast in which people
• 0:22
  Speaker 2: who know the markets explain the markets.
• 0:25
  Speaker 3: I'm Maren sum Zetweb.
• 0:26
  Speaker 2: This week, we're focusing on artificial intelligence and the role
• 0:29
  Speaker 2: we really hope it's going to play in boosting productivity.
• 0:32
  Speaker 2: For a while, it seemed that the consensus was that
• 0:34
  Speaker 2: AI was going to revolutionize everything. Last year, Goldman Economists
• 0:38
  Speaker 2: estimated that AI would increase annual global GDP by seven
• 0:42
  Speaker 2: percent over ten years. The IMF predicted that AI has
• 0:47
  Speaker 2: the potential to reshape the global economy. Look at pretty
• 0:51
  Speaker 2: much any report from any boog organization and you will
• 0:54
  Speaker 2: see something very similar, the idea that this will change everything.
• 0:59
  Speaker 2: Here in the the labor government clearly believe it too.
• 1:02
  Speaker 2: They spent a good part of the recent Investment Summit
• 1:05
  Speaker 2: positioning themselves as a champion for AI, called it an opportunity,
• 1:09
  Speaker 2: saying that the country needs to run towards it. Not
• 1:13
  Speaker 2: everyone is so convinced that AI is a silver bullet
• 1:16
  Speaker 2: for all our productivity. Rose and I'm afraid that we
• 1:18
  Speaker 2: are cynics on this podcast. We rather believe that too,
• 1:21
  Speaker 2: and so does this week's guest Diane Coyle Professor. Dame
• 1:24
  Speaker 2: Diane Coyle is the Bennett Professor of Public Policy at
• 1:27
  Speaker 2: the University of Cambridge.
• 1:28
  Speaker 3: The first book of hers I read.
• 1:30
  Speaker 2: Was GDP, A brief but affectionate History, and I often
• 1:33
  Speaker 2: refer to that even now. Her latest Worse Cogs and
• 1:37
  Speaker 2: Monsters What Economics is and what it should be? That
• 1:39
  Speaker 2: explored the challenges for economics in the context of the
• 1:42
  Speaker 2: digital transformation.
• 1:43
  Speaker 3: I particularly liked the title.
• 1:45
  Speaker 2: She's got a new book coming out in April called
• 1:47
  Speaker 2: The Measure of Progress, and maybe we'll get her back
• 1:49
  Speaker 2: on to talk about that then. For now, her current
• 1:52
  Speaker 2: research focus is on productivity and on economic measurements.
• 1:56
  Speaker 3: Earlier this year she.
• 1:56
  Speaker 2: Authored a very good article that caught our eye, Don't
• 1:59
  Speaker 2: bank on AI being a quick fix.
• 2:02
  Speaker 3: For elusive growth? Diane, welcome to Merin Talks Money.
• 2:09
  Speaker 4: It's a real pleasure.
• 2:10
  Speaker 2: And now one of the reasons that we asked you
• 2:12
  Speaker 2: to come on, apart from all these wonderul books and
• 2:13
  Speaker 2: writings that we've talked about, was an article that caught
• 2:16
  Speaker 2: our eye earlier this year called Don't bank on AI
• 2:19
  Speaker 2: being a quick fix for elusive growth? And rather like
• 2:23
  Speaker 2: a lot of other people on this podcast, we've talked
• 2:25
  Speaker 2: a lot about how everything's going to be absolutely fine
• 2:28
  Speaker 2: as soon as productivity pops up, and some somehow, one
• 2:31
  Speaker 2: way or another, technology and AI in particular, will do
• 2:34
  Speaker 2: that for us. So we will go from productivity growth
• 2:37
  Speaker 2: being as a miserable one or one and a half
• 2:39
  Speaker 2: percent across the Western economy is up to two and
• 2:41
  Speaker 2: a half maybe more, and all the problems will disappears
• 2:46
  Speaker 2: going to happen?
• 2:47
  Speaker 4: Is it? Well? Technologies do drive productivity growth over the
• 2:52
  Speaker 4: longer term, but they tend to be much slower than
• 2:56
  Speaker 4: people often imagine when there's a lot of new and
• 3:00
  Speaker 4: exciting news about what's coming along. And the classic example
• 3:04
  Speaker 4: from economic history is electricity, which is a nineteenth century
• 3:07
  Speaker 4: technology fundamentally, but it didn't affect productivity growth until at
• 3:11
  Speaker 4: least the nineteen thirties. And AI might be a little
• 3:14
  Speaker 4: bit quicker, but I think they're still looking at a
• 3:17
  Speaker 4: decade or so before it starts to have a measurable
• 3:20
  Speaker 4: impact on productivity, and there are lots of reasons why
• 3:23
  Speaker 4: that uptake is slow. So I get quite concerned about
• 3:28
  Speaker 4: the silver bullet claims that you referred to, because it
• 3:32
  Speaker 4: raises expectations and they get disappointed, and then you get
• 3:34
  Speaker 4: a sort of backlash against technology. AI is an amazing technology.
• 3:39
  Speaker 4: It will be able to do some extraordinary things for us,
• 3:42
  Speaker 4: but it's not going to fix productivity. Is ye.
• 3:44
  Speaker 2: Should we go back briefly to electricity, because that's very
• 3:47
  Speaker 2: interesting and is that? Was that an infrastructure problem? Because
• 3:51
  Speaker 2: obviously connecting everyone electricity is a huge infrastructure problem. Got
• 3:55
  Speaker 2: to be expensive, it's difficult, and you can't immediately see
• 3:58
  Speaker 2: the impact of it. So you get that presumably Jacob
• 4:01
  Speaker 2: effect right where you've got to spend all the money
• 4:03
  Speaker 2: up front and the productivity doesn't appear until everyone's connected,
• 4:06
  Speaker 2: and then it comes. So that's why it takes many decades,
• 4:09
  Speaker 2: did take many decades.
• 4:11
  Speaker 4: There's several reasons for the delay. Infrastructure is certainly one
• 4:13
  Speaker 4: of them. So with electricity, it was building out the networks.
• 4:17
  Speaker 4: With AI, it's building the data centers and the computational power,
• 4:22
  Speaker 4: but also actually getting the data in order because AI
• 4:26
  Speaker 4: eats data, that's its fuel, and a lot of the
• 4:30
  Speaker 4: data that we have isn't available, isn't interoperable, isn't in
• 4:34
  Speaker 4: a good shape, isn't high quality, And so there's a
• 4:38
  Speaker 4: big challenge in that sort of intangible infrastructure that we're
• 4:41
  Speaker 4: going to need for AI. But there are other investments
• 4:45
  Speaker 4: that are needed as well. In the electricity example, it
• 4:48
  Speaker 4: was building new kinds of factories because you can have
• 4:52
  Speaker 4: a dynamo on each machine, and the assembly line was
• 4:56
  Speaker 4: part of what made the electricity revolution create productivity gains.
• 5:02
  Speaker 4: But then also once you had to build new factories,
• 5:07
  Speaker 4: you needed to build new transportation networks to get workers
• 5:09
  Speaker 4: to them. And even then there's a reorganization of work
• 5:14
  Speaker 4: that's needed, so the different kinds of jobs in factories
• 5:17
  Speaker 4: and so on. So it's infrastructure directly involved, it's other
• 5:21
  Speaker 4: kinds of infrastructure and reshaping the economy. And then it's
• 5:24
  Speaker 4: what actually did people's jobs involved from what are the
• 5:26
  Speaker 4: skills that needed for that? So all of those things
• 5:29
  Speaker 4: need to come together, and that's why you get delayed.
• 5:32
  Speaker 2: Let's go to the optimistic end first, So go back
• 5:35
  Speaker 2: to that bit and talk about the main things that
• 5:37
  Speaker 2: people expect AI to do for us. So if you
• 5:40
  Speaker 2: divided into generative and predictive, it's really generative AI that
• 5:45
  Speaker 2: is the thing that we really expected productivity to gain
• 5:48
  Speaker 2: gains to come from.
• 5:49
  Speaker 3: Should they come.
• 5:50
  Speaker 4: Yes, and the models are still improving dramatically over time
• 5:56
  Speaker 4: according to the benchmarks that the industry itself uses to
• 6:00
  Speaker 4: tess that. So I've just come back from a week
• 6:03
  Speaker 4: in Silicon Valley, and every time I've been there in
• 6:06
  Speaker 4: the last couple of years, people have been saying, oh, well,
• 6:09
  Speaker 4: the thing that the next generation of models will do
• 6:11
  Speaker 4: for us is absolutely amazing, And then that has been true.
• 6:15
  Speaker 4: So I think technical advances are still extraordinary. But then
• 6:18
  Speaker 4: to think about the economic impact, you need to think
• 6:20
  Speaker 4: about what will that actually do in specific activities in
• 6:24
  Speaker 4: the economy. So we've got the language models, and they
• 6:28
  Speaker 4: are clearly able to do many kinds of administative task,
• 6:32
  Speaker 4: so they could create productivity that way. There are visualization models,
• 6:36
  Speaker 4: so let's think about what tasks involve visual inspection, or
• 6:41
  Speaker 4: where visualization would help improve productivity. There might be things
• 6:46
  Speaker 4: about three D real time visualization that are going to
• 6:49
  Speaker 4: become much more possible. So mapping from what the technology
• 6:52
  Speaker 4: can do in general to what are the actual things
• 6:56
  Speaker 4: in a workplace it might be able to do for
• 6:58
  Speaker 4: people is part of the journey.
• 7:00
  Speaker 2: I suppose what I'm really trying to ask is, as
• 7:02
  Speaker 2: an ordinary worker, an ordinary person, what is it that
• 7:07
  Speaker 2: we will see change?
• 7:08
  Speaker 3: What will happen around us? I mean, we know.
• 7:10
  Speaker 2: What's happened with digitalization and the rise of the smartphone,
• 7:14
  Speaker 2: all those kinds of things. We can see the difference
• 7:16
  Speaker 2: around us. We understand how that productivity changed, right. We
• 7:19
  Speaker 2: understand how the smartphone changed how we work. We understand
• 7:22
  Speaker 2: how our computers have changed things. We understand how cloud
• 7:25
  Speaker 2: computing has changed the way we work and operate, etc.
• 7:27
  Speaker 2: But as generative AI moves forward, what is it that
• 7:30
  Speaker 2: we will we will see and we will feel what
• 7:32
  Speaker 2: will be different for us?
• 7:34
  Speaker 4: Let's think about it in two ways. What products are
• 7:37
  Speaker 4: we going to experience as consumers that will make things
• 7:39
  Speaker 4: much better? And for there, I'd be looking for time
• 7:43
  Speaker 4: saving things, things that I don't like doing that AI
• 7:47
  Speaker 4: and robots might be able to do for me. Hans
• 7:49
  Speaker 4: Rostling a long time ago did a wonderful ted talk
• 7:52
  Speaker 4: about the impact of the washing machine on people's lives
• 7:54
  Speaker 4: and how extraordinary it's been. At the moment, AI can't
• 7:57
  Speaker 4: do that because it's not embedded in robotics. It's abstract,
• 8:01
  Speaker 4: it's not computers. Work is starting on that kind of
• 8:06
  Speaker 4: embodied AI and down the line, if that can do
• 8:10
  Speaker 4: my laundry for me with a super washing machine that
• 8:13
  Speaker 4: will also fold and iron clothes. That's going to be
• 8:15
  Speaker 4: something quite amazing in terms of daily life applications in medicine,
• 8:20
  Speaker 4: so a lot of opportunities for medical scanning, monitoring, tailored
• 8:27
  Speaker 4: personalized medications. Huge potential there to improve the multi of
• 8:31
  Speaker 4: people's lives. So that's one way to think about it,
• 8:34
  Speaker 4: and then the other way to think about it is
• 8:35
  Speaker 4: all the process innovations it can help us to introduce
• 8:38
  Speaker 4: at work that will save time do boring work much
• 8:41
  Speaker 4: more cheaply, and then they're free up the time for
• 8:45
  Speaker 4: workers to do things that are going to add more
• 8:47
  Speaker 4: value for their customers. So if you think about the
• 8:51
  Speaker 4: medical examples, well, it saved doctors time writing letters so
• 8:55
  Speaker 4: that they can spend more time with their patients. If
• 8:58
  Speaker 4: you work in manufacturing, is it going to speed up
• 9:01
  Speaker 4: those processes or help you track waste or energy use
• 9:05
  Speaker 4: in better ways than are currently available. And that's the
• 9:08
  Speaker 4: kind of process innovation that allowed the just in time
• 9:12
  Speaker 4: revolution and manufacturing productivity in the nineteen seventies and early eighties.
• 9:17
  Speaker 4: So thinking about it in those two ways will give
• 9:19
  Speaker 4: us some clues about where we can hope for ultimately
• 9:22
  Speaker 4: productivity gains.
• 9:24
  Speaker 2: My AI reverse the trying trend from a consumer's point
• 9:30
  Speaker 2: of view of having pretty much all the work that
• 9:32
  Speaker 2: used to be done by companies when you buy something,
• 9:35
  Speaker 2: or book something, or rain something being outsourced to you,
• 9:39
  Speaker 2: and the maddening difficulty that that caused with me, might
• 9:41
  Speaker 2: say is one of the big downsides of the product
• 9:44
  Speaker 2: of the technology revolutions.
• 9:45
  Speaker 4: So far, it might. It depends on the incentives of
• 9:50
  Speaker 4: the companies to do so, of course, but we've all
• 9:51
  Speaker 4: paid that time tax of getting into voicemail hell and
• 9:54
  Speaker 4: not being able to get talk to a human being.
• 9:56
  Speaker 2: Or chatbot and talking to a chat but I'm begging
• 9:59
  Speaker 2: it to let you talk to a person so you
• 10:00
  Speaker 2: can get to the end of the process.
• 10:02
  Speaker 3: But not happening.
• 10:03
  Speaker 4: So one model might be what's happened in retailing, or
• 10:06
  Speaker 4: is happening in retailing where you used to go to
• 10:08
  Speaker 4: the shop and somebody would scan your groceries and possibly
• 10:13
  Speaker 4: even put them in a bag for you back in
• 10:14
  Speaker 4: the old days, and increasingly that labor has been outsourced
• 10:18
  Speaker 4: to us, and so now we've got the automatic checkouts
• 10:20
  Speaker 4: where you do your own scanning and packing and you've
• 10:22
  Speaker 4: got to persuade the machine that you haven't put the
• 10:24
  Speaker 4: item in the bag the wrong way. But now there
• 10:27
  Speaker 4: are stores where you just take the stuff off the
• 10:29
  Speaker 4: shelf and go away with it, and so that then
• 10:32
  Speaker 4: would be labor saving. How we measured these in productivity
• 10:36
  Speaker 4: obviously has a different effect because the automatic checkouts and
• 10:40
  Speaker 4: the shops are going to increase the measured productivity of
• 10:44
  Speaker 4: the retailers because it's not measuring your unpaid labor in
• 10:47
  Speaker 4: doing that scanning and packing yourself. So we don't know
• 10:51
  Speaker 4: yet whether that AI enabled or you walk into the
• 10:54
  Speaker 4: store and take it out is going to last. Maybe not,
• 10:57
  Speaker 4: but that would be the kind of journey we might
• 10:58
  Speaker 4: hope for. Elsewhere, there are some call center experiments with
• 11:02
  Speaker 4: AI where it actually seems to be improving the experience
• 11:06
  Speaker 4: that consumers have because the AI can train call center
• 11:09
  Speaker 4: workers and gives them better scripts so they become better
• 11:12
  Speaker 4: at doing what they're doing. But I think you know
• 11:14
  Speaker 4: that's what we need to see, that's what's going to
• 11:16
  Speaker 4: help us.
• 11:17
  Speaker 2: Interesting, let's go back to what you just said about
• 11:19
  Speaker 2: the measurement of productivity, because it's a very interesting point
• 11:23
  Speaker 2: that these making us do the work improve the productivity
• 11:27
  Speaker 2: of the firm, and if you then take that work
• 11:30
  Speaker 2: back in house using AI, it may make no difference
• 11:33
  Speaker 2: at all to what looks like productivity. So the change
• 11:37
  Speaker 2: has happened, but it's not caught in the numbers. And
• 11:40
  Speaker 2: this is one of the things that you specialize in
• 11:41
  Speaker 2: writing about how the numbers do not count to the
• 11:44
  Speaker 2: change as digitalization and technology moves.
• 11:47
  Speaker 4: Toward, the numbers really entirely miss the digital revolution in
• 11:53
  Speaker 4: our experience as consumers and workers. So there are all
• 11:57
  Speaker 4: kinds of phenomenon that have been enabled by the digital
• 12:01
  Speaker 4: from global production networks that rely on communications and logistics
• 12:06
  Speaker 4: that have been enabled by digital to the way that
• 12:10
  Speaker 4: many manufacturing companies wrap services around their products it's called
• 12:13
  Speaker 4: servitization in the literature, to all the platform models and
• 12:17
  Speaker 4: these free goods, and how do we take account of
• 12:21
  Speaker 4: the zero price that we pay for the free goods?
• 12:23
  Speaker 4: But what are the other implications of all of that
• 12:27
  Speaker 4: just not captured in the economic statistics. It's always been
• 12:31
  Speaker 4: the case that it's really hard to measure the major
• 12:34
  Speaker 4: impacts of big innovations in statistics. The statistics are good
• 12:37
  Speaker 4: at small changes, but not big changes, And so economists
• 12:41
  Speaker 4: would also look at things like improved life expectancy or
• 12:46
  Speaker 4: the quality of life as well to supplement GDP figures.
• 12:50
  Speaker 4: But at the moment, this extraordinary change in how we
• 12:54
  Speaker 4: lead our life since the smartphone arrived in two thousand
• 12:57
  Speaker 4: and seven is pretty much invisible. And that's a lot
• 13:00
  Speaker 4: of what my next book is going to be about.
• 13:02
  Speaker 3: Okay, And what are the solution as to that? How
• 13:05
  Speaker 3: do we pick it up?
• 13:05
  Speaker 2: How do we change the way we measure I mean,
• 13:08
  Speaker 2: in a very minor way. We get very irritated on
• 13:10
  Speaker 2: this podcast a lot by GDP measurements because we're maddened
• 13:13
  Speaker 2: by people constantly talking about GDP instead of GDPP ahead.
• 13:17
  Speaker 2: But that's a very minor problem in the context of
• 13:20
  Speaker 2: what you're talking about. So how do you reframe the
• 13:22
  Speaker 2: measurement of economic progress?
• 13:25
  Speaker 4: Well, a lot of it is about collecting new data
• 13:27
  Speaker 4: and new statistics, and we've got to use the technology
• 13:29
  Speaker 4: itself to help us to do that. It is slowly starting.
• 13:33
  Speaker 4: So on things like global production chains, the impact of
• 13:39
  Speaker 4: a tariff on imports is now going to be different
• 13:43
  Speaker 4: than it would have been in the nineteen sixties because
• 13:45
  Speaker 4: to manufacture anything, you've got to import a lot of
• 13:47
  Speaker 4: the components, but the data are starting to capture what's
• 13:50
  Speaker 4: called value added in trade, how much do you need
• 13:53
  Speaker 4: to import to manufacture particular kind of export. So that's
• 13:57
  Speaker 4: slowly starting. But there's all kinds of new days to
• 14:01
  Speaker 4: collecting credit card data, using administrative data like taxes, using
• 14:06
  Speaker 4: mobile phone data, webscraping, and the research into how to
• 14:11
  Speaker 4: develop statistics using these new methods is kind of in
• 14:14
  Speaker 4: its early days. So I think we're probably in for
• 14:16
  Speaker 4: a I don't know, twenty year gap before we've got
• 14:19
  Speaker 4: a settled set of statistics on the digital economy.
• 14:23
  Speaker 2: Okay, So if we were in optimistic mode, might we
• 14:28
  Speaker 2: say that, in fact, living standards are improving much faster
• 14:32
  Speaker 2: than our current statistics tell us they are.
• 14:34
  Speaker 4: I think they're improving in some ways and for some people,
• 14:37
  Speaker 4: faster than statistics tell us, But in other ways they
• 14:41
  Speaker 4: are less good at telling us what's going on. The
• 14:44
  Speaker 4: obvious example people always give is that statistics don't tell
• 14:47
  Speaker 4: us anything about the environmental costs of recent economic growth,
• 14:51
  Speaker 4: which are going to hit productivity because places will get
• 14:56
  Speaker 4: flooded much more than they used to be. Agricultural productivity
• 14:59
  Speaker 4: will decl line because soil quality and biodiversity is declining,
• 15:03
  Speaker 4: so they will have real economic consequences, none of which
• 15:06
  Speaker 4: are measured. So there are pluses and minuses, and we
• 15:10
  Speaker 4: need really a broader framework to think about all of
• 15:13
  Speaker 4: these things that are outside the traditional standardized manufacturing economy
• 15:18
  Speaker 4: for which these statistics were devised in the nineteen forties.
• 15:21
  Speaker 3: Okay, that wasn't the answer I was looking for.
• 15:22
  Speaker 2: I was looking for you to tell me that the
• 15:24
  Speaker 2: statistics were hiding progress, as in effect, you're telling me
• 15:27
  Speaker 2: they're hiding a type of decline.
• 15:29
  Speaker 4: They're hiding both the hiding decline and progress. And if
• 15:33
  Speaker 4: you think even about things like medical innovations, something that's
• 15:38
  Speaker 4: now pretty basic, like the ability to do a really
• 15:41
  Speaker 4: straightforward cataract operational hip replacement operation, which has become a
• 15:46
  Speaker 4: much more standardized and straightforward process because of advances and technology.
• 15:51
  Speaker 4: They're not big, shiny advances like generative AI models, but
• 15:53
  Speaker 4: they're really important improvements in the quality of people's lives.
• 15:57
  Speaker 2: Well, let's move from that, just briefly, because no one's
• 16:00
  Speaker 2: well on this for too long, to the NHS and
• 16:02
  Speaker 2: the possibility the possibility that there are ways that AI
• 16:06
  Speaker 2: might be able to help us with the NHS in
• 16:10
  Speaker 2: a way that constant large lugs of cash do not.
• 16:13
  Speaker 4: Yes, So it's obviously one of the hopes for AI
• 16:18
  Speaker 4: that it will enable big improvements in public service productivity
• 16:22
  Speaker 4: and particularly in the NHS, And we already talked a
• 16:26
  Speaker 4: little bit about some of the potential there. There are
• 16:30
  Speaker 4: already quite well established gains to be made from having
• 16:34
  Speaker 4: AI screen test results and provide that information to doctors
• 16:40
  Speaker 4: or making decisions, a lot of potential for administrative simplicity.
• 16:45
  Speaker 4: But to me, it's a question of organizational than it
• 16:48
  Speaker 4: is of technology. It boggles my mind that it isn't
• 16:52
  Speaker 4: already the case that the NHS has not mandated all
• 16:56
  Speaker 4: of the hospital trust and GP surgeries to use interoper
• 17:01
  Speaker 4: technology and standardized forms of data. So even getting to
• 17:05
  Speaker 4: the basics of how easy is it going to be
• 17:09
  Speaker 4: to adopt AI, we're not there yet. And it is
• 17:13
  Speaker 4: clear that there's been underin investment in very basic technology.
• 17:17
  Speaker 4: My husband was in hospital last year and we noticed
• 17:20
  Speaker 4: that the nurses had to spend all of their time
• 17:22
  Speaker 4: with their backs to patients, typing things into computers with
• 17:25
  Speaker 4: these really user unfriendly kinds of interfaces that were very
• 17:29
  Speaker 4: out of date. So is there potential for AI to
• 17:32
  Speaker 4: fix that, sure, but that's going to need some investment
• 17:36
  Speaker 4: of money, but also time to restructure those information flows.
• 17:39
  Speaker 4: Within the NHS. There's an issue about data and data sharing,
• 17:45
  Speaker 4: and there is real conservatism and risk aversion across the
• 17:49
  Speaker 4: NHS and linking up people's data. Recent announcements about the
• 17:53
  Speaker 4: NHS the NHS app bringing that together for individual patients.
• 17:57
  Speaker 4: That's really welcome, but I think that fundamental lack of
• 18:00
  Speaker 4: trust in the possibility of data sharing. How secure will
• 18:04
  Speaker 4: it be? Who's going to profit from that? Will it
• 18:06
  Speaker 4: be private sector companies or will you? Will patients see
• 18:09
  Speaker 4: some benefit? That's a really big hurdle to overcome yea.
• 18:13
  Speaker 2: And not helped by the way that public trust in
• 18:15
  Speaker 2: the NHS has fallen very dramatically over four or five years. Yeah,
• 18:19
  Speaker 2: last four or five years, So how would one then
• 18:21
  Speaker 2: trust them further?
• 18:22
  Speaker 4: Absolutely? And then there are questions about organization and financing.
• 18:27
  Speaker 4: So there's lots of innovation in health tech. People are
• 18:29
  Speaker 4: coming up with some amazing new products, but they've got
• 18:32
  Speaker 4: to be paid for. So what's the business model that
• 18:35
  Speaker 4: will enable the NHS to invest in those kinds of
• 18:38
  Speaker 4: new innovations that might save them a ton of money,
• 18:41
  Speaker 4: but their budgeting structures don't easily allow that, and then
• 18:45
  Speaker 4: who's got the authority to make decisions? So AI is
• 18:47
  Speaker 4: an information technology. It uses data and it creates useful
• 18:51
  Speaker 4: information which will allow people to make decisions that can
• 18:55
  Speaker 4: improve productivity. But that's going to be decentralized across the
• 19:00
  Speaker 4: public services and including in health. They're quite hierarchical for
• 19:04
  Speaker 4: reasons of accountability and also training. So how can the
• 19:09
  Speaker 4: skills needed and the authority needed to take decisions be delegated.
• 19:13
  Speaker 4: So there's no point giving nurses amazing information through AI
• 19:17
  Speaker 4: products if they can't then do something with it. If
• 19:20
  Speaker 4: they still have to call up doctor to get them
• 19:22
  Speaker 4: to take the decision, they're still going to be bottlenecked there.
• 19:25
  Speaker 4: So there's a whole range of barriers alongside this huge potential,
• 19:29
  Speaker 4: and obviously the state of the NHS and the public
• 19:32
  Speaker 4: confidence and it means this is one that we've got
• 19:34
  Speaker 4: to crack as sin as possible.
• 19:36
  Speaker 2: I imagine briefly, when I was introducing the current government
• 19:40
  Speaker 2: talking a lot about wanting the UK to be an
• 19:42
  Speaker 2: AI leader and how important that is and how they
• 19:45
  Speaker 2: can help the state can help drive that. What do
• 19:48
  Speaker 2: you think the state's role is here in driving investment
• 19:51
  Speaker 2: into AI or improving the sector in the UK.
• 19:54
  Speaker 4: So multiple roles. Actually, we do have a really good
• 19:58
  Speaker 4: AI sector at the research frontier and in parts of AI,
• 20:02
  Speaker 4: so the government has funded a lot of that research
• 20:04
  Speaker 4: and should carry on doing so. Except it's got broad
• 20:07
  Speaker 4: benefits and want to enable more startups. There's an issue
• 20:10
  Speaker 4: about how do those companies grow. Many startups in tech
• 20:14
  Speaker 4: still hope that they would be brought out by a
• 20:16
  Speaker 4: big American company, and we need to be able to
• 20:18
  Speaker 4: grow our own and understand which niches we have real
• 20:23
  Speaker 4: advantage in and can export in. So that's again something
• 20:26
  Speaker 4: that government needs to help with because public markets are
• 20:30
  Speaker 4: making it very difficult for companies to grow across the board.
• 20:33
  Speaker 4: But then there's a whole area of things about coordinating
• 20:37
  Speaker 4: activities de risking investments. Is, for example, using public procurement
• 20:41
  Speaker 4: in the NHS to encourage innovation and say there will
• 20:46
  Speaker 4: be a market for new kinds of AI products. That
• 20:49
  Speaker 4: it doesn't have to specify which ones are going to
• 20:52
  Speaker 4: be brought or exactly which technological approaches needed, but the
• 20:56
  Speaker 4: fact of it's called an advanced market commitment. In my world,
• 21:00
  Speaker 4: and the fact that there will be a market through
• 21:02
  Speaker 4: public procurement can also really help encourage us better technology
• 21:06
  Speaker 4: and you know, public services can drive the take up
• 21:10
  Speaker 4: of technology, get things to scale, make it cheaper for
• 21:13
  Speaker 4: other users, demonstrate effectiveness. So this is an area where
• 21:18
  Speaker 4: I think government action can make a big difference as
• 21:21
  Speaker 4: long as it's realistic. So I think if the government
• 21:24
  Speaker 4: to say we have this silver bullet and it's going
• 21:26
  Speaker 4: to change things overnight, that that's very dangerous and they'll
• 21:29
  Speaker 4: be backlash. But if it's got a realistic approach in
• 21:34
  Speaker 4: you know, strategic industrial policy kind of framework, that could
• 21:36
  Speaker 4: be really powerful.
• 21:38
  Speaker 2: Okay, interesting, and what about the concerns around competition and
• 21:43
  Speaker 2: the idea that what with the you know, the amount
• 21:46
  Speaker 2: of investment, the amount of data.
• 21:48
  Speaker 3: Except with that AI.
• 21:49
  Speaker 2: Progress required that you have a situation where a big
• 21:52
  Speaker 2: tech this turns into big AI and we have the
• 21:55
  Speaker 2: same problems that we've anti competitiveness problems that we've had
• 21:58
  Speaker 2: with big tech and the aim as well, where's the
• 22:01
  Speaker 2: government's were all there to preempt that.
• 22:03
  Speaker 4: So it's a big issue. We've now got the Digital
• 22:06
  Speaker 4: Markets legislation that allows the CMA to take action around
• 22:11
  Speaker 4: certain kinds of behavior by big tech companies that have
• 22:14
  Speaker 4: got a dominant position. If you're looking at generative AI
• 22:17
  Speaker 4: and the large models, that the scale needed at the
• 22:20
  Speaker 4: moment is ginormous, so that problem of dominant domination of
• 22:25
  Speaker 4: markets becomes even worse. So more of the same in
• 22:29
  Speaker 4: terms of monitoring what's happening and looking out for things
• 22:34
  Speaker 4: like algorithmic collusion between the different models remains important. There's
• 22:41
  Speaker 4: a lot of scope for competition at the level of
• 22:43
  Speaker 4: applications that use large models, and that's a nascent market
• 22:49
  Speaker 4: with generitive AI, so there's a lot more scope to
• 22:51
  Speaker 4: make sure that is competitive.
• 22:53
  Speaker 2: The other thing that I think is quite important to
• 22:55
  Speaker 2: talk about is the energy hunger of AI. And you
• 22:58
  Speaker 2: mentioned earlier that day is the fuel of AI, but
• 23:01
  Speaker 2: of course that the fuel of AI is fuel. And
• 23:03
  Speaker 2: you also we talked earlier about the externalities and bringing
• 23:06
  Speaker 2: those into the equation. So if you start to look
• 23:09
  Speaker 2: at the productivity gains that might come from AI, you
• 23:12
  Speaker 2: kind of have to bring all that into the equation.
• 23:15
  Speaker 4: Absolutely, And among the most extraordinary stories recently have been
• 23:18
  Speaker 4: Microsoft reopening the three Mile Island Nuclear plant and Google
• 23:21
  Speaker 4: commissioning a whole fleet of small modular nuclear reactors. So
• 23:26
  Speaker 4: I've seen charts where the prediction is that the use
• 23:31
  Speaker 4: of AI as it grows will end up using all
• 23:33
  Speaker 4: of the electricity generated in the United States, which clearly
• 23:36
  Speaker 4: is not going to happen. So the energy supply might increase.
• 23:40
  Speaker 4: There'll be a lot of incentive for people to come
• 23:43
  Speaker 4: up with much more energy efficient types of models and
• 23:46
  Speaker 4: AI processes, and that is also starting to happen. But
• 23:50
  Speaker 4: otherwise it's a real constraint on the use of AI
• 23:53
  Speaker 4: because this might not be the most efficient way we
• 23:56
  Speaker 4: want to use our energy. There are lots of other uses.
• 23:58
  Speaker 4: And when you read stories about development along the End
• 24:01
  Speaker 4: four corridor not being possible because there's no scope for
• 24:05
  Speaker 4: new grid connections due to all the data centers, then
• 24:07
  Speaker 4: you start thinking, well, perhaps there'll be political reprioritization here.
• 24:11
  Speaker 4: AI is not going to sweep all before it, so
• 24:14
  Speaker 4: we certainly start need to start thinking about the energy efficiency.
• 24:18
  Speaker 4: I don't like the way that Google, when you do
• 24:21
  Speaker 4: a search now gives you an AI summary at the top,
• 24:24
  Speaker 4: because that means you're getting the same information as before,
• 24:27
  Speaker 4: but with turn times the energy.
• 24:28
  Speaker 3: Use you know, I hate that too. I've only noticed
• 24:31
  Speaker 3: it relatively recently.
• 24:32
  Speaker 2: It comes up and you look at it and you
• 24:34
  Speaker 2: know that that bit at the top that you didn't
• 24:35
  Speaker 2: ask for is using. I'm not entirely sure what the
• 24:38
  Speaker 2: figures are, but people keep telling me that a generative
• 24:41
  Speaker 2: search takes a thousand times the energy of an ordinary search.
• 24:44
  Speaker 2: As you look at that and you know you can
• 24:46
  Speaker 2: feel the energy uses, that doesn't feel good.
• 24:48
  Speaker 4: It's not good. And so one, you know, really key
• 24:52
  Speaker 4: imperative is keeping an eye on that energy use. And
• 24:56
  Speaker 4: I think governments will actually face choices about about their
• 25:00
  Speaker 4: energy grids, which are you know, public, key public policy decisions.
• 25:05
  Speaker 2: You've seen that in some of the EU comments already
• 25:07
  Speaker 2: on AI have been looking at the energy factor. But
• 25:11
  Speaker 2: I suppose the selver aligning, if you're that way inclined
• 25:14
  Speaker 2: to think it's the self aligning. Is this resurgence of
• 25:17
  Speaker 2: nuclear energy and the drive forward with SMRs et cetera.
• 25:20
  Speaker 2: That seems like a good thing.
• 25:21
  Speaker 4: Seems like a good thing. And obviously nuclear is controversial
• 25:24
  Speaker 4: with some people, but in a fully renewable world will
• 25:28
  Speaker 4: need some baseline generation which nuclear can provide a zero emissions.
• 25:32
  Speaker 2: And if the very rich, very cash rich tech companies
• 25:36
  Speaker 2: are prepared to be the ones who go first and
• 25:39
  Speaker 2: finance some of the research into this, and that that
• 25:41
  Speaker 2: seems like a finally a positive of the vast amount
• 25:44
  Speaker 2: of cash that they've earned over the last decade.
• 25:46
  Speaker 4: It does.
• 25:46
  Speaker 3: Yes, Yeah, we've talked.
• 25:48
  Speaker 2: A lot about things that AI might be able to
• 25:50
  Speaker 2: do that it could do. What we haven't talked about
• 25:52
  Speaker 2: is things that people believe AI can do, which it can't.
• 25:57
  Speaker 2: What is it that we might be expecting that is
• 26:00
  Speaker 2: very unlike.
• 26:01
  Speaker 4: That's a good question. What do you have in mind?
• 26:03
  Speaker 3: I do know it if you're particular in mind.
• 26:05
  Speaker 2: I just every time someone talks about AI and they
• 26:07
  Speaker 2: talk about, well, it can do this, where it can
• 26:09
  Speaker 2: do that. It can change the NHS, it can redrive productivity,
• 26:12
  Speaker 2: it can change the way we educate people. It can
• 26:15
  Speaker 2: change the way we manage the time. It will change
• 26:18
  Speaker 2: It'll completely change the way we run our agricultural economy,
• 26:20
  Speaker 2: for example. You know, you hear nothing but transform, transformation
• 26:25
  Speaker 2: or optimism. And because I'm not an expert in this,
• 26:29
  Speaker 2: when I hear people telling me it's going to change
• 26:31
  Speaker 2: the way we farm, it's going to change the way
• 26:32
  Speaker 2: we go to space, it's going to change the way
• 26:34
  Speaker 2: we mind mind for metals on the moon, etc. I'm
• 26:37
  Speaker 2: perfectly happy to believe all that because this is not
• 26:40
  Speaker 2: my area of expertise, and it seems entirely possible that
• 26:43
  Speaker 2: if you can manage data this effectively, and if you
• 26:46
  Speaker 2: can have models that can that can genuinely learn to
• 26:50
  Speaker 2: do things that we haven't yet learned to do, then
• 26:52
  Speaker 2: all these things could be possible. But I also know,
• 26:55
  Speaker 2: because you fall victim to optimism all the time, that
• 26:58
  Speaker 2: inside this belief that it can do anything, there's bound
• 27:01
  Speaker 2: to be things that definitely can't do well.
• 27:04
  Speaker 4: It might do all of these things if you again
• 27:06
  Speaker 4: look back to those historical examples. The railways enabled urbanization
• 27:11
  Speaker 4: because they meant that cows didn't need to live in
• 27:14
  Speaker 4: the city centers to provide fresh milk. They could live
• 27:16
  Speaker 4: out in the countryside, and the railways the milk train
• 27:18
  Speaker 4: would bring in the fresh milk. Who would have predicted
• 27:21
  Speaker 4: that when the first railways were built. So, yes, the
• 27:24
  Speaker 4: tech might do all kinds of amazing things. The question
• 27:26
  Speaker 4: I always ask when I hear the hype is how
• 27:30
  Speaker 4: is it going to do that? Given that this is
• 27:32
  Speaker 4: an information technology, So what you need to talk about
• 27:36
  Speaker 4: is how does having a different, better, faster flow of information,
• 27:42
  Speaker 4: change what we do to produce the amazing outcomes that
• 27:46
  Speaker 4: you're talking about, and a lot of the hype merchants
• 27:49
  Speaker 4: actually can't answer the how question.
• 27:52
  Speaker 2: Well, it's one of those things we will gradually find out,
• 27:54
  Speaker 2: won't we.
• 27:55
  Speaker 4: Yeah, And I mean as always that one of the
• 27:58
  Speaker 4: main barriers is people don't like changing what they do.
• 28:01
  Speaker 4: So anything that requires fundamental change and now you do
• 28:05
  Speaker 4: your job with leadir life can can be very slow.
• 28:08
  Speaker 4: It's partly fear because the kind of upfront cost of
• 28:12
  Speaker 4: using AI is so high. Still, there are a lot
• 28:14
  Speaker 4: of things for which it's still much cheaper to use
• 28:18
  Speaker 4: the traditional forms of production and traditional human labor. And
• 28:21
  Speaker 4: so even if there are amazing changes down the road,
• 28:24
  Speaker 4: as we were saying earlier, it be pretty slow.
• 28:27
  Speaker 3: Do you worry about the job market? Do you worry that?
• 28:29
  Speaker 2: I mean, obviously, as you say, it will be slow.
• 28:32
  Speaker 2: But in ten years will we have an unemployment problem
• 28:35
  Speaker 2: or will the economy do it? It always does simply
• 28:37
  Speaker 2: create you in different types of jobs, always has done so.
• 28:41
  Speaker 4: In klind towards optimism in terms of number of jobs.
• 28:44
  Speaker 4: I don't think this will be a sudden change in
• 28:47
  Speaker 4: the way that the wave of de industrialization in the
• 28:49
  Speaker 4: late seventies and early eighties was, which obviously policy could
• 28:53
  Speaker 4: not cope with. That are just too many people became
• 28:55
  Speaker 4: unemployed too quickly, So I think this will be more
• 28:58
  Speaker 4: gradual and ours adjustment in the labor market. I guess
• 29:02
  Speaker 4: my bigger concern is about the quality of work and
• 29:05
  Speaker 4: what people get paid. Because the benchmarks for AI are
• 29:09
  Speaker 4: set in terms of how well does it match what
• 29:13
  Speaker 4: humans do. That creates a kind of bias towards substituting
• 29:17
  Speaker 4: for humans doing what they do at lower cost, and
• 29:22
  Speaker 4: instead it would be better to think about benchmarks that
• 29:26
  Speaker 4: improve outcomes that both humans and AI can deliver together.
• 29:32
  Speaker 4: But if we go down the substituting for humans route,
• 29:34
  Speaker 4: then the impact on wages and job quality would be
• 29:38
  Speaker 4: something to worry about.
• 29:39
  Speaker 5: What's your greatest hope for AI when you look at
• 29:42
  Speaker 5: ten years, What's the thing that will have been most
• 29:44
  Speaker 5: transformative that will change Not things we can't say, the
• 29:47
  Speaker 5: things think we can see, things we can feel.
• 29:50
  Speaker 4: I would hope that AI will have saved us a
• 29:53
  Speaker 4: lot of time doing things that we don't like doing,
• 29:56
  Speaker 4: to create time for us to do things that we
• 29:58
  Speaker 4: really enjoy. So that this might be we've got more
• 30:03
  Speaker 4: leisure time because technologies have, over history increase the amount
• 30:07
  Speaker 4: of leisure and reduce the amount of work that we
• 30:10
  Speaker 4: have to do. Or it might be that our work
• 30:13
  Speaker 4: becomes more enjoyable we can focus on the more satisfying
• 30:15
  Speaker 4: parts of that. So if I were a medical person,
• 30:19
  Speaker 4: I would hope that AI had taken over all of
• 30:22
  Speaker 4: my ADMIN and left me with shorter shifts but more
• 30:27
  Speaker 4: time to spend with patients and give them the quality
• 30:30
  Speaker 4: of care that I would want to give them.
• 30:31
  Speaker 3: Goodness, I tell you, well, we'd all like that.
• 30:34
  Speaker 2: Diane, thank you so much for joining ask It.
• 30:36
  Speaker 4: I'm fascinating, Thank you, Thank you lovely to talk to you.
• 30:44
  Speaker 2: Thanks for listening to this week's Mirin Talks Money.
• 30:46
  Speaker 3: If you like our show, rate to review.
• 30:47
  Speaker 2: And subscribe wherever you listen to the podcasts and keep
• 30:50
  Speaker 2: sending questions or comments to.
• 30:51
  Speaker 3: Merror Money at Bloomberg dot net. You can also follow
• 30:54
  Speaker 3: me and John or Twitter or X.
• 30:56
  Speaker 2: I'm at Marinus w and John is John Underscore X.
• 31:00
  Speaker 2: This episode was hosted by me marenzum zet Web. It
• 31:03
  Speaker 2: was produced by Someasadi production support and sound designed by
• 31:06
  Speaker 2: Moses and m and special thanks of course to Diane
• 31:09
  Speaker 2: Coyle