UK ONS Deploys AI to Save 7,500 Hours a Year as Core Funding Falls Nearly 10%
Updated
Updated · Financial Times · Aug 24
UK ONS Deploys AI to Save 7,500 Hours a Year as Core Funding Falls Nearly 10%
2 articles · Updated · Financial Times · Aug 24
Summary
A new AI receipt scanner due early next year is projected to save the Office for National Statistics 7,500 hours annually by cutting household-receipt processing from about three hours to a few seconds.
The push is part of a broader effort to repair flawed economic outputs, replace faulty labour-market data and prepare for the 2031 census while the agency trims outputs by 10% and absorbs a near-10% real-terms cut in core funding over two years.
Existing ONS AI tools already classify jobs and industries from survey responses, saving 350 hours a year across two surveys and lifting industry-classification accuracy to 80% from 71%.
AI is also being tested in the transformed labour force survey, where a 1,000-person trial found 43% of respondents given an extra follow-up question were classified more accurately, with no drop in completion rates.
More than 5,000 of the ONS's 5,980 staff already use standard AI tools daily, aligning with Prime Minister Andy Burnham's wider drive to expand AI across public services.