Bill Gates is urging governments to treat advanced AI as an immediate policy problem, arguing that its biological, cyber, social and labor-market capabilities have already crossed thresholds that should have triggered stronger safeguards.
In an interview accompanying a new essay, the former Microsoft chief said he was especially concerned about models that can design novel molecules and the potential misuse of that capability in bioterrorism. His warnings are judgments about risk, not evidence that the worst outcomes are inevitable.
Gates proposed several ways to cushion disruption. One is to reserve selected occupations for people, with each country deciding which roles should remain human. Another is to tax robots and the tokens consumed by AI systems when automation replaces work, directing some of the resulting economic value toward the public.
He still expects AI to improve agriculture, health care, education and access to bureaucracy. The practical tension is timing: Gates believes those benefits may eventually produce greater abundance, but says institutions are not moving quickly enough to manage the instability that could arrive first.