The AI Economy Has a Productivity Problem Before It Has a Labor Problem

Much of the economic discussion surrounding artificial intelligence focuses on job destruction.

That concern is understandable, but it may not be the first economic question that needs answering.

The more immediate question is whether the enormous investment in AI will produce enough additional productivity to justify its cost.

The distinction is important. If AI allows workers to accomplish substantially more in the same amount of time, the technology can raise productivity and, over time, increase economic growth. If companies spend heavily on AI without generating comparable improvements, the result could instead be an extraordinary misallocation of capital.

There are already signs of a massive investment cycle. AI infrastructure, enterprise adoption, and demand for computing capacity are driving significant new investment.

Investment itself is not evidence of a bubble. But neither is technological progress evidence that all investment will earn attractive returns.

The history of technological revolutions is full of examples in which infrastructure was built faster than profitable demand emerged.

That does not make the infrastructure useless. It does mean investors can be wrong about timing.

The labor market presents a related challenge.

If AI substitutes for certain forms of skilled labor, the gains may initially accrue to companies and highly productive workers. Over time, however, competition can redistribute those gains through lower prices, higher wages, new industries, or new employment opportunities.

The transition between those stages is where economic policy matters.

Education systems must adapt. Workers need opportunities to acquire complementary skills. Firms need incentives to invest in productivity rather than simply reducing payrolls. Policymakers need to think seriously about whether existing institutions are capable of managing rapid technological transitions.

The most extreme predictions on either side are therefore unlikely to be useful.

AI probably will not eliminate human labor.

It also probably will not leave the labor market unchanged.

The eventual economic effect will depend on what humans do with the technology.

That may be the most important variable in the entire AI debate.