A fast rise, measured carefully
The Office for National Statistics reports that self-declared AI use among UK businesses with at least ten employees rose from roughly 12% in late 2023 to around 35% by 2026. A separate government study produced a lower headline estimate because it used a broader sample and a different definition. The figures are not contradictory so much as a warning: an adoption rate is meaningful only when the survey population, question and technology definition travel with it.
Use is not transformation
Many businesses start with writing, summarisation or customer-service support. Those uses can save time, but they do not automatically change how a company prices products, manages inventory or allocates capital. A firm may count as an adopter while only a small group uses one tool occasionally. Deeper adoption means redesigning a process, measuring the result and creating secure access to reliable data. The hardest cost is often organisational rather than technical.
The metric that matters next
Britain should continue to track adoption, but it also needs measures of depth: how many functions changed, what proportion of staff use approved tools, the error rate and the verified time saved. Small firms need practical vendor comparison and help calculating a return, not general encouragement to buy AI. A large share of British business has moved beyond curiosity. The productivity story now depends on whether experiments become dependable operating systems.
UK TECH TRENDIndependent analysis for the British technology market.
Continue to all articles
