The Alan Turing Institute, Department for Transport, and Office for National Statistics published research on June 2, 2025, identifying that around 40% of public sector time could be supported by generative AI. The study analysed 91 work activities using ONS Public Sector Time Use Survey pilot data, to assess where generative AI can meaningfully reduce administrative burden for time-pressed public sector staff.
Education workers could see the most benefit, with up to 49% of time spent on teaching activities potentially supported by generative AI. Healthcare workers show 33% potential benefit, while social care activities could see nearly half their workload supported. The research found generative AI particularly useful for roles with high administrative task levels, including routine email management where AI could decrease time spent by 70% from the current 30 minutes daily average.
Technology Secretary Peter Kyle stated: "This research shows just how far AI can go in supporting our public servants - freeing up time and energy for the work that really matters. From lesson planning in schools, drafting routine emails, and managing case files in social care, generative AI could save a significant amount of time across public sector tasks."
Specific applications include education staff spending 100 minutes daily preparing classroom lesson plans, with 75% potentially supported by generative AI. Activities like organising meetings, processing forms, and updating databases could be almost entirely supported. Time for creating or updating records, databases, case files, and documents shows near-complete support potential.
The study found 38% of frontline workers' activities could be supported compared to 46% for non-frontline workers. However, researchers emphasised careful implementation consideration for decisions directly impacting individual liberties and well-being, such as welfare access decisions, and high-stakes communication including parliamentary correspondence.
Youmna Hashem, Research Associate in AI for Public Services at the Alan Turing Institute, noted: "Our research shows that generative AI has the potential to greatly support the delivery of public sector work, assisting time-pressed staff with the completion of administrative tasks and freeing them up to focus on other elements of their jobs. However, it is vital for these technologies to be embedded in ways that are safe, responsible, and which take into account the many complexities of public sector work."
The research provides methodology for measuring generative AI adoption opportunities across different public sector contexts. Implementation requires careful consideration of human involvement, judgment, and discretion requirements, particularly in healthcare where doctors and nurses need high levels of human interaction with little room for error.
The study establishes a framework for systematic AI adoption across UK public services, potentially transforming government efficiency and service delivery. The 40% time support potential represents significant productivity gains, while highlighting the need for responsible implementation considering public sector complexities. Success could position the UK as a leader in government AI adoption, while demonstrating measurable benefits for AI vendors targeting public sector markets.