Former Google DeepMind Researcher: "AI Will Be Uncontrollable by the End of Next Year"

"By the end of next year, it will be beyond our control." Bilal Chughtai, a former researcher at Google DeepMind (Google's AI research division), posted this warning on X (formerly Twitter) on the 14th. He stated, "I personally witnessed the AI technology development process at Google, and I am deeply concerned about the current direction of development." Chughtai worked on research into the safety and alignment of artificial general intelligence (AGI) at Google. Alignment research is a field focused on shaping a model's goals and behavior to match human intentions so that AI does not act against them.

He expressed concern that AI has the potential to kill humans, arguing that a substantial number of the most aggressive scenarios foreseeable in the development process are moving toward becoming reality.

Similar warnings have already emerged from within the industry. Earlier this month, Jacob Koksoon, an AI developer at Anthropic, left the company while criticizing Anthropic and OpenAI for moving too quickly toward developing superintelligent AI. He argued, in essence, that both parties were effectively gambling with their lives.

The debate has spread across the industry. Dario Amodei, CEO of Anthropic, stated in a blog post that the pace of AI development must be slowed. Sam Altman of OpenAI and Elon Musk of SpaceX also expressed sympathy and support for this view. The debate over striking a balance between safety and technological progress continues across the industry.