The Office for National Statistics (ONS) is leaning on AI tools to improve the quality of its data and reduce its operating costs, according to the Financial Times.
The paper reported that amid a real-terms spending cut of almost 10 per cent over the next two years, the ONS is looking to boost the productivity of staff by applying AI to its survey production process.
New measures will include using AI to process household receipts used in the Living Costs and Food Survey, an annual measure of consumer spending patterns and the cost of living.
Andrew Banks, lead data scientist at the ONS, told the FT that AI will complete this task in seconds, down from the average of three hours ONS workers currently spend manually checking each family’s receipts.
He added that this use case alone would save around 7,500 hours of work each year.
James Benford, director-general for economic statistics at the ONS, told the FT: “There are big quality issues that are still outstanding and we’ve got to implement new standards for GDP and deliver the next census, transformed labour force survey, and statistical business register.”
“Using AI in the production of statistics is helping us do more, but it’s also improving quality,” he added.
In May, the ONS announced ex-Barclays chief data officer Luke Ashton as its new director general for digital, data and technology. Ashton was given a remit to digitally transform the ONS, overseeing its transition away from legacy systems and assessing the potential for AI implementation at the organisation.
The ONS already uses Google’s AI model Gemini for automatic classification of survey responses. Banks told the FT this had helped take the accuracy of industry classification in surveys from 71 per cent to 80 per cent, and saved staff 350 hours of manual work.
Benford and Ashton are overseeing a more widespread application of the technology across the survey process. The FT reported that 120 data scientists at the organisation now use AI assistants for coding tasks daily and 83 per cent of all ONS staff use generative AI tools.
Recent months have seen the ONS admit to major errors in its current survey process, which have skewed vital overviews of the UK’s economic standing. In June, Benford admitted that data collection errors for the Labour Force Survey in May and early June had lowered response rates by 19 per cent.
This has forced the ONS to estimate more data, reducing the overall accuracy of UK labour market data in the short term.
