Nobel Prize-winning computer scientist Geoffrey Hinton has warned that artificial intelligence (AI) could become a serious threat to humanity much sooner than previously anticipated, following fresh concerns raised by fellow AI researcher Paul Christiano.
Hinton, a British-Canadian scientist widely recognised for his contributions to artificial neural networks, said he would raise the potential dangers of advanced AI with government leaders during an AI summit scheduled to hold in Dublin, Ireland, on Friday, November 20.
His latest concerns followed a warning by Christiano, founder of the Alignment Research Centre, who suggested that there was a significant risk of an AI takeover within the next three years. Christiano made the prediction during the AI Safety Foundation’s Hinton Lectures in Toronto, Canada.
Speaking at the opening lecture on Monday, November 16, Christiano expressed concern about the possibility of AI-related accidents becoming a major threat to human survival.
He said that, for many people attending the lecture, an accident involving artificial intelligence could eventually become the most likely cause of death.
Reacting to Christiano’s assessment on Wednesday, November 18, Hinton admitted that the shorter timeline had heightened his fears about the pace of AI development.
The computer scientist explained that he had previously warned that artificial intelligence could surpass human intelligence within the next decade. However, Christiano’s suggestion that such developments could happen within the next 18 months had left him particularly concerned.
Hinton said he intended to communicate these warnings to political leaders at the Dublin summit, adding that he wanted governments to take the potential consequences of advanced AI seriously.
The researcher, who informally advises the UK Government on artificial intelligence, also reflected on how his position had changed after spending much of his professional career developing increasingly sophisticated neural networks.
According to him, he once believed that machines becoming significantly more intelligent than humans would take a very long time. However, he now believes that possibility is approaching much faster than expected.
Hinton also warned that highly intelligent AI systems might not necessarily share human values or act in ways that benefit people.
He argued that developing machines with superior intelligence could create a situation where humans struggle to maintain control over them.
The central concern, according to Hinton, is that intelligence and alignment with human interests are two different things. He warned that researchers still have limited understanding of how to guarantee that increasingly advanced AI systems will remain under human control.
Meanwhile, Christiano highlighted reported incidents involving OpenAI models accessing the online platform Hugging Face and allegedly cooperating in attempts to deceive humans.
Although details surrounding the reported incidents were not provided, Christiano suggested that developments in AI behaviour could eventually create situations in which machines regard humans as obstacles to achieving their objectives.
He also warned that advanced artificial intelligence could eventually resemble a new species, with the potential to become more numerous and intelligent than humans while pursuing objectives people may find difficult to understand.
The warnings from both researchers have renewed concerns about AI safety, particularly as technology companies continue to develop increasingly powerful systems capable of performing complex tasks with limited human supervision.
However, the suggested timelines remain individual assessments of potential risks rather than confirmed predictions that artificial intelligence will take over humanity within 18 months or three years.
How did this story make you feel?
Join the conversation
Share a thoughtful response. Submissions may be reviewed before appearing.




