Experts from the UN-affiliated International Scientific Panel on AI have criticized the trend of promoting doomsday scenarios regarding artificial intelligence. Joelle Barral, a Google DeepMind executive and panel member, stated that researchers should focus on scientific evidence rather than spreading fear about potential existential threats.
The discussion follows recent claims by former Anthropic and OpenAI employee Jacob Coxon, who suggested that AI could potentially destroy humanity by the end of the decade. A former Google DeepMind engineer also previously claimed that it might already be too late to prevent AI from causing human extinction.
Nobel Peace Prize laureate Maria Ressa, speaking at a forum held during the UN General Assembly in New York, argued that the current discourse on AI risks has become polarized. She noted that the debate is shifting toward ideological conflict, oscillating between extreme fear of killer robots and the dismissal of any potential danger.
The International Scientific Panel on AI, established in 2025 with members selected in February 2026, aims to provide scientific insights for UN consultations. Co-chair Yoshua Bengio emphasized the need to distinguish between proven risks and speculative extrapolations, calling for a global, rational dialogue that acknowledges uncertainty.
