Artificial intelligence (AI) is fast becoming a central plank of UK public sector reform. From automating back-office processes to improving fraud detection and speeding up citizen services, the potential use cases are clear.
But across government, many AI initiatives are still trapped in experimentation rather than generating measurable outcomes. According to data specialist Denodo, trust is emerging as one of the biggest obstacles to wider adoption.
“Trust means different things to different people, but for government it is particularly important because it deals with sovereignty, traceability, auditability and data sharing,” said Denodo product marketing director, Errol Rodericks.
That matters particularly in government, where public bodies must balance innovation with compliance, accountability and secure collaboration across departments and agencies.
The data problem behind the AI problem
Much of the debate around public sector AI has focused on models, ethics frameworks and procurement. But Rodericks argues the real challenge sits elsewhere: fragmented data estates.
“When we dig deep, everything boils down to data,” he said. “It all comes back to the quality of the data, the availability of the data, and the need to access data without having to move it.”
Many departments still operate with siloed legacy systems, inconsistent records and limited interoperability. Without access to accurate, timely and shareable data, even the most sophisticated AI tools struggle to deliver value.
For the public sector, errors in welfare, policing or healthcare decisions can have serious consequences.
Where AI could make the biggest difference
Despite the barriers, there are several areas where well-governed AI could have immediate impact in UK government.
If you liked this content…
Think Digital Government: Key Takeaways
Civil servants need AI confidence, not just coding skills, say government leaders
How should senior leaders approach AI? Practical steps for public sector leaders
Think AI for Government: Key Takeaways
AI won’t save government until it gets control of its documents
Rodericks points to fraud and improper payments as a major opportunity, where AI can help detect anomalies across tax, benefits and procurement systems.
He also highlights operational productivity, through automating case handling, document processing and administrative workflows, alongside improved citizen outcomes through better targeting of support and earlier interventions.
“My favourite is public safeguarding and risk prevention. Those are the outcomes the public sector should be aiming to do,” he said.
Joining up data from multiple sources could help agencies identify emerging risks more quickly, from serious violence to vulnerable households needing support.
Moving beyond centralisation
One longstanding instinct in government transformation programmes has been to centralise data into a single repository before innovation can begin. But Rodericks warns that this model can slow progress.
“Whenever they bring data together, it all migrates towards a central lakehouse or some kind of repository, and that simply doesn’t work for public sector workloads,” he said.
Instead, he argues that departments should focus on securely accessing data where it resides, rather than embarking on lengthy migration programmes. That approach can reduce duplication, accelerate deployment and help public bodies comply with data-sharing obligations without the disruption of wholesale systems change.
From pilots to outcomes
For digital leaders, CIOs, chief data officers and policymakers, the question is no longer whether to explore AI. It is how to operationalise it responsibly.
Those issues will be explored in an upcoming webinar hosted by Denodo on May 26, where Rodericks and industry analyst Mike Ferguson will discuss why government AI initiatives stall and how public bodies can deliver trusted results at scale.
