Silicon Valley optimists are betting that an era of AI-powered superintelligence will bring about explosive worldwide economic growth. Henry Curr, our economics editor, joins Jason Palmer, co-host of The Intelligence podcast, to discuss if their gamble will pay off. 00:00 – Can AI supercharge global economic growth? 01:15 – What are the predictions for how AI will impact the economy? 02:01 – What will cause explosive economic growth? 03:39 – Will jobs be at risk from AI? 05:22 – Should people buy into AI companies? 07:09 – What is important to look at in the real world? 08:37 – Is the AI boom similar to the internet boom? Listen to the full episode: https://econ.st/41Shtjz The economics of superintelligence: https://econ.st/4mUX7hW How America’s AI boom is squeezing the rest of the economy: https://econ.st/4mpaFlM Sign up to our weekly newsletter: https://econ.st/3J0xzBr Sign up to our Money Talks newsletter: https://econ.st/4oLIsY6
Can AI supercharge global economic growth?
In 1700, it would have seemed natural that economies only ever stood still. In the common era up to that point, output had expanded by just 0.1% per year on average. Then, steam engines began to puff. Global growth quintupled to 0.5% per year between 1700 and 1820. By the end of that century, it had reached 1.9%. And in the 20th century, output grew by 2.8% on average. The long history of the world economy is one of expansion at an increasing rate. And if you believe the inhabitants of Silicon Valley, the world’s economy is about to be transformed again. And that’s because AI is going to get so powerful that it is going to be able to carry out all sorts of tasks that are currently done by humans and eventually all tasks. And the implication of that is that you get explosive economic growth. I’m Jason Palmer, co-host of The Intelligence Podcast. And I’m Henry Kerr, economics editor, The Economist. Today, we’re going to carry out a little thought experiment looking into how the world economy will develop if the most outlandish predictions of the AI world actually come true. Remember to listen to the full episode. Click the link in the description below. Henry, let’s just
What are the predictions for how AI will impact the economy?
start with what the predictions are as they stand now. Well, if you believe the inhabitants of Silicon Valley, the world economy is about to be transformed. And that’s because AI is going to get so powerful that it is going to be able to carry out all sorts of tasks uh that are currently carried out by humans and eventually all tasks that are currently carried out by humans. And the implication of that in some e economic models is that you get completely explosive economic growth. We’re used to growth of 2% or maybe 3% in a good year in advanced economies over the past half century century or so. Uh some of these models churn out numbers like 20 or 30% growth. Carrying out our thought experiment then
What will cause explosive economic growth?
how do we get from a 2% 3% world to a 20 30% world? What does the middle of that journey look like? Well, I think the best way to answer that is to think about what enabled humanity to go from the very low growth pre-industrial revolution to the two to three% growth norm today. And part of that picture early on was the so-called accumulation of labor. Uh the size of the economy was very closely linked to the size of the population. And a bigger population led to the creation of more ideas because you have more people sitting around thinking uh and more output enabled death rates to come down and enabled birth rates early in the the industrial revolution. In theory, if your labor force, as you might call it, is made up of lots of AI agents who are really capable. In theory, you don’t have to wait around for generations to pass for your for your workforce to grow in the same way as you would have done in the earlier era. So the people who come up with these kind of numbers talk about the accumulation of AI workers being far more rapid than the accumulation of human workers. So what starts to happen is you get massive investment in uh the the production of AI agents, data centers, energy and so on. That pays off and produces a lot of growth because it automates so many tasks and then that payoff is reinvested into still more AI power and that loop can turn really quickly unlike when you’re looking at population accumulation uh where it takes a while for that to happen.
Will jobs be at risk from AI?
So two things there. One is that for the people who work um in automatable tasks, they are then going to have trouble more quickly. But that still leaves a bunch of things that AIs can’t do. AI plumbers don’t yet exist. Yes. So there are various extremes you to which you can push this thought experiment. Lots of people in Silicon Valley believe this. Super intelligence basically replaces everybody. Uh you have an AI that makes itself better. uh you have an AI that solves all robotics engineering challenges and you don’t need humans anymore. In reality, what you’re much more likely to see is intermediate phases which last a very long time in which there are bottlenecks. That might be bottlenecks to how fast the AIs can improve or how much investment in AI can take place. Or it may simply be um uh things like regulatory bottlenecks to having robots displace uh humans. uh or it might be that there are fundamental limits to what AIs and and robots can do. And uh typically what’s happened in the history of the world economy is that when you have very rapid productivity growth in one type of activity and not much productivity growth in another type of activity, then the low productivity growth sectors still see wages go up a lot. Uh and this is referred to as as cost disease. Uh but if you’re a worker being displaced by an AI and you’re having to find a new job, you might quite like cost disease because it means there’s lots of highly remunerated stuff still around for you to do. But uh that’s still a lot of disruption happening if you have workers moving from one sector to another as uh AIS expand and and displace more uh of the labor force. So looking at um Silicon
Should people buy into AI companies?
Valley valuations on these AI companies and so on. Should I in your view become a plumber or buy into these companies? Well, I wouldn’t become a plumber just yet on the basis of a thought experiment. You could buy into the companies. There are a lot of people in Silicon Valley who think that’s essential because in the long run they think labor is not going to have any value. The only thing that’s going to have any value is uh owning capital, having a stake in these AI companies and things that are complimentary to them. Um and that is uh justified by what you find in some of the economic models of explosive growth based on automation. The returns to capital go up. But it’s a little bit complicated by the fact that those economic models also show uh that in an explosive growth scenario, interest rates should go up a lot. And one way of thinking about this is is if you’re going to have explosive economic growth, you’re and you’re going to have AI powering that, you need a ton of data centers. you need a lot of energy production. Uh if you’re going to have a big economy, you need more infrastructure. Uh so there’s all sorts of demands for capital. But you have a lot of people who think they’re about to get rich because of AI sending economic growth to the moon. And so no one really wants to save because why save for tomorrow if you’re going to be rich tomorrow? A reduced demand to save increased demand for investment means there’s a kind of capital shortage that pushes up interest rates. And what do higher interest rates tend to do? They tend to reduce asset prices. So it’s a it’s it’s you know you can tell the story multiple ways here and uh it’s it’s not entirely clear even if you got capital exactly what you should buy if what you think is going to happen is explosive growth because it depends on these obscure parameters in the models which economists and finance professors disagree about
What is important to look at in the real world?
coming out of the models and looking at the real world what what to watch. How how do you sort of figure out which which end of things are we’re going to actually see as some of this stuff does or maybe doesn’t come to fruition? I think the interest rate story is quite important here because although it is the case at the moment that you have this very uh extreme abolance in the US stock market, high valuations of AI firms which would you know seem to suggest the market buying into the thesis somewhat. in the uh money markets you do not have the pricing in of explosive economic growth and a big increase in interest rates and and I think that interest rate story is quite compelling in theory. So the thing you should watch for whether or not markets are really starting to believe an explosive growth story would be the uh long-term bond yield which should should rise quite a lot if uh if explosive growth is coming. If it’s just really valuable AI companies, well, you can explain that, I think, with a story where yes, these companies are really profitable and AI is the next big big thing, but the economy as a whole isn’t exploding. It’s just that AI is the latest technology that keeps the economy on a on a normal growth path. It happens to be AI today, uh, just like in the past it would have been manufacturing, automation, uh, or the internet or or or or whatever, electricity. Um, and today it’s going to be AI. watch the bond markets. I’d say at the mention of the internet that was
Is the AI boom similar to the internet boom?
there was as much froth around this the the notion of the internet when it was new and how it was going to change productivity, how it’s going to change the world, change the markets, change the economies and so on and kind of didn’t in the long run. Is that instructive to your mind or is AI a sufficiently different technology that we should consider it differently? Yes. Well, I found myself hesitating when I mentioned the internet there for precisely uh precisely this reason. Um it it It’s clear that the internet both transformed the world and did not impact e economic growth all that much at least in a way that’s observable and some people do tell this story where AI will be a similar sort of thing lots of disruption not much measured economic growth and the economists who studied this in the 2010s uh tended to come up with estimates where essentially all the benefit of the internet and uh social media free services went to consumers but uh uh but but not in a way that shows up in your measured economic output. I think with AI, you know, it’s it’s it’s possible to imagine something similar uh happening. You know, if AI makes internet search much better, it’s a similar sort of thing. Um if it makes just the access of uh knowledge easier, then again, it’s a similar sort of thing. The key question here as it is for the explosive growth scenario more generally is does AI help humanity really push forward the frontier of knowledge? Um and if it does then I would say that probably puts AI in a different bucket. You know, if it can uh be creative, come up with research ideas, accelerate the pace of transformation, then you have a potentially quite powerful impact on on longr run living standards uh from from that in a way that perhaps the internet, at least in observer in an observable way, didn’t bring about.