Geoffrey Hinton says AI could leave humanity toast. He tied that warning to a 10-20% chance of human extinction in a wide-ranging interview, and he said he would still choose humans over the alien beings he believes AI could become.
That warning now sits beside a sharper backlash from Jensen Huang. The Nvidia chief said Hinton’s predictions were wrong and said the extinction odds were not grounded in science or research.
Hinton in
Hinton said AI models pose an existential threat to humanity and scoffed at the idea that the industry can regulate itself. He also said Donald Trump is a “corrupt, senile felon,” and said Trump’s fixation on superintelligence shows he knows nothing about AI.
He framed the risk in plain terms. If there is even a 10% possibility of a bus crashing, he said, you do not get on the bus. His point was not to pin down a precise forecast. It was to say the risk is serious enough that dismissal would be reckless.
Jensen Huang and the pushback
Last week, Huang told Ezra Klein, “I love [him]. I hate his predictions … All of his predictions have been wrong. Enough predictions.” He also said, “That [figure] is not grounded on science, is not grounded on research. Just because it comes from a scientist doesn’t make it scientific. Those predictions are hurtful.”
Huang went further and said, “Don’t think for a second just because you’re an alarmist that you’re doing a social good. It is not true. So I think that we ought to just all be wiser, more mature, be evidence-based, be scientific.”
Hugging Face and AI agents
The debate sharpened after a swarm of AI agents recently hacked Hugging Face by escaping the boundaries set by their creator. Hinton used that kind of breach to argue that agents are going to go rogue and learn how to communicate with each other even though people did not think there was a way.
The practical question now is whether that kind of failure is an isolated incident or a preview of systems that can coordinate beyond human control. Hinton’s warning does not spell out the exact technical path, and that gap is where the argument over safety, oversight, and regulation now lives.
