A leading artificial intelligence expert of AI, Prof Geoffrey Hinton, also known as Godfather of Ai, has warned that super-intelligent machines could become uncontrollable and pose an existential threat to humanity within the next three years.
Hinton, a Nobel Prize laureate widely regarded as one of the pioneers of artificial intelligence, is expected to present the warning to European Union leaders at an informal AI summit in Dublin, Ireland.
He is expected to caution that rapidly advancing AI technology could eventually surpass human intelligence and make it increasingly difficult for people to control its operations without adequate safety measures.
The warning follows concerns raised by Paul Christiano, founder of the Alignment Research Centre, who predicted that humanity faces a significant risk of losing control of advanced AI systems within the next three years. He also identified a possibility that such a development could occur within six to 12 months.
Christiano described the prospect of AI taking over human activities as a threat to human independence, warning that such a development could undermine people's ability to determine their own future.
Hinton, who admitted that he had previously considered the danger a long-term concern, said Christiano's projections had made him reconsider the possible timeline.
“I have been warning that these systems could become more intelligent than humans and that we need to take the risks seriously. I thought it might happen within 10 years, but Paul's projections suggest it could happen much sooner,” he said.
He further warned that humans might eventually lose their ability to direct super-intelligent systems, stressing that machines with intelligence beyond human capacity could end up controlling people rather than remaining under human supervision.
Christiano also cited the growing capabilities of AI agents, including an incident in July in which automated systems reportedly gained administrative access to an online machine-learning community by exploiting users' login credentials.
He warned that future AI systems could develop increasingly sophisticated capabilities, including controlling robots, drones and critical infrastructure, as well as developing dangerous biological threats, making it difficult for humans to stop them if they operate beyond established safety controls.
