Public sector organizations across the UK remain wary of AI’s ability to deliver short-term productivity gains, despite sustained investment and political pressure to use the technology to improve efficiency.
This is according to new research from Snowflake, based on a survey of 500 senior decision makers conducted by YouGov.
The findings highlight the widening gap between ambition and execution, with only 23 per cent of UK organizations reporting productivity gains from AI at scale, and 45 per cent seeing benefits limited to pilots or narrow use cases.
The report comes as ministers continue to place the UK at the heart of economic growth and public service reform, including an ambition to boost the UK economy by £47bn a year through the introduction of AI.
The public sector takes a cautious approach
The research suggests that the public sector is taking a more cautious and governance-driven approach to AI, and that productivity improvements are expected to take longer than in some other sectors.
More than half (52%) of public sector leaders said AI will not significantly improve productivity for at least two years, and 66% reported that ethics and safety considerations will largely drive adoption decisions.
Concerns about trust are also prominent, with 53% saying the safety of AI output is a major barrier to trust.
Although this warning is in line with regulatory and public accountability requirements, it may delay the realization of efficiency gains compared to less regulated sectors.
Structural barriers outweigh technology challenges
The report finds that across all sectors, the main obstacles to scaling AI are organizational rather than technical.
If you like this content…
Key barriers to progress include lack of skills, poor data quality, siled teams, and unclear ownership. Only 19% of respondents identified technology itself as a major constraint.
Responsibility for AI is often fragmented between senior executives, IT teams, and data teams, limiting accountability and slowing decision-making. This is a particularly acute challenge in complex public sector organizations.
Just 24% of respondents said they prioritize their AI efforts using a clear framework that aligns with their organization’s goals.
Investment continues despite limited returns
Despite these challenges, trust in AI remains high. Most organizations expect investment to increase over the next 12 to 24 months, but only 1% plan to reduce spending.
Cost reduction is now the primary measure of success, cited by 44% of respondents, with 26% prioritizing revenue growth.
Dr Fabien Stefani, who contributed to the study, said the findings reflected the typical lag between technological progress and measurable productivity gains.
He noted that organizations need time to adapt their workflows, governance models, and employee skills before the benefits are realized at scale.
“The power of AI systems depends on the ability of the people who develop, apply and manage them,” he said, noting that the continued shortage of AI-related skills is a significant constraint.
