A recent UK Government AI experiment revealed that certain tasks can be fully automated, saving up to 13 working days per person annually.

The results are documented in two reports: Mapping the Potential by the Alan Turing Institute, and The Government Digital Service’s cross-government experiment of Microsoft 365 Copilot

Between September and December 2024, 20,000 UK civil servants participated in the trial. The findings identified four fully automatable activities:

• Organising meetings and meeting administration
• Creating or updating appointments or bookings
• Completing or processing forms
• Creating or updating records, databases, or case files

Key Insights from the Experiment

• 82% of users expressed a preference not to return to pre-AI workflows.
• Time savings were highest in:
  • Document drafting: up to 24 minutes/day
  • Presentation preparation: 19 minutes/day
• On average, 13 working days were saved per person annually—equivalent to 1,000 working years saved across all participants (assuming a 260-day working year).

Economic Implications

The 2024 median annual salary for a UK civil servant was £33,980. For 20,000 staff, the total annual payroll equates to: 20,000 × £33,980 = £679,600,000

Assuming a 260-day working year, the daily cost per civil servant is: £33,980 ÷ 260 ≈ £130.69

A 13-day annual saving per person translates to: 13 × £130.69 ≈ £1,699 per person/year

For 20,000 civil servants, the total annual saving is: 20,000 × £1,699 ≈ £33,980,000

Some Policy Implications

• Reallocation of budgetary resources to other priority areas
• Reskilling staff for roles where AI still requires a human-in-the-loop
• Faster public service delivery, enabled by freeing up 13 working days per employee

This AI experiment demonstrates tangible productivity gains but also invites deeper reflection on how public-sector roles might evolve in the years to come.