Latest / Trumponomics / Why AI Isn't Actually Boosting Productivity

Transcript

Why AI Isn't Actually Boosting Productivity · Trumponomics

0:02
Speaker 1: Bloomberg Audio Studios, Podcasts, radio News.
0:17
Speaker 2: I'm Stephanie Flanders, head of Government and Economics at Bloomberg,
0:21
Speaker 2: and this is Trumpnomics, the podcast that looks at everything
0:24
Speaker 2: in the economic world of Donald Trump. Today. I wanted
0:28
Speaker 2: to dig into two seemingly conflicting stories about the global
0:33
Speaker 2: economy that are often embedded in the conversations on this show.
0:36
Speaker 2: I mean, one story that we hear often is that
0:39
Speaker 2: we live in an era of extraordinary innovation, new more
0:43
Speaker 2: powerful AI models unveiled almost daily, and the US and
0:47
Speaker 2: China seemingly neck and neck in the race to dominate
0:51
Speaker 2: that transformative technology and reap the benefits. I mean, the
0:55
Speaker 2: challenge in that world is not that we have too
0:57
Speaker 2: little innovation, but too much, and how we got absorb
1:00
Speaker 2: it into the economy without blowing up everything that we
1:03
Speaker 2: hold deer. But another story, which we heard more often
1:08
Speaker 2: after the global financial crisis but has definitely not gone away,
1:11
Speaker 2: as one of stagnation of declining competition in large chunks
1:17
Speaker 2: of the economy, more and more meddling by governments. And
1:20
Speaker 2: of course we've heard that, particularly in Europe. But I
1:22
Speaker 2: think if you think about the rise of monopoly power
1:26
Speaker 2: more and more tariffs getting in the way of global trade.
1:29
Speaker 2: I mean, we've certainly seen all of that in Trump's
1:31
Speaker 2: American economy, and not to mention the more direct involvement
1:36
Speaker 2: in the affairs of business that the president and his
1:39
Speaker 2: cabinet have engaged in. So I was interested to try
1:43
Speaker 2: and tease out some of these themes and try and
1:45
Speaker 2: get a sense of are we in a very innovative
1:47
Speaker 2: era that means we're now about to have loads of
1:51
Speaker 2: growth or actually potentially running the risk of something more negative.
1:57
Speaker 2: I wanted to discuss it with Carl Benedict Free, who's
2:00
Speaker 2: the author of a book that came out last year Actually,
2:02
Speaker 2: How Progress Ends, Technology, Innovation, and the Face of Nations.
2:08
Speaker 2: He's also an associate professor at Oxford University and often
2:11
Speaker 2: writes for the Ft and other places. Car, thank you
2:14
Speaker 2: very much for joining Trump andomics.
2:17
Speaker 1: Good to be with you.
2:25
Speaker 2: Before we get into this, I think it's probably helpful
2:27
Speaker 2: for people to understand what the argument of your book is,
2:31
Speaker 2: and specifically your thesis around how technological and economic progress
2:38
Speaker 2: happen and the patterns that countries have tended to go through.
2:42
Speaker 1: So a key theme of the book is that the
2:44
Speaker 1: institutions need to just as technology moves along, right, and
2:47
Speaker 1: so you can grow for for a long period of
2:51
Speaker 1: time just by adopting a scaling technology invented elsewhere. And
2:55
Speaker 1: so the Soviet Union did that quite successfully over four decades.
2:59
Speaker 1: It took ample advantage of the Fortomotive Company's open door
3:03
Speaker 1: policy and transfer that knowledge to the Soviet Union build
3:09
Speaker 1: a vehicles industry of its own. And in the age
3:12
Speaker 1: of mass production, the Soviet system actually worked fairly well
3:16
Speaker 1: because you know, when technology is mature, when production is
3:20
Speaker 1: fairly standardized, then you can hold factory managers accountable essentially
3:26
Speaker 1: just by benchmarking performance. But as the returns to the
3:31
Speaker 1: mass production system petered out globally in the nineteen seventies
3:36
Speaker 1: and something new was needed for growth, that system did
3:39
Speaker 1: no longer work particularly well because when something is novel,
3:44
Speaker 1: when you're dealing with new technology, well, first of all,
3:46
Speaker 1: how do you benchmark performance? How do you hold factory
3:50
Speaker 1: managers accountable? And secondly, you know, that new thing that
3:54
Speaker 1: was needed for further growth was the computer revolution. To
3:58
Speaker 1: each Soviet contributeus were essentially non Why is that well,
4:03
Speaker 1: A big reason is that the Soviet Union provided very
4:07
Speaker 1: little room for decentralized exploration. And so if you were
4:11
Speaker 1: an engineer in the Soviet Union, you could go to
4:13
Speaker 1: the Red Arm and ask for funding. If they declined,
4:16
Speaker 1: well maybe two or three other options. If they declined,
4:20
Speaker 1: then your idea would die with you. And that's quite
4:23
Speaker 1: different from the American system of more decentralized finance, where
4:28
Speaker 1: you know, Besma Venture famously the client to invest in
4:31
Speaker 1: Google back in nineteen ninety nine. They probably regretted today,
4:34
Speaker 1: but it also illustrates that Google was not a safe
4:38
Speaker 1: bet at the time. Alta Vista and Yahoo they were
4:41
Speaker 1: dominating search, and so somebody you know, needed to invest
4:48
Speaker 1: to show that Google would actually catch on. And the
4:52
Speaker 1: fact that Besman didn't invest, though, didn't mean the end
4:54
Speaker 1: of Google because others stepped in. And so when you
4:58
Speaker 1: explore new technology, you need more people exploring different potential
5:04
Speaker 1: technological trajectories. So that's the exploration phase. But once you've
5:09
Speaker 1: settled on the product type, then you need to scale that.
5:12
Speaker 1: And that scaling in the Soviet Union did reasonably well.
5:17
Speaker 1: But when that runs into diminishing, it turns you need
5:20
Speaker 1: to get onto the new cycle. So you need to
5:22
Speaker 1: move from more centralized consolidation to more decentralized system again,
5:28
Speaker 1: and it's those shifts in institutions that make progress very
5:33
Speaker 1: hard to sustain.
5:35
Speaker 2: If you just look at the title of your book first,
5:38
Speaker 2: anyone looking at that, particularly this year, how progress ends
5:42
Speaker 2: technology innovation in the face of nations, you'd have to think, well,
5:46
Speaker 2: however it ends, it doesn't seem to be ending at
5:48
Speaker 2: the moment. So what's your take on the world we're
5:53
Speaker 2: looking at now? Are we at the end of progress
5:55
Speaker 2: or just the beginning of amazing progress?
5:57
Speaker 1: Well, the title of the book is not men to
6:00
Speaker 1: suggesting that progress is inevitably about to end. It's more
6:04
Speaker 1: a reflection of the fact that progress is unnatural. Right,
6:07
Speaker 1: If progress was inevitable, it would not have taken two
6:10
Speaker 1: hundred thousand years to having an industrial revolution. If progress
6:14
Speaker 1: was inevitable, the most of the world would be rich
6:16
Speaker 1: and prosperous. If progress was inevitable, Britain, the country where
6:20
Speaker 1: I live, would not have suffered two decades of productivity detegnation.
6:24
Speaker 1: So progress is clearly not inevitable, and it's clearly not
6:27
Speaker 1: inevitable even despite the acceleration we've seen in innovation. That dimension, right,
6:33
Speaker 1: So if you look at patenting, if you look at
6:35
Speaker 1: scientific publications, any measure of inventive output, all those indicators
6:42
Speaker 1: are up. Yet if you look at the economy as
6:45
Speaker 1: pointing in a very different direction, and even measures of
6:48
Speaker 1: research productivity and breaking through innovation are down. So we're
6:52
Speaker 1: getting more in terms of scientific and inventive output, but
6:58
Speaker 1: it seems that we're getting less transformational output. And I
7:02
Speaker 1: think that's important to keep in mind when we're looking
7:05
Speaker 1: at the potential impact of the economy on artificial intelligence
7:11
Speaker 1: as well, because in many ways the computer revolution was
7:14
Speaker 1: more transformative than the AI we have today. Right The
7:18
Speaker 1: computer and the Internet connected the best scientists and inventors
7:22
Speaker 1: around the world. It streamlined the research the process enormously.
7:26
Speaker 1: It gave us access to the weld store knowledge essentially
7:30
Speaker 1: in our pockets. And what do we get out of that?
7:34
Speaker 1: Basically a decade long productivity upsurge mostly confined to the
7:42
Speaker 1: United States. Now. I do think that AI will show
7:46
Speaker 1: up in the productivity statistics eventually, but the question is
7:51
Speaker 1: by how much and for how long? And so you know,
7:54
Speaker 1: any productivity upsurge, you'll take it. But I think if
7:59
Speaker 1: you believe that we are entering a new renaissance for
8:02
Speaker 1: economic growth. You're likely to be mistaken because AI is
8:09
Speaker 1: actually likely to give less of a productivity boost even
8:14
Speaker 1: the computer revolution. And the reason is this, with AI,
8:17
Speaker 1: still need verification. So AI automates, you know, a lot
8:22
Speaker 1: of knowledge work, but in the end of that, you
8:24
Speaker 1: need to verify the outputs, and so it's the time
8:27
Speaker 1: saving minus the time for verification. The computer revolution was different, right,
8:33
Speaker 1: you know, I could sit around and wait for hours, days,
8:37
Speaker 1: even weeks for new material to arrive for me to
8:40
Speaker 1: start my research, and the Internet gave me access to
8:44
Speaker 1: that instantaneously, and so it's automated downtime. AI is not
8:47
Speaker 1: automating that downtime. It's automating the production, and you still
8:51
Speaker 1: need a verification as well.
8:53
Speaker 2: I'm sure we'll get onto some of the implications of
8:56
Speaker 2: AI and also sort of the bearing of some of
8:59
Speaker 2: your research on that, because you sort of look back
9:01
Speaker 2: at thousands of years of what has produced innovation and
9:04
Speaker 2: then what's then translated that into growth. But just what
9:08
Speaker 2: you said about that sort of disconnect around, we've seen
9:12
Speaker 2: more invention at some level, lots more invention, lots more
9:15
Speaker 2: patenting and everything else, and yet seemingly at the micro level,
9:19
Speaker 2: at the real level of the economy, still quite a
9:22
Speaker 2: lot of low productivity and lack of translation of that
9:27
Speaker 2: technology into productivity. And thinking about that contrast between the
9:32
Speaker 2: US and China, there was a story that was told
9:35
Speaker 2: for many years that probably is still being told about
9:38
Speaker 2: the innate advantages of the US in technological progress that
9:44
Speaker 2: was always around. You know, China might be very good
9:47
Speaker 2: at copying things, but when it comes to actual innovation
9:51
Speaker 2: at the frontier, the US is always going to have
9:55
Speaker 2: an advantage over a sort of centralized system China. There's
9:59
Speaker 2: quite a lot support for that in your book, But
10:03
Speaker 2: I guess if you look all around, you would have
10:05
Speaker 2: to sort of say, well, maybe Ai is going to
10:07
Speaker 2: be the exception, because they seem to be getting pretty
10:09
Speaker 2: close to the frontier on AI without having the kind
10:13
Speaker 2: of decentralized system that the US has.
10:16
Speaker 1: Well, I think, first of all, though clearly is innovation
10:18
Speaker 1: in China, But China is also a country of one
10:21
Speaker 1: point four billion people, so it would be absolutely extraordinary
10:25
Speaker 1: if there was no innovation going on in China whatsoever.
10:30
Speaker 1: But in addition to that, though China is not as
10:33
Speaker 1: centralized as is commonly believed, and so I like the
10:38
Speaker 1: Soviet Union, which also had one party system, where every
10:42
Speaker 1: industry was managed centrally from Moscow, whether it was railroads
10:46
Speaker 1: or steel. The Chinese economy is much more decentralized, and
10:52
Speaker 1: provinces and provincial governors and mayors have much greater autonomy
10:58
Speaker 1: than the counterparts would have had. And so what you
11:03
Speaker 1: have in China system where provincial governors are competing against
11:11
Speaker 1: each other based on often growth targets for promotion inside
11:17
Speaker 1: the one party system, and so what that creates is
11:21
Speaker 1: essentially a tournament of political competition that is very hard
11:28
Speaker 1: to replicate anywhere else. And so Europe does industrial policy,
11:33
Speaker 1: it often tends to be anti competitive, right where you
11:36
Speaker 1: German rearmament essentially means plowing funds into Rhymtal. In China,
11:42
Speaker 1: on the other hand, industrial policy can often be pro
11:45
Speaker 1: competitive because you have provinces seeding new firms that are
11:49
Speaker 1: competing against each other, and so I think that's often
11:54
Speaker 1: underappreciated part of the Chinese economy. That said, though, when
11:59
Speaker 1: you look get you know the firms which are leading
12:03
Speaker 1: in innovation in China, it's mostly startups, it's mostly privately
12:10
Speaker 1: funded firms, it's often foreign funded firms as well, and
12:15
Speaker 1: so in that sense, China is not that different from
12:19
Speaker 1: either Europe and the United States. What is different is
12:23
Speaker 1: that it doesn't have the rule of law, and so
12:26
Speaker 1: building political connections is more important in China because you
12:31
Speaker 1: need to have a seat of the table when priorities changed,
12:34
Speaker 1: and priorities changed from time to time. But perhaps somewhat paradoxically,
12:38
Speaker 1: what people believed in the two thousands that China would
12:41
Speaker 1: could become more like the United States, the opposite seems
12:44
Speaker 1: to be happening, and the United States today looks more
12:48
Speaker 1: like the political capitalism and that you have in China.
12:52
Speaker 1: And you know, being on the right side of the
12:54
Speaker 1: current administration is important, which is probably one of the
12:59
Speaker 1: reasons since that firms like Open the Eye are saying
13:02
Speaker 1: that we are open to the government taking five percent
13:06
Speaker 1: stake in the company, because if they are on your side,
13:08
Speaker 1: well then you're on the right side of things from
13:13
Speaker 1: your perspective at least.
13:14
Speaker 2: That's a quite important theme in your book, which is
13:18
Speaker 2: that ultimately has undermined in the past a country's position
13:25
Speaker 2: at the sort of cutting edge of innovation but also
13:28
Speaker 2: of growth. And you have the example of the UK
13:32
Speaker 2: after the industry revolution is that you have vested interests
13:36
Speaker 2: build up which slowly kind of resists disruption and competition
13:42
Speaker 2: and new entrance to the tow industries, which you then
13:45
Speaker 2: you know, gradually lose your innovative edge. You seem to
13:49
Speaker 2: see scope for that happening both in the US and
13:53
Speaker 2: in China at the moment, But I guess for very
13:54
Speaker 2: different reasons.
13:56
Speaker 1: Yeah, And I think, you know, I think that's the
13:58
Speaker 1: natural order of things, right. So if you look at historically,
14:00
Speaker 1: the leaders in bicycles didn't become leaders in automotive, although
14:04
Speaker 1: they try, right. Legacy media companies did not lead the
14:08
Speaker 1: social media revolution, The legacy car companies did not lead
14:12
Speaker 1: in electric vehicles. The legacy retailers did not lead in
14:15
Speaker 1: e commerce. So it's a clear pattern, the pattern that
14:19
Speaker 1: it's new firms that tend to develop new technology and
14:24
Speaker 1: new kinds of industries, and old industries have an incentive
14:29
Speaker 1: or ever incentive to prevent that sort of competition. And
14:33
Speaker 1: so that's why we're see in the rise of killer
14:35
Speaker 1: acquisitions in the United States, for example, whereby incumbents buy
14:39
Speaker 1: up promising startups just to shut them down. That's why
14:43
Speaker 1: we see a revolving door between the US Patent and
14:46
Speaker 1: Trademark Office and some incumbents, whereby you know, patent examiners
14:51
Speaker 1: grant them low quality patents and then take on jobs
14:55
Speaker 1: for these firms in return. And that's all that. It's
15:00
Speaker 1: creating barriers to entry for new companies, and it helps
15:04
Speaker 1: explain why we see in a decline in business dynamis
15:08
Speaker 1: despite the fact that every technology from the personal computer,
15:12
Speaker 1: the internet, the cloud, and now ai will have made
15:15
Speaker 1: it much cheaper to set up a company and operate
15:18
Speaker 1: a firm, and yet we're see less entry and so
15:20
Speaker 1: those barriers are clearly at work in the US and
15:24
Speaker 1: they were at work in China as well, although in
15:28
Speaker 1: China you have the added component that the priorities of
15:35
Speaker 1: the CCP has changed over the past fifteen years or
15:39
Speaker 1: so from basically economic targets to targets that concern political
15:47
Speaker 1: means around self sufficiency and national security, common prosperity and
15:54
Speaker 1: so on. And what that means in the Chinese case
15:58
Speaker 1: is that that creates it greater reliance on state owned enterprises,
16:04
Speaker 1: because private firms are for the most part less keen
16:07
Speaker 1: on pursuing national non economic objectives, and by any measure,
16:13
Speaker 1: state owned enterprises in China have been less innovative and
16:19
Speaker 1: less productive, and that I think is unlikely to change.
16:26
Speaker 2: And one set of sort of arguments against or points
16:38
Speaker 2: against your thesis would be sort of in the realm
16:42
Speaker 2: of this time is different, or that AI might be
16:46
Speaker 2: a different kind of technology. You make the example of
16:48
Speaker 2: the bicycles to cars and another sort of classic technological shifts.
16:55
Speaker 2: And you know, one argument that you could make just
16:58
Speaker 2: looking at the way has evolved over the last few
17:02
Speaker 2: years and even the last few months, is it seems
17:04
Speaker 2: like scale matters more than anything else in this the
17:08
Speaker 2: development of this technology. And whereas in the past we
17:12
Speaker 2: might have thought that it was going to be about
17:14
Speaker 2: different innovations around different ways of doing things, the big
17:20
Speaker 2: developments in AI seem to be coming just from throwing
17:23
Speaker 2: a lot of resource at llm's and you know it's
17:28
Speaker 2: going to and that the gains go to the country
17:31
Speaker 2: that has the most energy and the most chips, and
17:34
Speaker 2: at the moment, certainly the most energy is China doesn't
17:37
Speaker 2: currently have the most chips. But doesn't seem to suggest
17:41
Speaker 2: that you need a lot of innovation or sort of
17:43
Speaker 2: new entrance, because actually the new entrants can't necessarily afford
17:47
Speaker 2: to put in all these enormous fixed costs. So what's
17:50
Speaker 2: the chance that this is just a different kind of technology.
17:53
Speaker 2: It's going to be innovating itself if it's given enough
17:56
Speaker 2: resource to do it.
17:57
Speaker 1: So, if what you've said is correct and clear, what
18:00
Speaker 1: I've written and said is wrong, right, And so I
18:03
Speaker 1: don't believe that that is the case though. And so
18:07
Speaker 1: you know, if the world was just a static distribution
18:10
Speaker 1: of events, you could probably you know, brute force things, right,
18:13
Speaker 1: So more you know, more compute, more data world, you know,
18:16
Speaker 1: eventually get you there, right. But the world is not
18:18
Speaker 1: just a static distribution of events. It's changing all the time, right,
18:22
Speaker 1: And so you know, my job today may be very
18:26
Speaker 1: different from my job tomorrow. Not all work is like that.
18:30
Speaker 1: And so there's a lot of things that static and
18:33
Speaker 1: you can automate that, but you know a lot of
18:35
Speaker 1: things also requires some degree of resilience, right, And so
18:39
Speaker 1: if you use AI to manage supply chains, that might,
18:42
Speaker 1: you know, work quite well, and then all of a
18:43
Speaker 1: sudden you have pandemic and you don't know what on
18:46
Speaker 1: earth is going on. And we see this, you know,
18:49
Speaker 1: even in more confined spaces, right, So most people know
18:52
Speaker 1: of AlphaGo that beat licid all forty one back in
18:59
Speaker 1: twenty six, and so you know, by that time already
19:03
Speaker 1: AI had achieved super human performance in GO. Few people
19:08
Speaker 1: know that actually, just a couple of years ago, human
19:11
Speaker 1: amateurs using standard computers be the best available GO programs
19:18
Speaker 1: quite easily by exposing them to new positions, new concepts
19:22
Speaker 1: that they would not have encountered in training. And so
19:25
Speaker 1: that rises a sort of fundamental question of you know,
19:28
Speaker 1: even in cases where we achieve super human performance, we
19:31
Speaker 1: cannot be sure if that's actually going to be true
19:34
Speaker 1: tomorrow when circumstances change. And so, you know, humans we
19:39
Speaker 1: are capable of learning from just a few examples. We
19:43
Speaker 1: are very very data efficient, and I think you will
19:47
Speaker 1: have to get to AI that is capable of that,
19:52
Speaker 1: and that will need some innovation. And so right now
19:56
Speaker 1: we don't really know what the path forward in AI
20:00
Speaker 1: looks like. It may be that's you know, large language models,
20:03
Speaker 1: it's the future of AI. It may be that small
20:06
Speaker 1: language models, it might be world models, it might be
20:08
Speaker 1: something entirely different. And so I think what we need
20:12
Speaker 1: is greater data efficiency, and that's clearly not something we're
20:15
Speaker 1: going to get through just through scaling. AI is still
20:19
Speaker 1: you know, waiting for what I call the separate condenser moment,
20:23
Speaker 1: because you know, this first Industrial Revolution, early on steam engines,
20:28
Speaker 1: they were tremendously energy and efficient. They were basically just
20:31
Speaker 1: used to drain coal mines. They couldn't really be used
20:34
Speaker 1: for anything else, and it took the separate condenser to
20:37
Speaker 1: make them energy efficient for them to be applied to
20:40
Speaker 1: transportation later on railroads and steamships, et cetera. And I
20:45
Speaker 1: think AI is still waiting for that moment, and that's
20:48
Speaker 1: a question of you know, further innovation, not just scaling.
20:51
Speaker 2: It's also it's a very good documentary about the AlphaGo experience,
20:55
Speaker 2: and it was this sort of classic example that stayed
20:59
Speaker 2: in people's mind of like when the humans lost gets
21:02
Speaker 2: this new technology. I was surprised to read in the
21:05
Speaker 2: book the idea that you actually had had a fight
21:08
Speaker 2: back by people who were using techniques that AlphaGo had
21:13
Speaker 2: not been able to surmise from it from what it
21:16
Speaker 2: was learning from, And sort of interesting to me that
21:19
Speaker 2: people are determined to think that humans are going to
21:21
Speaker 2: lose this race rather than jumping on those kind of examples.
21:26
Speaker 2: So it seems like you would you know, Again, I
21:30
Speaker 2: mentioned at the start there's this kind of dizzying array
21:33
Speaker 2: of new models of AI, and then debates about where,
21:37
Speaker 2: you know, how close the US is to China when
21:41
Speaker 2: you're trying to think about AI's impact on the world
21:44
Speaker 2: and whether the US is ahead of China or behind China.
21:48
Speaker 2: Is that even the right way to think about it
21:50
Speaker 2: when we look at the sheer power of these models
21:53
Speaker 2: as a gauge to that, or should we be looking
21:56
Speaker 2: at other aspects of their economies.
21:58
Speaker 1: So unless we get to a stage where some firm,
22:04
Speaker 1: you know, some company gets onto curve where it really,
22:08
Speaker 1: you know, just pulls away from the rest, I don't
22:12
Speaker 1: think you know whether the US or China or you know,
22:17
Speaker 1: open R or and Tropic or Google is three months
22:21
Speaker 1: ahead or not. If one place really pulls ahead, that
22:26
Speaker 1: could have a meaningful impact, But it's not clear that
22:30
Speaker 1: that's going to be within the sort of space of
22:33
Speaker 1: large language models. It might be something else. And then
22:36
Speaker 1: there's a question of adoption. And obviously, in the end
22:39
Speaker 1: of the day, the use of a technology is what
22:41
Speaker 1: drives productivity, but they're also the question is adoption for
22:45
Speaker 1: what if people adopted for email, it's not going to
22:49
Speaker 1: drive growth in a meaningful way, but you know, it may
22:53
Speaker 1: look good in sense that the adoption rate is high.
22:55
Speaker 1: But if people adopt it for scientific research in ways
22:59
Speaker 1: that develop new products and new technologies, that's obviously a
23:04
Speaker 1: different matter. And I think there, you know, incentives matter
23:07
Speaker 1: a lot, and I think that has some bearing on
23:10
Speaker 1: why we're in this sort of global productivity stagnation since
23:14
Speaker 1: the computer revolution. Really, so what you really see with
23:18
Speaker 1: the computer era is that inventors and scientists have taken
23:23
Speaker 1: on more projects since then. So when you get a
23:25
Speaker 1: new powerful productivity tool, right, you can be one of
23:27
Speaker 1: two things. You can either use it to drill deeper
23:30
Speaker 1: or dig deeper, or you could use it to your
23:33
Speaker 1: drill more holes. And if you do more projects, just
23:37
Speaker 1: drill more holes. In the end of the day, your
23:40
Speaker 1: attention is going to be more thinly spread across multiple projects,
23:45
Speaker 1: and as a result of that, you are actually less
23:48
Speaker 1: likely to make a breakthrough at any given time. We
23:52
Speaker 1: see that in the data. AI seems to have had
23:54
Speaker 1: the same effect. AI means that we can do more things,
23:58
Speaker 1: but people seem to be using it to do more
24:00
Speaker 1: things rather than digging deeper, and so in academia the
24:05
Speaker 1: incentive is publish a parish, and so we shouldn't be
24:07
Speaker 1: surprised if people use it to produce more output rather
24:11
Speaker 1: than you know, spending the next ten years maybe producing nothing,
24:15
Speaker 1: by coming out with a real breakthrough by the end
24:18
Speaker 1: of that period.
24:19
Speaker 2: Finishing this line of thought, if you're country, as most
24:25
Speaker 2: countries are, that are not home to companies that are
24:30
Speaker 2: absolutely at the forefront of producing this technology, probably your
24:35
Speaker 2: government is saying adoption is key. Certainly we hear this
24:38
Speaker 2: in UK, we hear this in Europe, and to your point,
24:42
Speaker 2: you know, the most important thing is going to be
24:44
Speaker 2: the way we adopt this technology, not whether we happen
24:46
Speaker 2: to own this or that piece of it. You know,
24:51
Speaker 2: what are the lessons from your analysis for governments that
24:56
Speaker 2: want to try and make the most of these technological
25:01
Speaker 2: advances and translate them into growth.
25:06
Speaker 1: So I think if you're behind the frontier, which is
25:09
Speaker 1: inevitably true for most places, and you know that was
25:14
Speaker 1: true d in the Second Industrial Revolution the computer revolution
25:17
Speaker 1: as well, the best thing you can do is trying
25:20
Speaker 1: to adopt technology invented elsewhere. And then the question is,
25:24
Speaker 1: obviously how easy is that and so during the post
25:26
Speaker 1: war period, the United States had an explicit policy through
25:31
Speaker 1: martial aid, where it shared technology with its allies, and
25:36
Speaker 1: that's contributed, i think to a meaningful degree to the
25:39
Speaker 1: post war miracle in Europe, also in Japan and Korea. Similarly,
25:46
Speaker 1: with the computer revolution, buying used technology has been fairy
25:51
Speaker 1: straightforward for most places. And so, you know, the reason
25:55
Speaker 1: that the entire world isn't rich and isn't you know,
26:00
Speaker 1: at the frontier in computers is not that computer technology
26:04
Speaker 1: is not unavailable to them, but that there are you know,
26:08
Speaker 1: some institutional or pre cultural constraints that prevent adoption or
26:13
Speaker 1: at least sort of prevents the creation of domestic firms
26:18
Speaker 1: and domestic industries around these new technologies. YI is not
26:24
Speaker 1: really different from that, but it might become different in
26:30
Speaker 1: the sense that there are clear national security concerns around that.
26:37
Speaker 1: And so you saw that recently with the Trump administration
26:41
Speaker 1: imposting restrictions on foreign use of Entropic's latest model. We
26:48
Speaker 1: can expect to see similar things happening going forward, perhaps
26:53
Speaker 1: at greatest scale. And that, you know, obviously means that
26:59
Speaker 1: you cannot really be dependent on the technology leader. You
27:05
Speaker 1: have to try to grow some domestic capacity with the
27:09
Speaker 1: eye with large language models. The easiest way of doing
27:12
Speaker 1: that is through open source or open weights. And you know,
27:16
Speaker 1: that's how China has closed the gap. It really embraced
27:20
Speaker 1: an open weight ecosystem in large part because of export
27:28
Speaker 1: controls on ships, which essentially forced it to go there.
27:32
Speaker 1: And so I think for many countries they will be
27:38
Speaker 1: probably pivoting either towards Chinese or European technology if they
27:44
Speaker 1: feel that America is an unreliable trading partner in technology,
27:52
Speaker 1: or they may try to build their own domestic open
27:55
Speaker 1: weight ecosystem, although that is going to be I think
28:01
Speaker 1: a harder approach for most places.
28:03
Speaker 2: Well, it's very thank you so much.
28:06
Speaker 1: It's been a pleasure. Thank you for having.
28:20
Speaker 2: Trumponomics was produced this week by Moses and and Samasadi
28:24
Speaker 2: with help from Amy Keen. Sound design was by Blake
28:28
Speaker 2: Maples and Kelly gary And. To help others find us,
28:31
Speaker 2: please rate and review us highly wherever you listen.