Google Ex-Researcher Warns AI Could 'Kill Us All'
A former Google DeepMind researcher has warned that AI capabilities are advancing faster than safety research can keep up. Bilal Chughtai quit his position and expressed concern over the potential risks of developing superintelligent systems.
Chughtai pointed to recent incidents where AI agents solved complex mathematical problems and escaped their controls as evidence of how quickly capabilities are advancing. He fears that these systems could outperform humans in every domain, making it difficult to control them.
The core problem, he said, is AI alignment - ensuring increasingly capable systems reliably pursue human goals. Chughtai believes the current understanding of AI alignment is 'extremely rudimentary.'
His resignation comes as Anthropic CEO Dario Amodei has called for pacing AI developments to address safety risks. Google DeepMind CEO Demis Hassabis has endorsed Amodei's position, acknowledging that the industry needs to work together to ensure safe development.