AI leaders call for slower development of frontier systems
Major AI laboratories have begun shifting away from the winner-take-all approach that has shaped the development of advanced models for years. Industry leaders are now calling for greater coordination and a slower pace, warning that frontier AI could soon become too dangerous or difficult to understand and control.
Anthropic’s Dario Amodei argued in a nearly 4,000-word essay that model capabilities should improve more slowly. He said commercial pressure could drive a race to the bottom and intensify catastrophic risks. OpenAI co-founder and CEO Sam Altman endorsed the argument and said similar discussions about pacing were already taking place at OpenAI.
Demis Hassabis, Alphabet’s chief scientist and a co-founder and chair of Google DeepMind, said the essay indicated an appropriate direction and repeated his support for an industry-wide standards body. Microsoft CEO Satya Nadella also backed research and deliberate pacing as part of making alignment a design goal. His comments came before Microsoft’s planned release of a detailed “humanist AI” code of conduct for its models.
Written by Sakhanda Wire AI from reporting by Ars Technica. Facts, quotes and full context are in the original article below.