UK Official Data: AI Adoption Tripled to 35%, But Depth Barely Moved

The share of UK businesses using AI nearly tripled in under three years — from around 12% to around 35%. Over the same period, the number of AI tools the average adopter actually runs crept from 1.4 to 1.6. That gap, between how many companies bought in and how deeply they use what they bought, is the most useful thing in the UK's newest official AI statistics — and it should change what you expect for your own job.

[Fact] The Office for National Statistics published "Artificial intelligence in UK businesses: 2023 to 2026" on 20 July 2026, drawing on the Business Insights and Conditions Survey (BICS) Wave 159, fielded from 5 to 28 June 2026 with a 26.7% response rate. BICS matters because it is run by a national statistical office rather than a software vendor or a consultancy chasing a headline. Its numbers tend to land lower, and more soberly, than the industry surveys that circulate on social media.

Adoption tripled. Usage did not.

[Fact] Among UK businesses with 10 or more employees, AI use rose from around 12% in September 2023 to around 35% in June 2026. [Fact] Businesses with 0 to 9 employees sit at 28%, and those with 250 or more employees at 49%.

Then the depth measures arrive, and the story turns.

[Fact] The average number of distinct AI technologies used per adopting business moved only from about 1.4 to 1.6 across the whole period. [Fact] Only 10% of AI-adopting businesses describe their own use as extensive. [Fact] Just 15% report that more than half of their employees use AI as part of daily work.

Read those three together and the typical "AI-adopting" UK business looks like this: one or two tools, used by a minority of staff, most of the time.

The tools, in the order businesses actually use them

[Fact] As of June 2026, among businesses with 10 or more employees: large language models 18%, visual content creation 16%, data processing using machine learning 12%, image processing using machine learning 6%, robotics 2%, other AI technologies 2%.

Robotics at 2% deserves a pause. The version of automation most people picture — machines on a floor replacing physical work — is the smallest category in the UK data by a wide margin. What businesses are actually buying is text and images.

Which jobs UK businesses name

This is where BICS goes further than most adoption surveys. ONS asked adopters which roles their AI use has affected, and the answers cluster tightly.

[Fact] Among businesses using visual content creation AI, over half report impacts on creative or design roles. [Fact] Among businesses using machine learning image processing, 53% report impacts on administrative or clerical roles and on creative or design roles. [Fact] Among businesses using machine learning for data processing, the most commonly reported impacts are administrative or clerical roles at 41% and data analysis roles at 39%.

Two role families absorb almost all of the reported effect: creative and design work, and administrative and clerical work. If you are a graphic designer, an art director, a photographer, an administrative assistant, or a data entry keyer, your occupation is being named directly by employers in an official national survey. That is a stronger signal than a model-based exposure score, because it comes from the people who sign the contracts.

[Fact] Data analysis roles at 39% put data scientists and market research analysts in the same conversation — though "affected" here plainly means tooling change more often than headcount change.

The headcount question, answered less dramatically than expected

[Fact] Around half of businesses report that AI has produced no change to their overall workforce headcount so far. [Fact] Around 7% of medium-sized businesses report a decrease. [Fact] Around 6% of businesses using AI specifically for operational improvements report a decrease in headcount, a higher share than among businesses using AI for other purposes.

Here is the obvious objection, and it is a fair one: headcount is a lagging and blunt instrument. A firm that quietly stops replacing people who leave will report "no change" for a year or more before the level moves. [Claim] The reported 7% is therefore better read as a floor on AI-attributed workforce reduction than as a full accounting of it. The role-level answers — over half naming creative or design impacts — are moving faster than the headcount answers, which is exactly the pattern you would expect if the first effect is fewer new openings rather than layoffs.

Where adoption actually concentrates

[Fact] Information and communication businesses report 58% adoption. [Fact] Construction reports 13%.

A four-and-a-half-fold spread between sectors, inside one country, in the same month. National averages are close to useless for personal planning; the sector you work in explains far more than the year does.

The training gap is the real story for workers

[Fact] Around 40% of medium- to large-sized businesses report integrating AI skills through training. [Fact] But only 11% of businesses report that more than half of their workforce has received AI-related training. [Fact] And 41% of businesses with 10 or more employees report no barriers at all to adoption.

Employers report almost no friction adopting AI, and almost no follow-through on training the people who have to use it. [Estimate] For most UK workers in an "AI-adopting" business, AI skills will remain something acquired on personal time rather than company time.

What this survey cannot tell you

Honest limits, because they matter for how much weight you put on the numbers. BICS asks businesses, not workers — an employer's view of "which roles were affected" is not the same as a worker's. ONS labels this analysis official statistics in development, and flags measurement uncertainty about how much AI activity falls inside the production boundary. "Impact on a role" is not a synonym for job loss; it covers task change, tooling change, and reallocation. And this is UK data, on UK firm structure, in a labour market with its own hiring rules.

What to do with this

If your occupation appears in the role list above, the practical move is not to panic about the 7%. It is to notice that adoption is wide and shallow, which means the near-term advantage sits with the worker who becomes the person in the team who actually operates the tools well — the 15% rather than the other 85%. Ask your employer directly whether you fall inside the 11% who get more than half-workforce training, or outside it. If outside, treat that as the answer about who is expected to close the skills gap.

Check your own occupation page for exposure and automation-risk figures alongside these UK employer reports, and watch the sector number more closely than the national one.

Sources

• Office for National Statistics, "Artificial intelligence in UK businesses: 2023 to 2026" (20 July 2026) — https://www.ons.gov.uk/businessindustryandtrade/business/businessservices/articles/artificialintelligenceinukbusinesses/2023to2026
• Office for National Statistics, "Management practices and the adoption of technology and artificial intelligence in UK firms" (24 March 2025) — https://www.ons.gov.uk/economy/economicoutputandproductivity/productivitymeasures/articles/managementpracticesandtheadoptionoftechnologyandartificialintelligenceinukfirms2023/2025-03-24
• Office for National Statistics, "Research into how artificial intelligence (AI) is affecting employment" (FOI response, 18 November 2025) — https://www.ons.gov.uk/aboutus/transparencyandgovernance/freedomofinformationfoi/researchintohowartificialintelligenceaiisaffectingemployment

---

AI-assisted analysis. Figures in this article are taken directly from the Office for National Statistics release cited above; the interpretation, occupational mapping, and caveats are our own. Percentages are rounded as published by ONS.