Geoffrey Hinton, widely known as a pioneer in AI, says it is reasonable to believe there is about a 10% chance that advanced AI could end humanity. In comments reported by outlets, he characterizes it as difficult to quantify how AI risk evolves, but argues against dismissing the possibility as extremely unlikely.
Hinton does not present a precise prediction method, and the reports emphasize the uncertainty in estimating probabilities for long-term, large-scale outcomes. Business Insider highlights his view that it would be “very foolish” to assume a near-zero risk, while noting that confidence in any single number is limited. Yahoo News frames the same remarks as part of his broader warning about existential risk.
Across the coverage, the key difference is mainly framing rather than substance: the outlets agree on the substance of Hinton’s claimed likelihood and his reasoning that the risk should be taken seriously despite uncertainty.
