A quarter of UK civil servants say artificial intelligence tools are already saving them time at work, according to the latest Civil Service People Survey.
Results from the survey, published by the Cabinet Office, show that 27% of employees believe using AI tools saves them at least one hour per week. The questions on AI were included in the survey for the first time, reflecting growing interest in how the technology is reshaping the public sector workforce.
The findings suggest productivity gains are already emerging from early adoption of AI tools across Whitehall, though the benefits appear uneven between departments.
The survey highlights significant differences in reported time savings across government organisations. Teams at the Department for Science, Innovation and Technology (DSIT) were the most likely to report gains from AI use, with 57% saying the technology saves them at least an hour each week.
By contrast, staff at larger operational departments reported lower levels of benefit. At HM Revenue and Customs and the Department for Work and Pensions (DWP), around 22% of respondents said AI tools delivered the same level of time savings.
The variation is attributed to differing levels of access to tools, digital capability, and the nature of departmental work, with policy and analytical roles typically easier to augment with generative AI than large-scale operational delivery functions.
Alongside questions about time savings, the survey also asked civil servants about the likely long-term impact of the technology on their work. Across the civil service, 76% of respondents said it was likely that AI would change how they perform their current role within the next five years.
The survey also found that 74% of civil servants believe they would be supported if they tried a new idea to transform how work is done, suggesting a relatively strong appetite for experimentation across departments.
The survey itself is one of the largest workforce exercises in government. In 2025, 343,961 officials responded out of 584,718 invited, giving a response rate of 59%.