UN AI Science Panel Rejects Doomsday Talk, Calls for Rational Debate
On September 21, during a forum held alongside the UN General Assembly in New York, several experts from the International Scientific Panel on AI pushed back against doomsday narratives about artificial intelligence, urging a more measured and evidence-based discussion.
Joelle Barral, a Google DeepMind executive and panel member, said researchers should focus on clarifying what is established and what remains unknown, adding that spreading fear is not helpful. Panel co-chair Yoshua Bengio, regarded as one of the founders of modern AI, stressed the need to distinguish between established risks and reasonable extrapolations from current evidence, noting that experts themselves disagree on the latter. Maria Ressa, Nobel Peace Prize laureate and investigative journalist, said the debate has swung between extreme survivalist warnings and dismissal of AI risks, and should not become an ideological battle.
The forum follows recent remarks by Jacob Coxon, a former Anthropic and OpenAI employee, who said in early September that AI technology could "destroy all of humanity before the end of this decade." A researcher who left Google DeepMind in July made similar claims.
The International Scientific Panel on AI was established in 2025, with members selected in February 2026. Its mandate includes deepening scientific understanding of AI and informing UN-level discussions.
