August 25, 2026, (Inside AI) — Donald Michie, a pioneer of machine intelligence, convened an international group of experts at Villa Serbelloni on Lake Como in 1972 to address the ethical and social dangers of artificial intelligence. The gathering, known as the Serbelloni group, mapped early concerns about political tyranny, erosion of human autonomy, and widespread social coercion driven by automation. Yet their warnings were largely ignored.
The group was led by Michie, then director of the University of Edinburgh's department of machine intelligence. According to his son, Jonathan Michie, professor of innovation and knowledge exchange at the University of Oxford, the experts identified risks that remain relevant today. The failure to act, he argues, was not accidental.
In 1973, the UK government cut research funding for machine learning, robotics, and AI following the Lighthill report. That report, later documented as having been established for this purpose, pushed researchers toward more immediately industrial work. Attempts to reconvene the Serbelloni group with humanities experts proved unsuccessful.
Jonathan Michie wrote that the international gathering tackled the ethical and social implications of AI decades before the current debate. He noted that the next year's funding cuts forced a shift in priorities. The early warnings about automation and coercion were left without a sustained research agenda.
Why Early AI Warnings Lost Their Audience
The Lighthill report, commissioned by the Science Research Council, was highly critical of AI research in Britain. It argued that AI had failed to deliver on its promises and recommended severe funding reductions. The result was a period now known as the AI winter, which stalled progress and silenced many voices.
Funding cuts were not the only factor. The Serbelloni group lacked institutional backing to continue its work. Without support from governments or universities, its findings faded from public view. The group's early mapping of risks such as political tyranny and loss of autonomy was never developed into policy.
Jonathan Michie's account suggests that the window for early intervention closed quickly. The industrial focus that followed left little room for ethical inquiry. Researchers who might have pursued these questions were redirected to short-term applications.
The Machine's Limits and Human Knowledge
Callum Brown, professor emeritus in history at the University of Glasgow, offers a different perspective. He argues that AI is not as intelligent as its promoters claim. A few well-chosen puzzles, he says, expose its crippling obsession with data rather than knowledge.
Brown connects the AI debate to the work of Lewis Mumford, an American intellectual who spent decades warning about the machine's impact on human purpose. Mumford corresponded with British humanist Frederic J. Osborn for over three decades, developing ideas about resisting induced passivity. Brown writes that the automatic machine has the capacity to make man passive and purposeless, much like living in an uninhibited city.
Mumford saw the city itself as a form of machine damaging people by the million. His warnings about technological passivity echo in current debates about AI. Brown insists that humans must resist induced passivity of all kinds, and that the myth of AI's superior intelligence still prevails.
The historical record shows that early warnings about AI were not only technical but deeply social. The Serbelloni group identified risks that are now central to global policy discussions. Yet the institutional and financial forces of the time prevented those warnings from taking root.
Today, as governments and corporations race to deploy AI systems, the lessons from 1972 remain relevant. The failure to listen then was not due to a lack of expertise. It was a failure of funding, attention, and political will. Whether the current generation of warnings will fare better remains an open question.
