AI for public good? Unpacking the data sharing and regulatory aspects of the AI Opportunities Action Plan.
Yesterday saw the publication of the AI Opportunities Action Plan. Authored by Matt Clifford, AI Advisor to Number 10, the plan sets out how the UK the harness the potential AI for public good, improving public services and kickstarting economic growth. With estimates that AI could grow the UK economy by an additional £400 billion by 2030 through enhancing innovation and productivity, there’s no wonder the government sees AI as “the single biggest lever” to deliver on its manifesto promises.
AI for good?
The vision contained in the Action Plan is one where the UK builds on its global leadership role in AI safety and governance, pioneers public sector AI product and service development in partnership with the private sector and builds both the compute and talent infrastructure required to harness the AI opportunity.
The ultimate goal of these efforts is to harness the potential of AI to improve people’s lives, especially when it comes to making public services better. As the Prime Minister said in his launch address: “The prize is within our grasp, and the prize is national renewal”.
Whilst Smart Data Foundry does not utilise private sector or public sector data for AI development, the Action Plan does contain two areas of particular interest which align with our core mission of using data for societal benefit:
The creation of a National Data Library from public sector data sources.
Fostering trust with proportionate regulation.
Establishing a National Data Library to improve access to valuable data assets.
There has long been political will to establish and grow a National Data Library for the UK. Public bodies hold huge amounts of data that are of benefit to academic research, policy makers, other public bodies and third sector organisations. The EU’s data.europa open data scheme has now been running for 10 years, and contains over 1.8 million datasets from across the member states.
A UK library would follow a similar model, bringing together public sector datasets from local councils, environmental bodies, the NHS, housing bodies and more. Currently, researchers face challenges in knowing what data is available to them and navigating siloed and complex data infrastructure to access data assets. The Action Plan proposes starting with five high-impact public data sets, but does not set out what these could be.
Bringing public sector data assets into a centralised library will ease discovery and create new opportunities for innovative research that utilises the ability to combine datasets from disparate sources. This is also of benefit to public sector AI innovation, potentially providing the nuanced and relevant data required to train AI tools which will improve public services.
The creation of a National Data Library also creates opportunities for research which combines public and private sector data and unlock insights which can tackle some of the UK’s most pressing challenges. Recommendation 12 proposes incentivising academia and industry to unlock more private sector datasets, and it appears that integrating public and private data sources for research and innovation could be a longer-term goal for the National Data Library. There is a wealth of private sector data which, when responsibly and securely shared with appropriate third parties in the public sector, academia and third sectors, can be put to work to enhance publicly available data; for example, see our work with East Renfrewshire Council and the Joseph Rowntree Foundation.
To support this longer-term goal, the paper proposes the creation of a new unit, UK Sovereign AI, which will spearhead public/private sector AI collaboration. Part of its remit will be to “package and provide responsible access to the most valuable UK-owned datasets and relevant research”.
The government’s response to the Action Plan states that “we will responsibly, securely and ethically unlock the value of public sector data assets to support AI research and innovation”. Details on the practicalities of the National Data Library are set to be published by Summer 2025, and we anticipate that these details will include how data will be processed and held responsibly, securely and ethically, and what data assets the government wishes to add to the library.
Regulation is essential to foster public trust of AI
Trust is critical when it comes to AI. The public want to see improvements to public services and economic growth, but they must trust that AI developed to deliver this is using their data in a proportionate, secure and ethical manner.
There are some technical solutions which can aid safe and ethical AI development, including synthetic data and regulatory sandboxes, but these are a tool rather than a strategy for ethical innovation.
The AI Opportunities Action Plan recognises that regulation does not have to mean stifling innovation; “clear rules provide clarity to businesses so they have the confidence to invest and bring new products and services to market.” To bring this to fruition, government departments will ensure that guidance on safe AI innovation is included in their strategic guidance to regulators. This is already in progress; last October the government set up the Regulatory Innovation Office (RIO). RIO will be empowered by the Department for Business and Trade (DBT) and Department for Science, Innovation, and Technology (DSIT) with the goal of driving the innovation agenda forward with regulators.
The impact of RIO, and the extent to which regulators adapt to this drive for innovation, remains to be seen. But what is clear is that the government is mindful of the need to balance risk with providing an innovation ecosystem which can move at the pace required to deliver the impact it’s looking for.
Final thoughts
The government has strongly endorsed rapid yet responsible AI development and adoption, pioneered via the public sector. The AI Opportunities Action Plan provides a comprehensive blueprint for it to follow. The realities of how it is implemented will no doubt be softened slightly by the parliamentary process, with the regulatory aspect likely to see spirited cross-bench debate.
However, at the heart of this plan is the desire to use data to address some of our biggest challenges. This is a significant moment for all in the data and AI space, and we look forward to contributing to this important journey.
