Key Takeaways
- Jill Lepore’s August 18 Democracy v the machine article highlights a long‑standing failure to heed early warnings about AI’s societal dangers.
- In 1972 an international group of experts, the “Serbelloni group,” convened at Villa Serbelloni on Lake Como to map ethical and political risks of AI, including political tyranny, loss of human autonomy, and automation‑driven social coercion.
- The group was led by Donald Michie, director of the University of Edinburgh’s Department of Machine Intelligence, whose son Jonathan Michie recalls that the experts “mapped out early concerns regarding potential political tyranny, the erosion of human autonomy and widespread social coercion driven by automation.”
- The following year the UK cut funding for machine‑learning, robotics and AI after the Lighthill report, steering researchers toward more immediately industrial‑relevant work and preventing the Serbelloni group from reconvening with humanities scholars.
- Historian Lewis Mumford and town‑planner Frederic J. Osborn exchanged letters for over three decades, warning that automatic machines could render people passive and purposeless, a danger they saw mirrored in unchecked urbanisation.
- Contemporary AI, while powerful, remains “not that smart” and exhibits a “crippling obsession with data” rather than the kind of knowledge that characterises human understanding, as illustrated by simple puzzles that expose its limits.
- The Guardian invites readers to send letters in response to these pieces, underscoring the ongoing public debate about AI’s role in democracy and society.
Jill Lepore’s critique of AI and the warning ignored
Jill Lepore’s essay, “Democracy v the machine,” published in the Guardian on 18 August, asks why society failed to listen to early warnings about the threats posed by the AI revolution. As a Harvard professor, Lepore’s analysis is inevitably US‑centric, but she situates the current debate within a longer lineage of concern about technology’s power to reshape politics and everyday life. She writes, “why nobody listened to warnings about the threats posed from the AI revolution,” a question that frames the subsequent historical reflections offered by readers. Lepore’s piece serves as a catalyst for revisiting moments when scholars and policymakers could have steered AI development toward more humane outcomes.
The Serbelloni gathering and early AI ethics concerns
In response to Lepore, Jonathan Michie, Professor of Innovation & Knowledge Exchange at the University of Oxford, recalls a pivotal 1972 meeting that predates today’s AI hype. He notes, “In 1972 there was an international gathering of AI researchers and others in Italy to tackle the ethical and social implications of AI.” This “Serbelloni group” met at Villa Serbelloni on Lake Como and was led by his father, Donald Michie, then director of the University of Edinburgh’s Department of Machine Intelligence. According to Jonathan, the group “mapped out early concerns regarding potential political tyranny, the erosion of human autonomy and widespread social coercion driven by automation.” Their deliberations anticipated many of the anxieties now voiced about algorithmic governance, surveillance, and the concentration of power in tech platforms.
How the Lighthill report derailed AI research
The optimism of the Serbelloni conference was short‑lived. Jonathan Michie explains that the very next year the UK government slashed funding for machine learning, robotics and AI, a move directly tied to the controversial Lighthill report. He states, “the next year the UK cut research funding for machine learning, robotics and AI, following the Lighthill report, which had been established – it has since been documented – for this purpose.” The report, commissioned to assess the state of AI research, concluded that the field had overpromised and underdelivered, prompting policymakers to redirect scarce resources toward projects with immediate industrial applications. Consequently, the broader ethical and societal discussions initiated at Serbelloni were marginalised as researchers pursued narrowly technical goals.
Attempts to broaden the dialogue with humanities fell short
Jonathan Michie further laments that his father’s effort to revive the Serbelloni conversation with a stronger humanities presence proved unsuccessful. He writes, “My father’s attempts to reconvene the Serbelloni group to include experts from the humanities thus proved unsuccessful.” The withdrawal of funding and the shift toward applied AI limited opportunities for interdisciplinary exchange, leaving philosophical, political, and cultural critiques of AI under‑represented in the research agenda. This gap, he implies, contributed to the later emergence of AI systems that optimise for efficiency or profit without sufficient scrutiny of their impact on democratic institutions and individual agency.
Mumford and Osborn on the dehumanizing power of machines
Callum Brown, Professor Emeritus in History at the University of Glasgow, draws a line from the early AI debates to the intellectual legacy of Lewis Mumford and Frederic J. Osborn. He recalls that Mumford, an American intellectual who devoted his life to restraining the uninhibited city, “hooked up with the British humanist Frederic J Osborn in a letter exchange that lasted over three decades, developing a sane town‑planning system.” Brown quotes Mumford’s insight: “the automatic machine had the capacity to make man passive and purposeless in the same manner as did living in the city, and humans must resist induced passivity of all kinds.” This warning—that automation could erode human purpose and civic engagement—echoes the Serbelloni group’s fears about political tyranny and social coercion, suggesting that the dilemma is not new but has merely changed technological guise.
Contemporary AI: data obsession versus human knowledge
Brown brings the historical critique into the present, arguing that today’s AI, despite its impressive performance, remains fundamentally limited. He observes, “Beware – AI is really not that smart, and a few well‑chosen puzzles for it to solve expose its crippling obsession with data, not with the humans’ forte, knowledge.” By pointing to simple tasks that trip up state‑of‑the‑art language models—such as reasoning about causality or grasping contextual nuance—he underscores that AI’s strength lies in pattern recognition within massive datasets, not in the kind of abstract, value‑laden understanding that humans bring to ethical decision‑making. This observation reinforces Lepore’s call to heed earlier warnings: without embedding humanistic insight into AI design, we risk amplifying the very passivity and erosion of autonomy that Mumford warned against.
Invitation for reader responses
The Guardian closes the correspondence with an invitation for public participation, reminding readers that the debate is ongoing. The note reads, “Have an opinion on anything you’ve read in the Guardian today? Please email us your letter and it will be considered for publication in our letters section.” This call‑to‑action reflects the democratic spirit that Lepore, Michie, Brown, and the historical figures they reference all seek to preserve: a society where citizens, scholars, and policymakers can collectively scrutinise and shape the trajectory of powerful technologies like AI.
https://www.theguardian.com/technology/2026/aug/25/why-the-threat-from-the-rampaging-ai-machine-went-ignored