Researchers at the Alan Turing Institute have found that generative artificial intelligence could help with tasks that take up nearly half of civil servants’ time at work.
The study from the data science hub investigated the scope for AI to assist public sector workers with their everyday tasks by looking at Office for National Statistics figures for time spent on particular activities, as reported by people with specific jobs.
According to a paper published this week, tasks that take up 45% of Probation Service officials’ time have “exposure” to AI – meaning that artificial intelligence could potentially assist.
For staff at HM Courts and Tribunals Service, the proportion of potentially AI-compatible tasks was 44%. At HM Prison Service the proportion was 43%.
For the wider civil service, the Turing Institute said a slightly higher proportion of officials’ time – 47% – involved tasks that AI could assist with.
The study, which was conducted in partnership with the Department for Transport, using data from the ONS Public Sector Time Use Survey pilot, where participants recorded how they spent their time during a typical working day across 91 activities.
Researchers then assessed the extent to which generative AI could be used to complete each activity, taking into consideration different methods and tools of work as well as the level of social interaction required within specific roles.
Only four of the 91 activities were assessed as having “full exposure” to generative AI, meaning that a reduction of 100% of the human time spent on the activity might be possible. Those activities were meeting organisation and administration; creating or updating appointments and bookings; completing or processing forms; and creating or updating records, databases or case files.
The Turing Institute said public sector workers spent an average of 30 minutes a day on emails, many of which are routine and require little human involvement and oversight and that using a generative AI tool to support initial drafting could decrease this time by 70%.
Tasks undertaken by teaching staff in primary and secondary education had a high proportion of “exposure” to generative AI (49%), according to the Turing Institute. It said the figure reflected time spent planning lessons, among other tasks.
Overall, 41% of public-sector officials’ working time was found to be taken up by tasks that AI can help with. Researchers said that figure could be broken down to 38% for frontline workers and 46% for non-frontline workers.
Youmna Hashem, research associate in AI for public services at the Turing Institute, said the study showed generative AI had great potential to lift the burden of administrative tasks, freeing up time for other work.
“If the government introduces generative AI effectively and staff receive the training and assurance they need to confidently use the technology, this could meaningfully transform the way that public sector time is spent,” she said.
However, Hashem cautioned that it would be “vital” for generative AI to be embedded in ways that are safe, responsible, and which take into account the many complexities of public sector work.
Microsoft counts the minutes
Separately, the Department for Science, Innovation and Technology this week released the findings of a trial of the Microsoft 365 Copilot “AI-powered digital companion”, which 20,000 civil servants took part in – including officials at the Department for Work and Pensions and Companies House.
The study found that using generative AI for everyday tasks – like drafting documents, summarising lengthy emails, updating records, and preparing reports – saved users an average of 26 minutes per day. DSIT said that on an annual basis the time-saving would add up to two weeks a year per civil servant.
Technology secretary Peter Kyle said the trial showed AI was “a present reality” rather than a “future promise”.
“Whether it’s helping draft documents, preparing lesson plans, or cutting down on routine admin, AI tools are saving civil servants time every day,” he said.
“That means we can focus more on delivering faster, more personalised support where it really counts.”
In January this year, DSIT said the public sector had the potential to secure £45bn a year in productivity savings from better digital services.
The department subsequently confirmed that 80% of the expected dividend would come from the automation of routine tasks currently dealt with by staff.
Last week, it emerged that documents supporting the DSIT projection suggested that 62% of tasks undertaken by the lowest-ranking civil servants were considered to be “routine” and by implication suitable for automation.