'Pause Giant AI Experiments'

Eight days after GPT-4, the Future of Life Institute published a letter asking all laboratories to pause, for at least six months, the training of any system more powerful than it. "Should we develop nonhuman minds that might eventually outnumber, outsmart, obsolete and replace us?" it asked, and proposed that the pause be used to build shared safety protocols audited by outside experts. Yoshua Bengio, Stuart Russell, Elon Musk, Steve Wozniak and the historian Yuval Noah Harari signed; by the summer the count passed 33,000.

No laboratory paused, and several signatories said they had not expected any to. Musk announced his own laboratory, xAI, in July. The letter's real effect was on the conversation: it put the words "existential risk" into every newsroom, it was followed in May by a one-sentence statement from the Center for AI Safety, signed by the heads of OpenAI, DeepMind and Anthropic, that the risk of extinction from AI should be a global priority, and it framed the hearings, summits and executive orders of the rest of the year.

Critics, including the authors of the stochastic-parrots paper, argued that the letter's science-fiction framing distracted from harms already happening. That argument, between long-term and present risk, organised the safety debate for the next two years.