Rich Whitehouse’s critique of AI is less about machines taking over than about who owns the machines, whose work built them, and what happens to everyone else. He acknowledges that language models can be useful while arguing that the industry’s promises obscure an appropriation of human labor and a drive to concentrate power.
His account connects several incentives:
- Training data and ownership: companies absorb other people’s writing and code, then defend their own systems against copying. Whitehouse sees their objections to distillation—training another model from a model’s outputs—as especially revealing.
- Regulatory capture: warnings of catastrophe and appeals to competition with China can help incumbents secure protection, exceptions, and barriers to competitors.
- Workers’ leverage: if automation succeeds, increased production does not automatically mean shared prosperity. People whose labor is no longer needed may also lose the bargaining power that lets them demand a share.
That last question is the essay’s strongest: why would a society organized around private ownership distribute automated abundance to people who can no longer withhold essential work? A technical breakthrough does not answer a political question about access, rights, and power.
Whitehouse goes further, declaring current models incapable of genuine intelligence and forecasting a convergence between labor displacement and elites’ indifference to mass climate deaths. The essay does not demonstrate those technical limits or substantiate a coordinated plan. Its ownership critique is worth engaging without treating its most sweeping forecasts—or its claim that virtually every model output is illegal—as established facts.
The conclusion is less puritanical than the opening. People may need to use these systems to keep their jobs; using a product does not require endorsing its supply chain. His request is awareness rather than self-hatred: “Do what you must do to stay alive, but try to maintain awareness and help others to do the same.”