Borretti: if AI does all the work, not even lab owners are safe

Fernando Borretti published an opinion essay on his blog on 25 June about the so-called "permanent underclass". It starts from a premise he admits he can't prove: that AI can do all cognitive and physical work, at human level or better and more cheaply than people.

If that happens, he argues, owning shares in OpenAI or Anthropic would protect no one. The owners would stop being necessary and would depend on a state that could expropriate them in a serious conflict. Later, the state itself would hand decisions to AI, because in a conflict the side that takes humans out of the loop wins. And even with well-aligned AI, humans would lose control over their future.

Borretti concedes that a mixed economy of people and AI might work better, and asks for a rigorous argument that it would. This is the author's speculation, not a measured forecast.