AI adoption tripled. Depth didn't.

The Office for National Statistics (ONS) published new numbers on AI adoption in the UK. Business use has nearly tripled since late 2023, from 12% to 35%.

That sounds like transformation. It is not. The average number of AI technologies used per adopting business has moved from 1.4 to 1.6 in the same three years. Only 10% of businesses using AI call their own use extensive. We’ve gone wide, not deep.

The gap is not spread evenly either.

Source: ONS - Artificial intelligence in UK businesses: 2023 to 2026

If you run a services or knowledge business, you’re probably closer to the top of that table than the bottom. That doesn’t mean you’re ahead of it. Even information and communication, the clear leader, is still averaging under two AI technologies per business. The leaders are shallow too.

Larger businesses should have the best handle on this - more headcount, more IT, more reporting. They do not. The ONS found bigger businesses are more likely to say they’re simply not sure how much of their workforce uses AI day to day, not less. Size doesn’t buy you visibility. It buys you more places for the answer to hide.

Most of this is happening bottom-up, not top-down. 55% of employees report using AI at work. Only 35% of businesses report official use. The employees are ahead of the org chart.

And where a business is deciding what to point AI at, it’s mostly the same decision. Close to 60% use it to improve existing operations - cut a cost, speed something up, automate a task. Far fewer use it to build something new or go after a new market. Larger businesses lean into efficiency hardest; smaller ones are, relatively, a little more willing to point it at growth.

We’ve made this point before. A few issues ago, looking at PwC’s IKEA data: cutting the staff freed up by their chatbot would have banked about €13 million. The headlines would have been about the job losses. Instead they retrained the same people as design consultants and generated €1.3 billion in revenue.

The ONS numbers show why so few businesses have got anywhere near that decision yet. Adoption is still early, still shallow, still uneven by sector, and mostly being decided one employee at a time, not one leadership team at a time. Efficiency isn’t a choice most leaders are making. It’s the default you get when nobody’s gone deep enough to ask the other question.

Productivity is the consolation prize - we made a similar point a few issues back using PwC’s IKEA data. Worth a re-read alongside this one.

Ask each of your team leads this week: what AI tools is your team using that I don’t know about? You’ll likely be surprised - and you can’t manage what you can’t see.