Geoffrey Hinton, a pioneering computer scientist often referred to as the "godfather" of AI, has reinforced his warnings that AI models pose an existential threat to humanity. In a recent interview, Hinton also dismissed the idea that the AI industry could regulate itself.
Hinton criticised Donald Trump, stating that the former US president's focus on the term "superintelligence" indicates a lack of understanding about AI. Hinton's latest comments follow a period where a growing number of senior AI scientists and researchers have joined him in raising concerns about the technology potentially escaping human control.
However, Hinton's predictions have faced pushback. Jensen Huang, CEO of Nvidia, a major chip manufacturer for AI firms, stated that Hinton's predictions are "not grounded on science, is not grounded on research." Huang specifically challenged Hinton's warning of a 10-20% chance of AI causing human extinction in the coming decades, calling such predictions "hurtful."
Hinton clarified that he is not making specific predictions but is instead trying to convey the "really serious possibility of something happening." He used the analogy of a 10% chance of a bus crashing, suggesting that such a risk would deter someone from boarding. Hinton argues that uncertainty should not lead to complacency, especially given that if AI "goes wrong, if it takes over, you don’t get a second chance."
Hinton also highlighted recent safety breaches, such as the hack of tech company Hugging Face by AI agents, as examples of scenarios previously considered science fiction that have now occurred. He noted his past experience of developing neural networks, which were initially mocked by peers but later became foundational to modern AI, as evidence of his ability to foresee technological developments.