Back to Resources
Research by the Federal Reserve Bank of St. Louis compared AI use among workers in the US and six European countries. In early 2026, around 36% of UK workers said they use AI for their job. Although this share was below the US figure of 43%, it was the highest among the six European countries that were surveyed.
But how does this really look inside UK businesses?
Well, there’s a new research from the Office for National Statistics (ONS) that gives us a closer look. The ONS research shows AI use is growing. But the findings also raise some questions about training, visibility and what successful AI adoption actually looks like.
In this blog, we’ll cover some of these findings:
🔹35% of UK businesses with 10 or more employees now use AI, up from 12% in 2023.
🔹58% of information and communication businesses use AI, the highest of any sector.
🔹55% of employees say they use AI for work or education.
🔹Close to 60% use AI to improve business operations.
🔹Only 11% of businesses with 10 or more employees have trained more than half their workforce on AI.
🔹15% of businesses reported that more than half their employees use AI as part of their daily work.
Let’s start with the positive findings first.
According to this research, around 35% of UK businesses with 10 or more employees now use some form of AI. Back in autumn 2023, that figure was about 12%. So, that’s almost three times higher in just under three years.
The information and communication sector, which includes IT, has one of the highest adoption rates, at 58%. It’s a broader category than ITSM, so this doesn’t exactly tell us how many service desks use AI.
The ONS doesn’t look at service desks specifically. But many of its findings land right on our doorstep.
For service desk leaders, the useful question is what this growing use means for the services we support, and whether AI is improving the delivered service.
Here are three findings that made us stop and think
#1 Your people may be one step ahead
The ONS shared these two interesting figures side by side. On one side, 55% of employed and self-employed people in the UK say they use AI for work or education, while only 35% of UK businesses with 10 or more employees report AI use.
Although they don’t exactly show how many employees are using unapproved tools, or how big the gap between official and informal workplace use is, it’s still an interesting comparison.
The ONS suggests that independent or informal use could help explain the difference. For a service desk, that’s a good reason to ask what people are already using rather than relying only on the approved software list.
If you’ve worked on a service desk, this might feel familiar. It’s not unusual that people find a tool that saves them time and start using it before approval is clear. That can resemble shadow IT, although these statistics don’t tell us how common it is.
The useful conversation here is about the task they’re trying to make easier. Are they summarising meetings, drafting customer replies or analysing a spreadsheet? Understanding the need makes it easier to offer a workable, approved option and spot where sensitive data could be exposed.
#2 Adoption is growing, but is more always better?
Among businesses with 10 or more employees that use AI, only 10% reported using it extensively. The average number of AI technologies used by adopting businesses grew modestly, from around 1.4 in late 2023 to 1.6 in June 2026.
We could say that this means many organisations have made a promising start, with plenty of room to explore what AI can do for them. But it could also mean businesses are being selective because only a few use cases deliver measurable value. Using more AI technologies doesn’t automatically mean an organisation is more mature.
The ONS itself cautions that counting technologies is an imperfect measure of how intensively AI is used. Sometimes using multiple tools may increase costs and complexity without improving it.
For the service desk, that means looking beyond whether a feature is switched on. If AI categorises tickets faster but sends more of them to the wrong team, the work hasn’t disappeared. Someone further down the line has to correct it.
#3 Training deserves a closer look
Based on the ONS research, only 11% of businesses say more than half of their workforce has received AI-related training. Meanwhile, 15% say more than half of their employees use AI in their daily work.
These statistics don’t tell us whether the daily users are the same people who received training. But they raise a question about whether support is keeping pace with use.
For service desk analysts, useful training goes beyond writing a better AI prompt. They need to know which tools are approved, what information can be shared, how to check an AI-generated answer and when to escalate a question.
Recurring tickets can help identify where guidance is missing. If people keep asking whether they can upload a customer file, the answer needs to come from an agreed policy and be easy to find. Analysts shouldn’t have to make that decision afresh each time.
If you’re wondering where to start, SDI’s Global Best Practice Standard v9 can help. Its AI concept covers practical areas such as training, clear responsibilities and checking that AI is working as intended. Use it to review the support your team needs, or try our AI maturity quiz to explore where to focus next.
The challenges worth exploring
None of this means we need to slow down. But it does help to take an honest look at where we are.
Skills are a common hurdle. Lack of expertise is one of the leading barriers to AI adoption. It’s reported by around 18% of businesses with 100 to 249 employees.
It’s worth reflecting on that second number for a moment.
It doesn’t mean businesses reporting no barriers are overlooking something. Could “no barriers” sometimes mean there’s simply more still to discover? Or they may already have the skills they need, and the figures don’t show whether training improved outcomes.
Concerns about jobs may be less pressing than expected. Around half of businesses say AI has had no effect on headcount. Among those using AI to improve operations, 63% report no change, and only 6% report a decrease.
For service desk leaders, it’s worth discussing how tasks might change before assuming time saved can become a staffing reduction. Checking AI outputs, maintaining knowledge and handling exceptions all take time too.
➡️ Download Free Guide: Eight AI Traps Every Service Desk Walks Into (And How to Avoid Them)
What this means for the service desk
When AI use grows faster than official adoption, it is natural to look for help. Without guidance, people often run into practical questions like the following ones:
- Am I allowed to use this?
- Is it OK to paste this customer data into an AI tool?
- Why can’t I install this AI add-in?
- The AI summary got this wrong. Who do I ask?
And those questions often end up with service desks. These aren’t just technical questions. They’re questions about trust, risk and confidence. And the service desk is often the first familiar place people go for answers.
There’s another angle too. Improving business operations is the most commonly reported purpose of AI use in the ONS findings. For IT teams, possible applications include ticket triage, knowledge support and self-service. These are service desk examples, rather than use cases measured by the research.
And much of that AI may not be built in-house. The ONS also found that information and communication businesses were more likely to access AI through purchased software or ready-to-use services. For a service desk, AI could arrive through existing ITSM tools they already use.
There’s a people angle too. As we saw earlier, AI is already affecting administrative and clerical roles. If an AI assistant draws on outdated articles, unclear instructions or content a user shouldn’t access, a convincing answer can still be wrong or inappropriate. Before expanding its use, check who owns the source material, how it is reviewed and which permissions apply.
We asked our LinkedIn community where they’d spend budget for one AI tool on the service desk. In a quick poll of 35 people, ticket triage and routing came out on top (40%), followed by keeping knowledge up to date (29%) and password and access requests (26%). It’s only a small snapshot, but it suggests service desk teams want AI to take on the repetitive work first, so analysts can focus on what matters most.
So, the service desk has two valuable roles to play. It can support colleagues who are using AI. And it can use AI to help deliver even better service itself. That’s an exciting opportunity, and one worth getting right.
Five ways to turn the findings into action
You don’t need a huge AI strategy to get started. Small, practical steps go a long way.
- Find out what people are already using. Rather than assuming, start by asking. A short survey or a few open conversations can tell you more than any policy document.
- Try to make the safe option the easy option. This means that if people are using AI informally, you can offer approved tools and clear guidance that will answer their everyday questions. You can also agree on who owns decisions about data, security and acceptable use, and give analysts clear escalation routes.
- Support your own team, then your colleagues. Your service desk analysts will often get the questions. Help them feel confident answering them. Provide them practical training on checking outputs, handling sensitive information and recognising when human judgement is needed.
- Start with one use case, then measure the overall result. Pick a real pain point and record a baseline before the pilot. For example, for ticket categorisation, track assignment accuracy, reassignments and resolution time alongside analyst time saved. For knowledge drafting, include review effort and corrections. Then you decide what result would justify expanding the pilot, and what would prompt a pause.
- Keep ownership and human help clear. Agree who is accountable when AI makes a recommendation or takes an action. Build in appropriate checks and give users an easy route to a person when the tool gets something wrong.
Final thoughts
AI adoption in the UK has almost tripled since 2023. That’s real progress, and the information and communication sector is leading the charge. So, we could say that the next chapter is less about how many businesses use AI, and more about how well they use it.
That means getting the foundations right: skills, trust, security, governance and good support.
And you don’t have to do this alone. Explore our AI webinars and resource library to learn from industry experts and peers.
⬇️ For more practical guidance, download Eight AI Traps Every Service Desk Walks Into (And How to Avoid Them).
Sources:
Office for National Statistics, Artificial intelligence in UK businesses: 2023 to 2026, released 20 July 2026. Based on the Business Insights and Conditions Survey (BICS). Figures refer to businesses with 10 or more employees unless stated otherwise.
Federal Reserve Bank of St. Louis, Mind the Gap: AI Adoption in Europe and the United States, published 30 March 2026. Worker survey carried out in January and February 2026.
