The real contest isn’t humans versus AI. It’s concentrated versus distributed power.
Daron Acemoglu is about as close as economics gets to a rock star. Still only 58, the MIT professor and Nobel laureate has spent his career asking why nations prosper, why technological progress sometimes benefits everyone and sometimes only the few, and what happens to democracy when economic power becomes concentrated.
Now, in What Happened to Liberal Democracy? Remaking a Politics of Shared Prosperity, he turns those questions on artificial intelligence.
Much of the AI debate is framed as humans versus machines: will AI destroy jobs or usher in extraordinary prosperity? Acemoglu argues that this misses the bigger question: who controls the technology, how is it used and who captures the gains?
There are, broadly, two possible AI futures.
In one, AI is primarily used to replace human labour. Companies automate tasks, employment becomes less secure, workers lose bargaining power and a growing share of the gains flows to the owners of technology and capital.
In the other, AI complements human intelligence. It helps nurses diagnose patients, teachers personalise instruction, engineers solve harder problems and workers acquire capabilities previously reserved for specialists. Instead of making people redundant, technology makes them more productive and valuable.
Acemoglu fears we are moving too far towards the first model.
That makes AI not simply an employment issue but a democratic one. A handful of companies accumulating extraordinary wealth, data and technological capability gain political as well as economic power.
The danger is a reinforcing cycle: technology → wealth → political influence → greater technological control → greater wealth.
This argument runs through Acemoglu’s earlier work. Why Nations Fail contrasted inclusive institutions that spread opportunity and power with extractive ones that concentrate them. Power and Progress argued that technological innovation does not automatically produce shared prosperity.
AI is the latest chapter in that struggle.
For Acemoglu, the crucial question is not how powerful AI becomes, but who becomes more powerful because of it.
An economy could enjoy spectacular AI-driven productivity, booming markets and rising GDP while millions of people experience insecure work, weaker bargaining power and declining economic importance. It might look successful economically while becoming dangerously unstable politically.
Acemoglu rejects both AI utopianism and fatalism. Technology does not determine its own destination. Companies choose investments, governments create incentives, universities set research priorities and institutions determine how the gains are shared.
The difficult question is whether we can steer AI towards making people more capable rather than simply making more of them unnecessary.
That may ultimately determine not only what AI does to our jobs, but what it does to democracy.
The defining struggle of the AI age isn’t human versus machine.
It is concentrated power versus distributed power.