Geoffrey Hinton, a leading figure in the development of modern artificial intelligence, warned that it might yet prove that humanity would have been better off never creating the technology. Speaking to The Guardian, the Nobel prize-winning computer scientist said AI could be close to triggering a large-scale incident in the near term, citing the possibility of temporary shutdowns of financial or power systems this year or next.
Hinton, often referred to as one of the “godfathers of AI” because of his pioneering work on artificial neural networks, stressed that he is not a “doomer” about the technology. He said AI also has the potential to produce a “wonderful future,” but added that “it might be that we would have been better off never developing it,” underscoring the tension he sees between AI’s benefits and its risks.
His remarks, which draw on the body of research that underpins many contemporary AI systems, add a high-profile voice to ongoing debates about how to manage and mitigate the hazards associated with increasingly capable machine learning models. Hinton’s warnings focused specifically on the risk of systemic disruptions, telling the paper that a major incident affecting critical infrastructure could happen imminently.
The intervention is likely to intensify calls from researchers, policymakers and industry figures for strengthened safety measures, oversight and further study of failure modes as AI systems are deployed more widely. While Hinton stopped short of advocating any specific policy in the interview, his dual emphasis on the technology’s promise and its potential for harm highlights the difficult trade-offs facing regulators and developers.
His comments come amid broader public and governmental scrutiny of AI, where concerns range from economic and labour impacts to cybersecurity and the integrity of essential services. Observers say striking a balance between encouraging innovation and preventing dangerous outcomes will be central to how societies respond to the rapidly evolving capabilities of AI.