Skip to content
Citizen Infrastructure

Manchester People's Panel for AI (2022–2025)

TL;DR. The People’s Panel for AI is a Manchester model for putting residents — especially those from digitally-excluded communities — in a position to hold public-sector AI to account. It began as a Manchester Metropolitan University research pilot (2022–23, funded by the Alan Turing Institute) and was adopted by Manchester City Council as part of its digital strategy (2024–25). Residents are trained on AI and data ethics, then sit across the table from the service leads and businesses proposing AI use-cases, who must defend them under structured questioning. It is less a one-off assembly than a recurring scrutiny mechanism for public-service AI — and it deliberately recruits the people usually left out.

At a glance

Section titled “At a glance”
WhenMMU pilot July 2022 – March 2023; Manchester City Council phase 2024–2025 (ongoing)
WhereGreater Manchester — Salford and Stockport (pilot), then the city of Manchester
WhoResidents recruited from digitally-excluded communities via targeted “AI roadshows” — not random sortition. The MMU pilot seated 9 panellists; the council phase ran 5 roadshows reaching more than 40 residents
What they didTwo days of training on AI, data, and ethics, then scrutiny of real AI use-cases: a service lead or business pitches a use-case and answers the panel’s questions, using “consequence scanning” to surface ethical risks
OrganisersManchester Metropolitan University (Dr Annabel Latham, Prof Keeley Crockett), funded by the Alan Turing Institute; then Manchester City Council’s digital-strategy team
OutcomePanel recommendations feed how participating services deploy AI; businesses reported changing design and ethics practices; a reusable Terms of Reference was published

What happened

Section titled “What happened”

The People’s Panel for AI began as a research project at Manchester Metropolitan University (July 2022 – March 2023), led by Dr Annabel Latham and Prof Keeley Crockett and funded by the Alan Turing Institute. The premise: diverse citizen voices — especially from marginalised communities — are largely absent from how AI is researched, built, and deployed. The team ran interactive “AI roadshows” with communities in Salford and Stockport, recruited nine residents into a panel, gave them two days of training on data, AI, and ethics, and convened four sessions in which tech businesses and researchers pitched their products and were questioned by the panel. Panellists came away more confident challenging AI systems; some businesses changed their design and ethics practices in response. Crucially, the team co-wrote a Terms of Reference (with the panel and the Greater Manchester Equality Alliance) so the model could be reused.

Manchester City Council then adopted that model. As part of its “Doing Digital Together” digital strategy — with a stated goal of becoming a world-leading digital city by 2026 — the council ran the panel at city scale, deliberately centring residents most at risk of digital exclusion (Greater Manchester has an estimated 1.2 million residents at risk, 450,000 without internet access). Working from the Digital Exclusion Risk Index, it held five free roadshows in community spaces across Manchester in May–June 2024, reaching more than 40 residents from a deliberately diverse range of backgrounds; pre- and post-session surveys showed understanding of, trust in, and confidence around AI each rising by about 30%. Residents who wanted to go further joined the panel, took the two-day training, and then heard service leads pitch real AI use-cases — using consequence scanning, a structured technique for surfacing a technology’s unintended effects — and pressed them with questions and concerns. Services including Transport for Greater Manchester, Tech Enabled Care, and Citizens Advice took part, and the panel’s recommendations are meant to feed back into how those services deploy AI. The council has continued developing the panel through 2025.

Why it matters

Section titled “Why it matters”

Most “AI and democracy” experiments use AI to help citizens deliberate. Manchester’s panel inverts that: it uses deliberation to hold AI to account — residents, not algorithms, sit in judgement of how the public sector wants to use the technology. Two features make it worth recording. First, it deliberately recruits from digitally-excluded communities — the people most exposed to AI’s harms and least represented in its design — rather than a random cross-section, treating inclusion as the point. Second, it pairs scrutiny with capability-building: residents are trained before they judge, so the panel doubles as digital-literacy infrastructure. As a recurring, reusable model with a public Terms of Reference, it is an early template for civic oversight of public-sector AI — closer to a standing citizens’ panel for technology than a one-off assembly.

Sources

Section titled “Sources”