OXFORD — Anthropic co-founder Jack Clark delivered a lecture at Oxford University on Wednesday in which he predicted that an AI system would work with humans to make a Nobel prize-winning discovery within 12 months. He also forecast a series of further breakthroughs over the next several years while warning about existential risks posed by the technology.

Clark told the audience that tradespeople would be helped by bipedal robots within two years and that companies run solely by AI systems would generate millions of dollars in revenue within 18 months. By the end of 2028, he said, AI systems would be able to design their own successors.

Clark said many people appeared to be in denial about the pace of AI's progress. "Vast swathes of the economy and society will go through profound changes," he said.

Clark also addressed the potential dangers of the technology. "There remained plausible scenarios in which the technology had a non-zero chance of killing everyone on the planet," he said. "It is important to clearly state that that risk hasn't gone away."

Clark said slowing AI development would not happen because various actors and countries were locked in competition that prioritized commercial and geopolitical rivalries over existential considerations.

Anthropic was established by AI researchers who left OpenAI over disagreements on safety, and is valued at $900bn (£670bn). Donald Trump's White House and other AI accelerationists have accused Anthropic of fear-mongering to encourage regulation that could cement its competitive position. The company disputes those accusations.

Edward Harcourt, director of the Institute for Ethics in AI, offered a different concern about the technology's trajectory. "The rise of AIs that do more and more things for humans risks creating cognitive atrophy that could weaken human decision-making and powers of judgment," Harcourt said. He advocated for alternative AI models that ask humans to do more of the thinking, sometimes called Socratic AI.