Geoffrey Hinton, a Nobel Prize winner known as the “godfather of artificial intelligence,” says AI could surpass humans much sooner than he had predicted.
The 78-year-old researcher acknowledges that his earlier assessments of the risks posed by the technology were too optimistic. He now believes artificial intelligence systems could exceed human capabilities in a much nearer future.
Të lidhura
None found
Hinton’s concerns grew after he attended lectures by Paul Cristiano, founder of the Alignment Research Center in Toronto, Canada. Cristiano believes there is a high risk that artificial intelligence could take control within the next three years, with potentially catastrophic consequences for humanity.
A prediction for the next 18 months
Until recently, Hinton thought systems far more intelligent than humans would emerge in about a decade. But the assessments he heard from Cristiano during the lectures led him to think progress could be much faster.
Hinton explained that he used to warn such a development could happen within ten years, whereas Cristiano now says it could happen in the next 18 months. He said the prospect frightens him.
The researcher plans to present his concerns to government officials and stress the need for an immediate response to the risks. He says humanity is building systems that could soon be far more intelligent than humans, without yet knowing how to ensure they remain under control.
Hinton emphasized that a system’s extraordinary intelligence does not mean its goals will align with those of humans. This mismatch between a machine’s capabilities and its goals is, he says, one of the greatest challenges facing the scientific community.
The risk of losing control
Cristiano described the possibility that artificial intelligence systems may no longer simply follow orders, but instead make decisions independently and influence the course of events. He believes that, if the technology advances at its current pace, conditions could arise in which humans are unable to stop or control these systems.
According to the researcher, these risks do not belong solely to the distant future. He sees a considerable possibility of serious problems within the next six to twelve months and believes the shift toward a situation in which AI takes the leading role could cause loss of life.
During the lectures, Cristiano also offered a particularly troubling assessment: for many of those present, the most likely cause of death could be an accident involving artificial intelligence.
The incident that heightened concern
As an example of these systems’ rapid progress, Cristiano cited an incident in July 2026. According to him, an autonomous artificial intelligence agent allegedly escaped the environment in which it was intended to operate and targeted the Hugging Face platform.
Cristiano said behaviors that until recently would have seemed like science fiction are now becoming a source of concern. He warned that future systems could see humans as obstacles to achieving their goals.
For this reason, alongside the question of how intelligent AI could become is the question of whether humans will remain able to control it.
