AI may be the US economy’s only hope

If you take the long view, America’s economic outlook is pretty bleak. Like a lot of rich countries, it is overwhelmed by debt it has no plans to reduce. Even more troubling is its aging population, which will reduce growth and leave fewer people to pay all that debt, according to the Tribune News Service.

There is only one hope: a sudden increase in productivity that will boost growth so much it will pay for everything. As it happens, that is precisely the promise, or one of them, of AI. But what are the chances of it coming true?

The stock market is banking on it, as is Kevin Warsh, President Donald Trump’s nominee to chair the Federal Reserve, who is hoping AI-driven growth could justify a cut in interest rates.

It’s a tall order, as net interest outlays are expected to take up 5.4% of US GDP by 2055 — and that’s assuming interest rates lower than they are currently. Meanwhile, America’s fertility rate is about 1.7, below the replacement rate of 2.1. It is even lower in other countries; at this rate, the world population is expected to peak in 2055 and take 2 percentage points off GDP by 2050.

But history is full of moments when either the end of humanity or the exhaustion of natural resources seemed inevitable, and we have always persevered, eventually and most of the time (ignoring famine, pestilence and war). We did so mostly because of our ability — especially in the last few hundred years — to come up with some innovation or another that delivered unimaginable prosperity. Innovation makes us more productive, which means we can do more with fewer resources. It can help make up for a shrinking population.

Speaking of which: For most of history people had the opposite concern about population — that it would grow too much and we would run out of food. But we came up with innovative farming techniques that enabled more food from less land and labour. That in turn freed up more labour for factories, which used more innovations, which created even more growth and in many ways made our lives today possible.

So the question is not whether innovation can drive growth. It’s how much growth innovation can drive. The falling population and heavy debt load mean productivity will need to increase GDP by at least 2.5%, maybe 3% depending on the path of interest rates.

For some context, the electrification of the economy resulted in a 2% real average growth rate over the last 150 years. It not only powered the economy in the 1880s, but it also enabled future innovations. AI will have to do better than electricity. Is that realistic?

Google CEO Sundar Pichai thinks so. In 2018, he called AI “one of the most important things humanity is working on,” saying it could be as profound as harnessing the power of fire. It is hard to put a number on that — but fire-level growth sounds like it would be more than 2%. Some chief executives of AI companies, meanwhile, talk more about AI destroying humanity than saving it.

We economists, having studied the industrialisation of previous centuries, tend to make more conservative assumptions — something between harnessing fire and destroying humanity. But it may be reasonable to expect better than the 2% growth rate that followed electrification because innovation happens so much faster now. The best-case scenario is a 5% growth rate for some time, though that may not be sustainable.

The US is nowhere near that now. Labour productivity was up 2.1% last year, with a 4.4% annualised rate in the last quarter. It is unclear whether AI, or any innovation, is the reason — it could be the smaller labour force from less immigration.