Report calls on government to scale AI training across the public sector

A new research report from Google, AI Works 2025, calls for an ambitious scale-up of AI training across the UK public sector, arguing that targeted upskilling is essential to close adoption gaps and unlock productivity gains.
The study, produced with policy consultancy Public First, outlines a series of detailed recommendations specifically aimed at central and local government leaders. It draws on pilots in education, small businesses and trade unions, and finds that a few hours of hands-on, sector-specific training can increase confidence and use of AI tools – particularly among older workers and women, who are currently underrepresented in AI adoption.
The report recommends that the UK Government embed AI training as a core offering for all public sector workers, including NHS staff, local authorities, and Civil Servants. It emphasises that training must be interactive, tailored to roles, and followed up with practical policies and support to ensure lasting behavioural change.
Among the more specific Civil Service reforms proposed:
- All Fast Streamers should complete a mandatory rotation in an AI team, or the Cabinet Office should assess whether a dedicated AI Fast Stream is needed.
- An AI lead should sit on the Executive Committee of every government department, with regular training for senior leaders to ensure informed and strategic adoption.
- Public Sector AI adoption trials should focus not on digital teams, but on hard-to-reach frontline workers and back-office staff, to surface real barriers and demonstrate practical impact.
The report also urges the Government to implement:
- Departmental AI adoption audits, coordinated by a new Centre for Digital Government, to assess progress and identify where AI could reduce manual work.
- Annual AI skills and adoption surveys, under Skills England, to monitor uptake and support policy alignment with broader goals such as the Tech Track initiative to increase digital roles in government.
Why it matters: from experimentation to transformation
Public First’s modelling suggests the UK could unlock up to £400 billion in growth by 2030 through AI, but half of that is contingent on widespread workforce adoption. Crucially, the report highlights that most AI use today is worker-led, with limited organisational support.
AI Works pilots demonstrated that a few hours of targeted training could double or triple daily AI usage, reduce admin burdens by over 120 hours per year per worker, and meaningfully shift perceptions of AI from fear to optimism. In the education sector, the time savings were equivalent to a 7% staff uplift.
