The UK House of Commons Public Accounts Committee has published a new report on ‘Use of AI in Government’ (2024-25, HC 356).
The report focuses on the specific situation in the UK and addresses issues closely related to the UK Government’s current ambitions to quickly roll out AI across the public sector.
However, most recommendations target general obstacles and pitfalls for AI deployment, acquisition, and assurance, and will thus be of interest in other countries.
The key conclusions of the report — which I would bet are largely applicable to most countries — include:
Out–of–date legacy technology and poor data quality and data–sharing is putting AI adoption in the public sector at risk.
Public trust is being jeopardised by slow progress on embedding transparency and establishing robust standards for AI adoption in the public sector.
There are persistent digital skills shortages in the public sector and current plans to address the skills gap may not be enough.
There is no systematic mechanism for bringing together learning from (failed) pilots and there are few examples of successful at–scale adoption across government.
There is a a long way to go to strengthen government’s approach to digital procurement to ensure value for money and a thriving AI supplier market.
Realising the benefits of AI across the public sector will require strong leadership.
The key recommendations in the report focus on the need to:
Deal with legacy technology and ICT systems before AI is overlaid on it.
Address the risks resulting from barriers to data–sharing and poor data quality.
Boost compliance with algorithmic transparency and disclosure requirements.
Strengthen spend controls for high–risk AI use cases to support safe and ethical roll–out.
Put effective plans in place to boost public sector digital skills sustainably.
Set up a mechanism for systematically gathering and disseminating intelligence on pilots and their evaluation.
Set out how to will identify common and scalable AI products and support their development and roll–out at scale.
Develop an effective procurement strategy that leverages buying power to the possible extent.
Ensure those taking procurement decisions across government have access to the right digital skills and knowledge.
Develop effective governance, leadership and ownership within central government.
Somehow, I am glad to see that these recommendations directly map onto the same areas of concern I have been highlighting in my recent research (eg here, here and here) and talks about these issues. The big question now is whether the (UK) government will find ways to meaningfully address (and fund!) the changes required if AI readiness in a real, practical sense is to be brought closer to the aspirations surrounding public sector AI use.
