Claims that AI will deliver significant productivity gains in the public sector are shaping major spending decisions, but the evidence underpinning them is often weak and unreliable, according to a new briefing from the Ada Lovelace Institute.
The thinktank warns that a single productivity estimate can drive decisions worth billions of pounds, reshape workforce planning for thousands of officers, and lock in technology choices for years – making the quality of that evidence a matter of significant public importance.
The briefing finds that most studies focus narrowly on time savings or cost reductions, while failing to account for wider outcomes such as service quality, equity, worker wellbeing or long-term value for money. It also raises concerns about industry influence on research design and the selective use of positive findings.
The Institute is calling on policymakers to demand higher standards of evidence before committing to major AI investments in public services.
