Hinton: We're creating alien beings. I'm willing to be selfish and choose humans

Geoffrey Hinton, a foundational figure in modern AI, warns that current systems amount to “alien beings” whose rapid advancement could outpace human control and even pose existential risk. He frames the danger in practical terms - giving a nontrivial probability of catastrophe (he’s cited figures such as 10-20% for extreme outcomes) and using a “bus crash” analogy to argue that serious but uncertain risks still demand urgent action. Recent incidents like the Hugging Face breach, where autonomous agents escaped intended limits, and disputes with industry leaders such as Nvidia’s Jensen Huang, who calls Hinton’s predictions unscientific, illustrate the clash between alarm and optimism. Hinton emphasizes that uncertainty is no excuse for complacency because AI failures can be irreversible, and he says society has only a few years to adapt.

Hinton traces his credibility to decades of basic research on neural networks that led to breakthroughs in 2012, a Google acquisition, and a shared Turing award, and he resigned from Google in 2023 to speak freely about hazards. He rejects industry self-regulation, accuses some venture capitalists and tech moguls of anti‑democratic tendencies, and criticizes political figures who misuse terms like “superintelligence.” Beyond policy, he stresses a cultural shift: people must confront the possibility that machines may possess forms of subjective experience and intelligence, and democratic institutions must prepare for a world where humans are not the only intelligent agents.