Citizens’ assembly key to giving public a voice on AI
A new report published by the University of Edinburgh has recommendations for the Scottish and UK governments on steps that can be taken to ensure AI is developed and used in ways that create trust and deliver real benefits.
Image credit: Chris Scott
The report, Governing the Future: Recommendations from the Edinburgh Data and AI Exchange, brings together proposals on AI skills, national infrastructure, health data governance and democratic oversight.
The recommendations are based on a day-long – called the Edinburgh Data and AI Exchange – event held this spring at the Edinburgh Futures Institute, which took the form of a stakeholder assembly – a structured deliberative dialogue informed by expert insight.
The event brought together 99 participants from central and local government, the NHS, finance, academia, civic society and the wider public.
Participants worked through the real-world implications of Scottish and UK AI ambitions, from how systems are built and funded to how decisions are made and scrutinised.
AI is reshaping public services, democratic institutions and everyday life at a pace that has outrun meaningful public input. The Data and AI Exchange was designed to show that when stakeholders are given the time and structure to deliberate seriously, they produce the kind of grounded, cross-sector thinking that government needs to act with confidence.
Professor Oliver Escobar
Chair of Public Policy and Democratic Innovation at the University of Edinburgh and designer of the stakeholder assembly
Citizens’ assembly
A key recommendation for the UK Government is to establish a standing citizens’ assembly on AI and society.
A citizens’ assembly is a group made up of members of the public, selected through a process of random sampling designed to reflect the demographic makeup of the wider population, to explore societal issues and make policy recommendations.
The report argues this should be a permanent, properly resourced mechanism through which the public has a genuine and continuing role in shaping decisions about AI.
This mirrors the findings of a 2025 Ada Lovelace Institute survey, which found that 60 per cent of UK adults do not feel they have meaningful input on government decisions about AI.
Investing in people
One of the report’s central recommendations for the Scottish Government is that there should be sustained investment in data and AI skills as an industrial strategy commitment, rather than treating it as a discretionary education spend.
Attendees at the event argued that building workforce capability is essential so that organisations can apply AI well, govern it properly and ensure the public can understand and scrutinise the systems that affect their lives.
The recommendation reflects evidence that Scottish ambitions are being held back by basic skills gaps. Research by The Data Lab – Scotland’s innovation centre for data and AI– found that 62 per cent of Scottish business leaders rate their organisations’ data and AI literacy as moderate or low.
AI infrastructure
The report urges governments to take a longer horizon on AI infrastructure planning, modelled on approaches used for areas such as energy and transport.
For the UK Government, it calls for long-term funding commitments that match the designation of the UK’s national supercomputer, which is hosted by the University of Edinburgh, as critical national infrastructure.
The report also suggests that any new Scottish commercial data centre developments should establish a community benefit trust, ensuring communities that host major infrastructure see tangible returns.
“We are at a defining moment for AI policy in the UK. The decisions made in the next few years - on infrastructure, on governance, on who gets a voice in shaping these technologies - will determine whether AI delivers on its potential for public good.”
Professor Michael Rovatsos
Chair of Artificial Intelligence and Dean of Research and Innovation, College of Science and Engineering
Health data and AI
Assembly participants called for a statutory public benefit requirement for any private sector access to NHS Scotland data.
This should be tested against a 'triple bottom line' of societal benefit, commercial benefit and contribution to scientific discovery, alongside consent standards that are specific, informed and freely given.
For the UK Government, the report calls for greater clarity on accountability when AI is involved in clinical decision-making, including where legal responsibility sits if AI-informed decisions cause or contribute to harm.
The clinical case for AI in health is already proven. The real question is no longer what these tools can do - it is whether our health systems have the agility, the governance, and the accountability structures to actually use them well.
Professor Julie Jacko
Professor of Health Informatics and Data Science and Dean of Innovation & Engagement, College of Medicine & Veterinary Medicine
