B (2:57)

In effect, for a lot of ordinary people, you go from a life living in a rural village where you have a small community and a church, relatively healthy air, decent drinking water. Now you've moved to a slum in London where you are working many more hours a day, backbreaking, very dangerous to your health. Potentially fatal infant mortality begins to rise as you move into London. The drinking water's no good, et cetera. So actually your experience as somebody coming into the Industrial Revolution is yes, the company's much more.

A (3:31)

Sounds great, but I'm not making money.

B (3:33)

But actually, objectively, would I rather be a ploughman in Cumbria or do I actually want to be, you know, working in some horrible gradgrind factory?

A (3:44)

Yeah, well, so going back to the thing you opened with in the very first episode of this series, one of the reasons I think AI is the great political question of our time is that it seems extremely likely to me that the aggregate benefits of this technology will be enormous. The distributional benefits, who benefits and when they benefit, I think are very up for grabs and will largely come down to questions of political economy.

B (4:07)

Yeah, well, so here's another example and I talked about the fact that we've gone from the same number of cars can be built by 10% of the people that would have built them 30 years ago. Right. So the productivity gain there is enormous. The productivity per worker has increased tenfold. But the salary of somebody working in a General Motors factory in Ohio in 1979 was approximately $20 in nominal terms, $20 in 1970s dollars an hour. Today, fast forward. In the same town in Ohio, working in the same site on the same factory, individuals are earning $17 an hour. Their salaries have dropped massively in real terms at the same time as their productivity has gone through the roof. There's no reason to believe that as AI makes companies more productive and they have fewer and fewer workers and they're creating more and more Value. That value is going to go to the person who is the software programmer or the person in the call center in Kenya or in Bangalore.

A (5:12)

Yeah, and this is why I think it is a political question. But I think the really important point is the answer to that is not to say, stop the train, I want to get off. Because one of the things I worry about a lot and a thing that I put in a lot to the action plan that I wrote for this government is that when you look at that, you say, how do you create options for the uk? I think the only answer is AI adoption. If you look at what is the UK good at? You've already listed some of them. Professional services, financial services. We could talk about life sciences. My worry is that we say, gosh, there's a lot of challenges that come from this. Let's be cautious, let's not adopt, let's try and preserve things as they are. But our competitors internationally don't do that. They accelerate, they adopt. We already know that the UK, especially UK SMEs, are very slow adopters of technology relative to their US peers. And there's a real risk to competitiveness in this country if we are very slow adopters. My strong intuition is that although there are lots of nuances to this and lots of challenges, we need to be in a position where we are capturing the benefits from this technology in order to be able to figure out how to divide those benefits.

B (6:23)

Okay, so let's take it one stage forward. Another problem that happens as we move from the industrial revolution and technology improves. We get into deindustrialization. So return to my car factory. 90% of those people are no longer needed in the car factory. They lose their jobs or Mrs. Thatcher and conservative governments close coal mines, famously. Now the story around doing that is pretty straightforward. These coal mines are no longer productive. And they also made the kind of arguments you make. Maybe it's not very nice working down a coal mine. It's dirty, it's hard. Who wants to work down a coal mine? And therefore people are going to be able to move on and do more productive, beneficial things and accept it doesn't happen. Because when the coal mine is closed, the coal miner doesn't end up suddenly retraining as a management consultant. When 150,000 truck drivers lose their job in Britain are replaced by autonomous vehicles, 150,000 45 year old truck drivers are suddenly not reimagining themselves as podcasters.

A (7:25)

Just as well by the sound of it.

B (7:26)

Just as well. Yeah. So this question of what happens to those people. And this was a theme, right? The way through the Industrial Revolution. Ultimately, it's true, new jobs emerge. But the idea that somebody aged 40, 45 is going to suddenly be retrained, find a new job, find a new career, seems pretty implausible to me.

A (7:49)

I think this is a huge challenge in some ways. I think a lot of the adjustment has already happened in that so many more of the jobs that we have today are these service sector jobs in the uk, where I personally believe that because of this jagged frontier, we're going to see actually a core human component of that being important for the foreseeable future. But I do think the thing I have felt very strongly is what are these service jobs?

B (8:15)

Because a lot of them seem to be under threat. I mean, you know, we go to the supermarket, we can see that there are fewer checkout people. We're now doing our own scanning, right. I've talked about law firms, I talked about consultancy firms. Where is this? I mean, there's a growth in the public sector, right. The government is spending like a drunken sailor, employing lots and lots more subservence, right?

A (8:35)

Indeed.

B (8:35)

But I don't really see where are all these services jobs.

A (8:38)

The thing that I think is hard about this for people who are broadly optimistic about technology, as I am, is that it's very hard to imagine futures where we get to reallocate resources on this scale. I think you can imagine having that conversation in the Industrial Revolution and people saying, well, where are these service sector jobs? And the answer is, actually, we have many problems in this country, but unemployment hasn't been one of them for a little while. Actually, we, in many cases, when I talk something to entrepreneurs, the big problem we have is actually a lack of skills. People can't fill the jobs that they have.

B (9:12)

But there's a problem there, which I mean the same probably true in the US and when we think about the fact that many Trump voters will say the problem has been globalization, China's taken our jobs. What they don't mean is there's no jobs. In fact, in the US the unemployment rate is quite low. What they say is that I've lost a well paid, dignified job. And if I try to get a well paid job, the employer says to me, I don't have the skills. And what does I don't have the skills mean? Well, it means I am a blue collar, working class guy from Ohio who didn't go to university and there's no jobs for me. Yeah, you don't have the skills. This sort of implies, well, somehow if the government set up a proper skills training program, then Matt's going to employ him in my new tech startup. And you're not going to employ me in new tech. No.

A (9:57)

But I do think that it's very hard to imagine the jobs of the future. But one thing that seems very clear to me is that a nation of confident users of AI will be much better equipped to make the adjustments that come from rapid AI progress. And something that I worry about, and part of the reason that although I share a lot of these concerns about adjustment, I think it's really important not to create fear of this, is that actually I think the best way for people to adjust is to. To use this technology in their jobs.

B (10:28)

My final anxiety is that you sometimes hear the tech people say, and when you talk about political economy, I think you might be hinting at this, that what could happen is that the companies make an enormous amount of money with few employees. We had the example of Dario saying he could imagine a billion dollar company with two employees. The government taxes that company. So the government has money sitting in the government coffers and then the government hands it out to the public. That's your political economy point, I guess, right? We redistribute the money from the billion dollar company with two employees to the average person.

A (11:05)

Because the challenge we have in the UK is we don't have any of those companies.

B (11:08)

Right. That'd be even worse. We'll get onto that when we get onto the geopolitics. But let's imagine in the best case scenario, we have one of these companies, right? The government then hands it out maybe through a universal basic income, I don't know, some welfare payment or some support. And then the idea that I'm sometimes told is that. And that's absolutely lovely because then your dad or whoever will have lots of lovely time where they'll be able to think about Shakespearean sentences like the rest is silence and pursue their deep interest in learning the piano or retrain as, I don't know, a yoga instructor or whatever they're going to do when they used to have a job driving a truck. I have a sort of instinct that humans really, really quite like working. We like being busy. And that this idea of us sort of, even if the money was available, even if these miracle companies emerged and the government had all the money to give us all a universal basic income, the number of us who actually would genuinely live satisfying, fulfilled lives reading Shakespeare and I agree.

A (12:12)

No, I don't think that universal basic income or anything like it can be the answer, I think. I guess again, I think I have quite moderate views on this. I'm playing the optimist here because of your sort of incipient doomerism, Rory. But I think we are underestimated. We have historically always underestimated both the breadth and depth of human wants and needs and the breadth and depth of human ingenuity in meeting those needs. And I think with the slightly more optimistic way of describing. What you're describing is a world where this was true would be a vastly wealthier world with vastly more disposable income to spend on all sorts of wants and needs that we can't imagine today.

B (12:52)

In the GDP figures. But you then said, and vastly more money to spend on wants and needs we can't imagine. At the moment, we seem to be in a world where Elon Musk ends up with a trillion dollars. In other words, we could end up in a world in which all the money that's made is hoovered up by some trillionaires in the United States.

A (13:10)

I will say, though, that, you know, all the evidence from economic history is that it's very easy to fixate on the inventor and owner of, of the technologies. But nearly all of the benefits of new technology diffuse into the economy. So there's like. And the kind of classic stat on this is like, only like 3% of the benefit of new technology is captured by the people that make it. Most of it just comes from you get clean drinking water, you get a mobile phone. You know, like, we, we get a lot of these benefits. And in some ways I do think that the task for us optimists is to have imaginations. And look, I worry about all the things you worry about. I worry about whether the UK is going to be a winner. I worry about whether we're going to have the economic setup that means that ordinary people benefit as much as they could from this technology. But I guess to slightly throw it back on you, what's the alternative? We can't stop the world and get off know what is the way through?

B (14:07)

Well, I would begin by working backwards and saying, are we describing a world that is better or worse? Before I'd get into the question of do we have any alternatives? If the world we're describing was a worse world for people in 10, 20 years time, then I think it is within the gift of politicians to stop it. And it could be stopped almost certainly by the President of the United States and Xi Jinping together if they wanted to. Because actually, at the moment we're at a very lucky stage. This stuff relies on these enormous data centers, this incredible amount of money, they can be spotted from space. We're still at a moment where this stuff can basically be switched off. If we chose to do so now, that would have huge economic implications because he just pointed out, and we switched off this stuff, these companies would collapse and all these companies that have bet on AI would be devastated and it would be a misery for many people that we know. But it could still be a perfectly valid public policy choice if we thought this stuff was going to make people's lives work. And it's a perfectly valid democratic choice for people. So I still think the question has to be do we think in 10, 20 years time we're going to think thank God for that stuff or not? And let me give you a counterexample. Right. Facebook. Do I think we're better off with Facebook? No. Do I think the world arguably could be better if we'd simply shut it down? Yes. I don't think that just because something is very popular and has made a huge amount of money for Mark Zuckerberg and some people are enjoying sharing photographs of their families and keeping up with their families, that makes it a net positive. In fact, I think there are many things that we do in the world, from putting plastics in the ocean through to polluting our soil, eating ultra processed food, which are extremely successful business models, where in retrospect we should have said no. And somebody saying to us, well, what's the alternative? America's going to produce the ultra processed food anyway? Isn't really the answer.

A (16:15)

Yeah, I mean, I think there's a question there about what can the UK do unilaterally? And I guess we'll get onto that in the next episode. I think the thing that I would say though is that there's a saying in venture capital which I quite like, which is pessimists sound, smart optimists make money. I guess my framing here would be, I do think that it's for me hard not to hear this and think we would have made this argument about steam electrification, you know, like really everything that built the modern world. And so, you know, will it get better or will it get worse?

B (16:52)

And that's what we have to get onto. Because I would have thought what's incumbent on me is to prove why this is different from steam electrification. And for that I would have to explain that actually creating artificial general intelligence is not just like electricity. Yeah.

A (17:04)

And if I believed that we were on the path to this frictionless world that you described, I think I would be more in your camp. I think we're in this. I think we live in a very complex, high friction world where actually intelligence is like so multidimensional and all this replacement is actually incredibly hard. And in the real world, when you're at the coalface, you're like, ah. And humans are just so versatile and so well evolved to deal with the world that we actually live in.

B (17:31)

So to some extent we don't need to worry because this stuff isn't going to work that well anyway.

A (17:34)

I think it will work extraordinarily well in domains that are extraordinarily valuable, but I do not think that we're heading for this Frictionless plug and play is a human replacement, and so I think humans will have comparative advantage in a bunch of things for a very, very long time.

B (17:51)

To listen to our full series on AI, sign up@therealstispolitics.com.