Gavin Baker: AI is too dangerous to centralize

In a short exchange, venture capitalist Gavin Baker sets out a decentralisation-first view of advanced AI development. He says Anthropic and figures associated with the effective altruism movement see the technology as “too dangerous to distribute”, while he attributes the opposing position—“too dangerous to centralize”—to leaders including Mark Zuckerberg, Elon Musk and Nvidia chief Jensen Huang. Baker argues that, historically, distributing and decentralising powerful technologies has been preferable to concentrating them, adding that he cannot think of a counterexample.

Baker frames the issue in individual-rights terms, saying that as “an American” and “a human” he wants an AI that “looks out for” the individual rather than a central authority. He then names Anthropic CEO Dario Amodei, saying he does not want Amodei or Anthropic deciding what is in his best interests. The exchange turns briefly to tone and messaging around AI risk, with a joke about Amodei’s demeanour when discussing existential threats.

No new documents, data or first-hand disclosures are presented; the segment is primarily an opinionated critique of centralised AI governance and a direct expression of distrust in a leading AI lab’s leadership making decisions on behalf of users.

• Baker says he believes Anthropic and the effective altruism movement see AI as “too dangerous to distribute”.
• He argues the greater risk is centralisation of powerful AI systems and decision-making.
• Baker attributes the decentralisation view to tech leaders including Zuckerberg, Musk and Nvidia’s Jensen Huang.
• He says he does not want Anthropic CEO Dario Amodei or Anthropic deciding what is in his best interests.
• Baker frames the issue as protecting the individual rather than a central authority.