Geoffrey Hinton warns AI could trigger major disruption

Geoffrey Hinton, one of the computer scientists known as the “godfathers of AI”, has warned that artificial intelligence could trigger a large-scale incident this year or next. Speaking to the Guardian, he cited a temporary blackout of financial or power systems as examples of the disruption he fears.

The Nobel prize winner said it was possible that humanity would ultimately have been better off never developing the technology. But he also rejected the description of himself as a “doomer”, saying AI retained the potential to lead to a “wonderful future”.

Hinton’s warning carries particular significance because his groundbreaking research into artificial neural networks helped underpin the technology he is now questioning. His remarks place the possibility of serious disruption alongside the promise of beneficial advances, rather than presenting either outcome as inevitable or already determined.

The examples he raised concern systems whose reliable operation is central to everyday life: finance and electricity. The supplied Guardian excerpt describes a warning about a possible future incident, not a report that an AI-related banking shutdown or power blackout has occurred. It does not identify a particular institution, electricity network or country as being at risk.

The excerpt also does not explain how AI might bring about such an incident or specify whether Hinton envisages deliberate misuse, a technical failure or some other cause. That distinction matters when interpreting his comments: the warning identifies potential consequences and a near-term timeframe, but the available material does not establish a detailed scenario or quantify its likelihood.

His broader concern extends beyond whether a single incident can be avoided. In saying “it might be that we would have been better off never developing it”, Hinton raised the possibility that AI’s overall consequences could outweigh its benefits. The wording is conditional, however, and should not be read as a conclusion that the technology has already caused more harm than good.

At the same time, his reference to a “wonderful future” leaves room for a very different outcome. Those two statements capture the tension in his assessment: the same field of research that offers substantial promise may also expose society to serious risks. His rejection of the “doomer” label is an important qualification to the warning, not a withdrawal of it.

For now, the immediate issue raised by Hinton’s comments is whether the disruption he fears materialises within the period he described. The supplied excerpt offers no detailed forecast or measures by which to assess that risk. His outlook remains explicitly uncertain: AI could help deliver a better future, but he believes a much less favourable judgment of its development remains possible.