DeepMind: Robert O'Callahan Resigns and Sounds the Alarm on AI

Robert O’Callahan, a technical expert at Google DeepMind in New Zealand, has left his position citing the risks associated with artificial intelligence. He criticizes the current pace of AI development and warns against the race towards superintelligence, amidst tensions surrounding effective altruism and "doomerism" in public debate.

An official memo targets effective altruism, according to Axios

A memo from the White House labels effective altruism as a marginal, cult-like group and presents Dario Amodei, CEO of Anthropic, as the face of AI-related "doomerism." Eliezer Yudkowsky and Nate Soares are associated with a movement described as controversial and criticized for prioritizing charity over redistribution.

O’Callahan opposes a "soon" superintelligence and lists risks

Robert O’Callahan believes that the danger associated with superintelligent AI exists, although it remains uncertain, and considers the goal of rapidly developing one to be profoundly irresponsible. He warns about cognitive submission, AI-induced psychosis and isolation, concentration of power, economic upheaval, cybersecurity threats, and a lack of accountability. According to him, AI causes more harm than it brings benefits, and many of his peers share the same concerns without expressing them publicly. He cites the essay If Anyone Builds It by Eliezer Yudkowsky and Nate Soares, but does not subscribe to their claim of an inevitable catastrophe.

Background and break: from chip design to "pro-human" work

Based in New Zealand, Robert O’Callahan was a technical expert at Google DeepMind and contributed to chip design tools that made AI more affordable and faster. He recently left his position, citing the risks associated with artificial intelligence and a pace of change he deems too high. He states that he can no longer justify his contribution to this work and now wishes to engage in unambiguously pro-human endeavors.