The proportion of British businesses with ten or more employees using artificial intelligence has almost tripled in the last few years, but adoption of the technology has been "relatively shallow", with little evidence yet of widespread headcount reductions, a new report from the Office for National Statistics (ONS) has found.

According to the ONS' Artificial intelligence in UK businesses: 2023 to 2026 report, self-reported use of AI in UK businesses with ten or more employees has increased from around 12% to around 35% since late 2023, with larger firms more likely to have adopted AI.

Almost half of businesses with 250 or more employees reported using AI, compared with 28% of firms with fewer than ten employees.

The technology has yet to become deeply embedded across most organisations, with the average number of AI technologies used per adopting business rising only modestly, from around 1.4 to around 1.6 since late 2023.

Just 10% of businesses using AI said they use it extensively, while only 15% reported that more than half of their workforce uses AI as part of their daily work.

The information and communication sector led adoption at 58%, while construction lagged behind at 13%. Large language models were the most widely used AI technology, ahead of visual content creation and machine learning for data processing.

The ONS also found little evidence that AI has yet led to widespread job losses, contrary to public perception. Around half of businesses said AI had no impact on workforce headcount, while only around 6% of companies using AI to improve business operations reported reducing staff numbers.

"A notable proportion of businesses report that the impact of AI on workforce headcount is either uncertain or not applicable, suggesting that for many firms the effects are either still emerging or not yet relevant to their operations," the report said.

The ONS added that it is working with other national statistical organisations, the AI Economics Institute, and other government departments to inform how it might reflect AI in macroeconomic statistics. The aim is to have identified the priority measurement topics for AI, including a clear recommendation on the feasibility and value of an AI-thematic account, by March 2027.

---