Hugging Face CEO Clément Delangue has questioned the outsized attention given to Jacob Cockson, a former Anthropic researcher who recently resigned over fears that AI could wipe out humanity. In a post on X late Wednesday, Delangue wrote, "Sorry, but having Jacob discuss AI extinction risk is like having your air-conditioner repairman discuss climate change."
Cockson, who previously worked at OpenAI, announced his resignation on Tuesday, saying he believes companies like Anthropic and OpenAI are "racing toward self-improving superintelligence" and gambling with human lives. "People developing AI genuinely believe that before the end of this decade, AI could kill us all. This is not a marketing gimmick," he said.
A call for context
Delangue, however, argues that Cockson's views should be seen in a fuller context. "I'm not saying it's necessarily meaningless or wrong, but we need to stay objective and listen to voices from all specialties across the ecosystem," he wrote. He clarified that comparing Cockson to an air-conditioner repairman was not meant to "discredit" him, but to emphasize the need for a more comprehensive discussion about AI risks.
Delangue noted that Cockson's work focused mainly on pretraining—training AI models on large datasets—rather than on risk research. "It just means we should view any single perspective in a larger context and also hear from the broader AI ecosystem," he added.
The exchange, reported by IThome (IThome), highlights a growing rift within the AI community over how to assess existential risks. While some researchers warn of catastrophic outcomes, others caution against overhyping doomsday scenarios that could distract from more immediate concerns like bias, misinformation, and job displacement.
What it means for tech watchers
For readers following the AI industry, this debate underscores the importance of evaluating claims based on expertise and evidence. Delangue's analogy serves as a reminder that not all opinions carry equal weight, even from insiders. As AI development accelerates, the conversation around safety and regulation is likely to intensify, with voices from various disciplines—engineers, ethicists, policymakers, and yes, even "repairmen"—contributing to the discourse.
Based on reporting by IT之家. Edited and published in English by geisou.
