Why AI in government is focused on services, not savings

AI is now firmly on the agenda across UK government, but its role is more measured and practical than some headlines suggest. Exclusive new research by Government Transformation Magazine, carried out in partnership with IBM, shows that public sector leaders are prioritising AI where it can deliver visible service improvements, while its use as a tool for cloud cost efficiency remains limited.

The findings are based on a survey of more than 100 senior digital, data and technology decision-makers across central government, alongside interviews with leaders from organisations including HM Revenue & Customs, the National Audit Office, Innovate UK and the North Sea Transition Authority.

Pragmatic adoption over hype

The research shows that AI is already in use across much of government. Two thirds of operational respondents say their organisation is using AI or automation in some form. However, these deployments are concentrated in areas with clear and immediate benefits, such as citizen-facing services and internal process efficiency.

At HM Revenue & Customs, for example, AI is being applied to high-volume interactions and internal workflows. James Mitton, Director General for Enterprise Transformation, describes in the report how digital assistants are now handling millions of interactions each day, while AI-supported tools are reducing administrative burdens for frontline staff.

This reflects a wider trend identified in the research. Strategic leaders see AI primarily as a way to improve service quality, manage demand and support staff, rather than as a lever for infrastructure optimisation.

Cloud efficiency remains a blind spot

Despite rising concern about cloud costs, only 19% of respondents say they are using AI or automation to optimise cloud usage directly. Techniques such as predictive scaling, anomaly detection and automated cost control remain the exception rather than the norm.

This is not due to a lack of awareness. Interviewees consistently point to structural barriers. Esra Kasapoglu, Executive Director of AI and Data Economy at Innovate UK, highlights the importance of data quality and trust, noting that AI can only scale effectively when its outputs are well understood and underpinned by reliable data.

Legacy infrastructure also continues to constrain progress. According to Yvonne Gallagher, Director of Digital at the National Audit Office, cloud migration alone does not guarantee value. Without rethinking services and demonstrating how time saved translates into money saved, efficiency gains remain difficult to evidence.

Confidence depends on proof

Senior leaders remain optimistic. More than three quarters of strategic respondents believe that genuine cloud efficiencies could be reinvested into better services or operational improvements. But that confidence is conditional.

Operational teams are clear that without credible, service-level evidence of savings, it is hard to justify further investment in AI-driven optimisation. As Nic Granger, Chief Financial Officer and CIO at the North Sea Transition Authority, notes, government needs to demonstrate that efficiencies are real and repeatable, not theoretical.

The research suggests that the next phase of AI adoption in government will depend less on experimentation and more on integration. Applying AI to cloud efficiency will require clearer ownership, stronger business cases and closer alignment between strategic ambition and operational delivery.

AI is already reshaping how public services are delivered. Whether it can also help government manage the cost and complexity of its digital infrastructure will be one of the defining challenges of the next few years.

Download the full report, Unlocking Cloud Efficiency for Smarter Government, to explore the findings and recommendations in detail.