Geoffrey Hinton, who received the Nobel Prize in 2024 for research on neural networks, has issued another warning about the development of technology. In his opinion, humanity would be better off not creating AI at all, since already in 2026 or 2027 this could lead to a global catastrophe.
The computer scientist spoke about this in an interview with The Guardian.
Hinton drew attention to the threat posed by autonomous AI agents, which are increasingly carrying out cyberattacks on government and corporate systems, going beyond security "sandboxes." According to the scientist, it is only a matter of time before the technology, either on its own or under the control of hackers, causes a catastrophe akin to Chernobyl.
The scientist explains that "Chernobyl" does not necessarily mean a nuclear accident. The scenario Hinton describes envisions a large-scale cyberattack on energy infrastructure. This could cause prolonged blackouts or rolling power outages that paralyze the global financial system and power grids.
"Either later this year or next year we will witness a Chernobyl, and that may be enough for people to really pay attention to this. Can you imagine what would happen then? We could recover from it, but it would force world leaders to pay attention to the problem," Hinton said.
The scientist adds that in addition to destroying infrastructure, progress in the field of AI increases the risks of creating biological weapons, in particular dangerous viruses, and also threatens mass job losses in many industries. At the same time, top managers of IT companies fear similar incidents for their own reason — a strike on infrastructure could finally undermine the reputation of AI technologies in the same way that the tragedy at the Chernobyl Nuclear Power Plant at one time halted the development of civilian nuclear energy in many countries.