Free AI courses answer the wrong question for small firms

On 28 January the Department for Science, Innovation and Technology (DSIT) opened its AI Skills Boost program to every adult in the UK. Technology secretary Liz Kendall set out the expansion in a speech at Bloomberg’s London headquarters, with a target of 10 million workers trained by 2030, at least 2 million of them employed by small and medium-sized enterprises (SMEs). The courses, checked against a Skills England benchmark, can take less than 20 minutes and teach people to use AI for drafting text, creating content and administrative tasks.

The partner list covers much of the software small firms already buy and most of the bodies that speak for them. Microsoft, Google, Sage, Intuit and Salesforce are among the founding partners; the Federation of Small Businesses (FSB), the British Chambers of Commerce, the Institute of Directors and Pax8 are among the new partners. Kendall’s summary was tidy: “We want AI to work for Britain, and that means ensuring Britons can work with AI.”

The same day, DSIT published research that complicates the premise. Its AI Adoption Research, carried out by IFF Research and Technopolis Group, surveyed 3,500 UK private sector businesses with at least five employees by phone between February and May 2025. Only 16% were using any AI technology. Asked what prevents or had prevented them from adopting AI, 71% said they had not identified a need for it and 60% cited limited skills. Among firms with no plans to adopt, the lack of a need rose to 81%.

That ordering matters. The government is paying to fix the second-biggest barrier, while the biggest is one a 20-minute course was never designed to touch. For the smallest firms, the missing piece is which job in the business AI should do – and a foundations badge does not certify that.

Skills are the second problem

The research shows a market split by size. Large (36%) and mid-sized (23%) businesses were more likely to be using AI, micro businesses (14%) less so. Just over half of all businesses (51%) did not see AI as relevant to their organization at all. The government’s own announcement, drawing on Office for National Statistics (ONS) data, puts micro businesses 45% less likely to adopt AI than large ones.

Skills are a real constraint, but they bite later. Among businesses already using AI, 54% said limited skills and expertise had held back wider use across the organization, rising to 64% at mid-sized firms. That is the audience a benchmarked course suits: firms that have chosen a use and now need more staff able to run it.

The FSB’s figure points the same way. “Small businesses want to make the most of AI, but just under half (46%) say they don’t yet have the skills or knowledge to use it well,” said Tina McKenzie, the FSB’s policy chair. Wanting to make the most of AI assumes a use for it. Most firms in DSIT’s survey said they had not found one.

Completions are the easy number

The program’s headline measure is courses. DSIT says 1 million have been delivered since June, a number it says is based on self-reported calculations from 11 industry partners. Completions are easy to count. They say little about whether a business now does anything differently on a Monday morning.

Spending tells a similar story. Just over a third (35%) of firms that use or plan to use AI did not know what they had spent on it in 2024. Among those that could put a figure on it, the median was £2,000. Those figures describe organizations trying things out. A plan waiting for trained staff would look different.

Nor is the training detached from the tools. Founding partners point to their own courses: Sage to its AI Fundamentals course, Salesforce to its Trailhead platform. Vendors have trained customers for decades and there is nothing improper in it, but it makes the product, rather than the process a small firm needs to fix, the likely starting point.

The process comes first

For owners and managing directors of small firms, the useful order is the reverse of the government’s. Pick one recurring task that costs real hours – quoting, chasing invoices, first-line customer email, tender responses – and decide what a good result looks like before anyone is trained. Then send the people who own that task on the course. A foundations badge earns its keep once it is attached to a job.

Mid-sized firms that already use AI are in a different position. DSIT’s data suggests skills are their binding constraint, and free benchmarked training is a cheap way to widen use beyond the early enthusiasts. They should take it, and judge it by whether the task changes, not by how many badges are earned.

The government wants Britain to be the fastest adopter of AI in the G7. Its own research says the bigger gap is purpose. Training can teach staff how to use AI. It cannot tell a business why.