TL;DR
- The UK published six sector-specific AI adoption plans on 8 June 2026, each written by an independent AI Champion. Links are below.
- The same barriers show up across every sector: skills and leadership gaps, unclear use cases, patchy data, the jump from pilot to production, cost, and workforce anxiety.
- The clearest message is that integration beats adoption. Using AI is not the same as building it into how the work gets done.
- Two models are worth borrowing: the creative sector’s augmentation-first approach, and manufacturing’s Scan, Pilot, Scale pathway.
- The workforce moves that matter most are protecting entry-level roles and growing people who understand both a subject and the technology.
- Adoption already runs well ahead of the 25 per cent national average in most priority sectors, but using AI and integrating it are not the same thing.
In June 2026 the UK government published a set of sector-specific AI adoption plans, written by independent AI Champions across six industries.
If you missed them, I would not worry. They landed in a crowded news cycle, and six detailed reports can feel like homework rather than help.
I have read through them so you do not have to, at least not yet. Below you will find links to every plan, the barriers they have in common, and the five points I think matter most for smaller businesses working out where AI fits.
In this article
The reports, and who wrote them
You can read each plan in full on GOV.UK:
- Advanced manufacturing, by Chris Dungey
- Clean energy (interim), by Lucy Yu
- Creative industries, by Sally Davies
- Digital and technologies, by Katie Gallagher OBE
- Life sciences, by Dave Hallett
- Professional and business services, by Shaheen Sayed
If you want the short version first, there is a summary of all six plans, and the government has published its response. The Financial Services plan is due summer 2026.
The barriers every sector runs into
Read across all six plans and the same problems keep surfacing. If your own AI plans have stalled, it helps to know you are not alone, and that the reasons are quite understandable.
The two biggest challenges are strikingly consistent: 71 per cent of firms cannot identify a clear use case, and 60 per cent point to a shortage of skills and expertise. Regulatory uncertainty is close behind, raised in almost 60 per cent of responses to the government’s technology adoption review.
Other considerations include:
- Skills and leadership are the most cited constraint, named by around 60 per cent of businesses, and the gap is not just technical. Businesses lack managers who can spot a viable use case, judge an AI output and redesign a workflow around it. High-tech sectors also struggle to find people who understand both a subject and the technology, such as computational biologists in life sciences (see below).
- Governance, trust and regulation make firms cautious. Safety-critical settings in energy and manufacturing need strong evidence before anything scales. Most professions carries the added weight of client trust and data security. The creative industries are held back by unresolved questions on copyright and digital replicas, and professional services by how rules such as GDPR apply to generative tools.
- Data is often not ready. Useful information sits trapped in legacy machinery, fragmented grids or unpublished datasets that lack the labelling a model needs.
- The jump from pilot to production is where most projects stall. Standing up an internal chatbot is easy. Embedding AI across core workflows and teams, with metrics that prove it works, is not.
- Cost and infrastructure can shut smaller players out. Demand for high-performance computing outstrips supply, and subscriptions, integration, legal advice and security add up quickly for an SME.
- A lack of clear use cases is the single most common barrier of all, cited by 71 per cent of businesses (note: and I’ve heard this particular challenge from the AI get-go). Many managers cannot yet see where AI adds real value rather than shifting effort around.
- Culture and job security decide whether any of this takes hold. In professional services, where value rests on human judgement, more than half of employees worry about being replaced, and that anxiety quietly slows adoption.
Below are some more insights extracted from the specific AI Adoption Plans:
AI adoption barriers across six UK sectors
Drawn from the 2026 independent AI adoption plans by the UK AI Champions.
Five practical lessons
The barriers are familiar and perhaps you could argue that there is little need to repeat them here. With that in mind, the more useful part is what the Champions suggest doing about them. I select the five ideas and suggestions, they stood out to me as relevant well beyond the sectors that produced them.
1. Integration matters more than adoption
The Digital and Technologies plan makes a point worth pinning to the wall: using AI and integrating AI are not the same thing.
Plenty of firms can say they use AI, but the depth is often shallow. Handing your team a chatbot is a start. The productivity gains come from building AI workflows into how specific tasks actually gets done, and that is a leadership job as much as a technical one. It is the difference between a business that has AI and one that runs on it.
Additional Insights
Professional services show the gap in numbers. Around half the workforce there is already using AI and 69 per cent of firms have expanded training, yet 75 per cent are not ready on the basics such as data and orchestration, and 70 per cent report little progress on redesigning how the work flows. Training is running ahead of the systems it depends on.
2. The creative sector’s augmentation-first approach
The Creative Industries plan, written by Sally Davies, takes a calmer line than the replace-everyone headlines suggest. Its guiding idea is augmentation first. As the plan puts it, AI should support human creativity, not displace it.
One example cited is the team at Revolution Software who remastered its 1996 game Broken Sword: they used AI to upscale thousands of character sprites from 480p to 4k, then had human artists refine the frames the tool could not handle, such as heads and hands.
Aardman uses a directed tool called Copycat to clean up problematic frames in post-production. It now covers 70 to 80 per cent of that specific task, which frees skilled staff for higher-value work.
A few more stats
The creative sector is the most active adopter of all, with 51 per cent of businesses using AI against a 33 per cent average across the economy. The spread inside it is wide, from 60 per cent in software down to 22 per cent in music and performing arts. Trust is part of the brake: 10.1 per cent of creative firms cite doubts about AI product safety or transparency, nearly double the 5.7 per cent seen across all businesses.
3. A way out of the pilot trap
Manufacturing has the clearest framework for getting past the pilot that never scales. The Advanced Manufacturing plan calls it Scan, Pilot, Scale: help firms spot where AI could pay, give them safe places to test it, then prove it at full scale.
At the centre are what the plan calls lighthouse sites, a handful of real factories running proven AI in live production, so other firms can see what good looks like and copy it rather than start from a blank page. There is also a fast-track to help smaller firms follow the same route without the same cost.
Smaller firms rarely have the capital to fail their way through a dozen pilots, so a working example they can copy is worth more than any amount of theory.
More Evidence
Concrete and relatable examples can do the heavy lifting, in a randomised trial of 515 startups, those shown how other firms had reorganised around AI went on to find 44 per cent more uses for it, and the strongest among them earned close to twice the revenue of the group left to work it out alone. The gains landed with firms already able to act, which is the point of a proven site: it hands a ready business the map instead of asking it to draw one.
4. Protecting the first rung of the ladder
A concern that runs across the Digital and Professional Services plans is what happens to entry-level work.
The tasks we have always handed to juniors, from drafting to routine analysis, are exactly the ones AI does well. Automate them without thinking, and you remove the training ground for the experts you will need in ten years.
The Plan’s answer is an Early Careers Jobs Alliance, a commitment to redesign junior roles rather than quietly delete them. Auto Trader is a useful example: it has reshaped its graduate and apprentice training so trainees learn to use AI as a working tool, and its intake has grown rather than shrunk.
About entry level positions
The concern is not abstract. In 2024 the number of 16 to 24-year-olds in computer programming fell by 44 per cent in a single year. The exposure is real at task level too: the OECD finds around half of tasks in information services are open to AI today, rising towards 80 per cent as tools improve. The question is whether junior roles are redesigned around that shift or quietly removed by it.
5. The people who can speak both languages
The Life Sciences and Digital plans keep returning to a particular shortage: people who combine deep subject knowledge with enough AI understanding to put it to work.
A brilliant scientist who cannot brief a model, or a capable engineer with no grasp of the science, each leaves value on the table.
The plans point to a gap in the middle, the engineers and data specialists who build and maintain models rather than just designing them. Technical skill on its own is limited. It needs the judgement of someone who knows the field well enough to spot when an output is wrong.
Additional Insights
Life sciences alone points to a need for more than 100,000 new and replacement workers, which makes AI look more like a pressure valve than a threat. Professional services show the same shape: 52.8 per cent of roles are likely to be significantly augmented by AI against 13.7 per cent at risk of direct substitution.
Where this leaves you
None of this asks you to overhaul your business next week. It does reward some honest reflection, ideally out loud and with your team.
Here is a practical way to start.
1. Take the plan for your sector, block out an hour, and read it with the people who do the work.
2. You are not after a finished strategy on day one. You are after honest answers to a few questions:
- Where are we only using AI, when we could be building it into how we work?
- Which everyday tasks would we trust it with?
- And who here is quietly worried about what it means for their role?
That last question matters more than most owners expect. The plans are clear that adoption stalls when people feel done to rather than brought in, so the conversation itself is part of the work.
Make it a standing item, once a quarter, with the plan open on the screen. Do that, and you will be a long way ahead of the firms still waiting for a perfect moment that never comes.
These are practical documents. The businesses that read them as a team, and act on what they find, are the ones likely to pull ahead.
About the author
Pascal Fintoni is Managing Director of ArcLight Marketing & Media Ltd, with 30+ years of experience in digital marketing and business development. In February 2023 he introduced AI into all his training and consultancy services, and has since delivered executive briefings and AI masterclasses for small businesses across the UK, as well as hosting AI conferences. He is a podcast and video producer, and the author of WebProud: The 5-Step Roadmap to Feeling Proud of Your Website Again!.
