AI, Markets, And What Remains Scarce

Every few months someone predicts that AI will make investing impossible. The argument is usually simple: everyone will use the same superintelligent investor, markets will become perfectly efficient, and excess returns will disappear because every opportunity is identified instantly.

I do not think it is that clean. Investment AIs will probably not become one universal brain. They will be personal, institutional, ideological, constrained, and trained around different assumptions. One investor wants safety. Another wants asymmetry. A fund wants edge. A company wants treasury discipline. A person wants to sleep at night.

Hedge funds will not run one model. They will run many. Companies will have their own systems. Individuals will have AIs trained on their philosophy, risk tolerance, time horizon, and objectives. Different data, different incentives, different definitions of success. That does not remove disagreement. It may amplify it.

Markets exist because intelligent people can look at the same information and arrive at different conclusions. AI may make every participant faster and better armed, but that does not automatically make everyone agree. The market may become more informed without becoming perfectly efficient.

The question that interests me more is what AI does to money itself. As a physicist, I like examining extremes. Suppose AI succeeds completely. First it replaces intellectual work. Then robots automate physical work. Agriculture, logistics, manufacturing, software, accounting, legal work: everything becomes progressively cheaper because fewer human hours are required.

One person could supervise thousands of machines. Eventually one person could manage an entire farm that once required hundreds of workers. If productivity keeps increasing and labor becomes abundant through AI and robotics, what exactly is money pricing? Land? Energy? Natural resources? Compute? Attention? Reputation? Trust?

Perhaps the economy gradually shifts from production to allocation. Producing another product becomes almost free. Deciding which product deserves attention becomes the difficult part. Intelligence does not disappear as a scarce resource. Human attention may become the battlefield.

That is where AI becomes fascinating to me. Investing may survive, but its objective could change. Capital may matter less as a claim on scarce production and more as a way to allocate energy, compute, trust, attention, and priority. The financial system may remain, while the things it prices become fundamentally different.

I do not know if this is where we are headed. But if AI reaches its logical extreme, the question is not only whether markets become smarter. The question is what remains scarce when intelligence and labor are no longer the obvious constraints.