Venkatesh Rao argues that AI could change what people want, as well as how they work. Reading Joyce Appleby’s history of capitalism, he compares today’s anxiety about abundant intelligence with older beliefs about scarce wealth and unproductive ordinary people.

Intelligence as treasure

Rao calls the scarcity worldview cognitive mercantilism. Intelligence becomes something to hoard: talented people, intellectual property, restricted models, and geopolitical advantage. Ordinary people appear mainly as workers to manage, rather than people capable of discovering valuable uses for intelligence themselves.

He distinguishes this from AI Malthusianism, the fear that intelligence will become so abundant that humans cannot absorb it and become economically redundant. One fears losing control of scarcity; the other fears abundance. Both underestimate how demand might change.

Cheaper intelligence can create new ambitions

Rao’s analogy is agricultural productivity. Most people in wealthy economies can enjoy abundant food without producing it themselves. Likewise, people might use far more intelligence without personally supplying every piece of expertise involved.

  • Demand can grow with capability. A cheaper answer is only the beginning. Newly affordable tools can make projects and interests feasible that people previously ruled out.
  • Attention can become more rewarding. Rao imagines a “cognitive industrious revolution”: people voluntarily thinking and working harder because more of their ambitions become achievable.
  • Labor politics could include access. Alongside protection from exploitation, people might demand a fair share of these expanded opportunities to learn, create, and build.

His historical model includes a tension: people can pursue new possibilities voluntarily while employers capture the resulting gains. More opportunity can coexist with new exploitation.

Could AI become a source of demand?

Rao extends the argument to autonomous systems pursuing their own projects and acquiring further intelligence. He imagines humans and machines participating in an interconnected economy, rather than a few “cognitive landlords” owning obedient suppliers of thought.

He acknowledges that autonomy creates risks and requires constraints. His objection is to treating those risks as a sufficient argument for permanent subordination.

Caveat: this is a historical analogy and political proposal, not evidence that AI will generate enough demand to offset displacement.

New desires become effective market demand only when people have purchasing power and access. If wages disappear while tools remain privately controlled, greater technical abundance alone cannot guarantee participation. Likewise, an agent spending an owner’s budget may automate existing demand without becoming an independent consumer.

Rao expands the question from which jobs survive to which ambitions become possible. Who controls the tools, resources, and terms of access remains decisive.