UK workplace AI use doubles, but the gains cluster at the top

TL;DR:

• Research commissioned by Google and run with Public First puts UK workplace AI adoption at 73%, up from 34% a year earlier.
• The study segments workers into four stages and finds the rewards — promotions, pay rises, faster progression — concentrate among the 15% it calls “trailblazers”.
• Google is tying the findings to “AI Works for Britain”, part of a stated partnership with the Government to train 10 million workers in AI skills by 2030.

Most of the UK workforce is now using AI at work, but the benefits are far from evenly shared. Research commissioned by Google and conducted with the consultancy Public First puts workplace adoption at 73%, more than double the 34% reported a year earlier — a figure worth weighing against the company’s interest in promoting AI use, but striking nonetheless.

A spectrum, not a switch

The study sorts workers into four groups: “spectators” (10%) not yet experimenting, “experimenters” (38%) testing simple tasks, “practitioners” (37%) using AI as a daily tool, and “trailblazers” (15%) finding new ways to work. According to the research, the trailblazers are 84% more likely to have been promoted in the past year and 55% more likely to have secured a pay rise, even after controlling for age, sector, gender and education. The gap, the authors argue, is behavioural and organisational rather than technical: many casual users never iterate prompts, and only a third say they have clear workplace guidance on responsible use.

That framing aligns neatly with Google’s commercial position — the report doubles as a case for its tools and a new Public First “AI skills quiz” — and its self-reported impact figures (£140bn of UK economic activity supported in 2025) should be read in that light. But the underlying tension it describes is real and visible elsewhere: UK firms remain ambivalent even as usage climbs, with most businesses using AI while calling it irrelevant in separate research this month.

For UK employers, the actionable point is the “permission to prompt” gap. If access to AI is widespread but guidance and encouragement are not, adoption stalls at shallow use — and the productivity dividend, along with the career rewards, accrues to a narrow group.

Looking forward

Google says the work feeds the Government’s goal of training 10 million workers in AI skills by 2030, building on its Digital Garage programme. Whether that target narrows the gap or entrenches it will depend less on access to tools than on whether employers give staff the time, guidance and remit to move beyond first-step experimentation.