OMM Newsroom
Nobel laureate Geoffrey Hinton has opposed accelerating the development of artificial superintelligence. He said there was not yet sufficient scientific evidence to ensure the safety and control of systems more capable than humans.
Hinton: Starting a superintelligence race would be “very foolish”
In an interview with The Times, the British-Canadian computer scientist, who is considered one of the pioneers of modern artificial intelligence, criticized beginning work on superintelligence before safety issues had been resolved.
“It would be very foolish to start developing superintelligence now. There is no scientific consensus on whether it can be created safely and kept under control,” Hinton said.
He said the main risk was not merely that powerful artificial intelligence systems could make mistakes. Systems more intelligent than humans could escape control and potentially become impossible to manage, posing a serious threat to humanity.
Risk of human extinction estimated at 10-20%
Hinton previously estimated the probability that artificial intelligence could cause the extinction of humanity within the next 30 years at between 10% and 20%. He based the forecast on the fact that technological development was advancing faster than expected.
The scientist believes humanity has no experience interacting with beings more intelligent than itself. For that reason, the emergence of machines that surpass human cognitive abilities would be fundamentally different from previous technological innovations.
Hinton also rejects the view that modern AI systems merely reproduce patterns found in their training data. In his assessment, advanced language models are capable of understanding semantic and abstract concepts.
Call for a more active government role
Hinton believes oversight of advanced artificial intelligence development should not be left solely to technology companies’ self-regulation. In his view, governments and regulatory bodies should determine the direction of the process.
The scientist compared technological innovation to a car’s accelerator and government regulation to the steering mechanism that determines the vehicle’s direction. He also said he did not believe market forces alone would eliminate the risks created by AI.
According to Hinton, commercial competition encourages companies to build increasingly powerful systems. International competition may also lead companies to avoid spending time on safety measures. He said the problem had become even more acute amid technological competition between China and Western countries.
Hinton’s work on artificial intelligence
Hinton worked at Google for many years and took part in research that contributed to the rapid development of modern machine learning. In 2023, he left the company, saying he wanted to speak more freely about the risks posed by artificial intelligence.
In 2024, Hinton and John Hopfield were awarded the Nobel Prize in Physics for discoveries and inventions that enabled machine learning with artificial neural networks.
Hinton believes humanity can still influence the development and management of increasingly powerful artificial intelligence. However, he said the race among companies to create systems that surpass human capabilities could reduce the time and opportunity available to establish safety mechanisms.
