There are two things happening in UK businesses right now, and most leaders are treating them as separate problems. They are not.
The first is a burnout and mental health crisis that has been building for the better part of a decade. The second is the fastest technology adoption most organisations have ever navigated — AI tools moving from novelty to expectation in a matter of months. Both are real. Both are demanding attention. And because they are happening simultaneously, they are colliding in ways that are producing risks most organisations are not yet naming, let alone managing.
This article is about where those two tracks meet, and what that collision means for founders and business leaders who are trying to get AI adoption right.
The two tracks
On the wellbeing side, the data is not ambiguous. UK sickness absence is at a fifteen-year high, averaging 9.4 days per employee (CIPD 2025). 41% of UK employers report that mental ill-health is now the leading cause of long-term absence in their organisation. The workforce arriving at AI adoption is not fresh and energised. It is, in many organisations, already at a tipping point.
On the AI side, adoption has accelerated faster than most forecasts predicted. 54% of UK firms are now actively using AI, up from 23% just three years ago (BCC 2026). 75% of knowledge workers are using AI tools at work. The technology is moving, with or without organisational guidance.
The pace is being set by the technology and by individual employees — not by organisations. That asymmetry is where the wellbeing risk lives.
Most organisations are managing these two things in parallel, in separate conversations. The wellbeing programme sits with HR. The AI adoption strategy sits with IT or operations. The two teams rarely talk. And in the gap between them, something is going wrong.
Four places where they collide
The collision between these two tracks is not abstract. It shows up in specific, concrete ways — and each one requires a different response.
The productivity paradox. The promise of AI is that it reduces workload. The reality, in the short term, is often the opposite. 62% of UK workers report that their workload has actually increased over the past year, despite AI adoption. The reason is straightforward: in a culture already driven by output, saved time does not become recovery time. It becomes more output. AI is being deployed into an exhausted workforce, and the efficiency gains are being absorbed by higher expectations rather than genuine relief.
62% of UK workers report their workload has increased over the past year, despite AI adoption. EY Future Workplace Index 2025
The dual training gap. Here is a number that should stop any founder in their tracks: only 26% of UK employees receive formal AI guidance or training from their organisation. At the same time, only 45% of managers have been trained to have mental health conversations with their teams. The line manager — the person who is expected to lead AI adoption from the front while supporting a team under increasing pressure — is equipped for neither side of this. They are caught in a vice, and most organisations have not noticed.
Psychological safety and shadow adoption. MHFA England's 2026 data shows that 44% of UK workers feel they cannot bring their whole self to work — a five-year low. When psychological safety is declining, employees who are confused or anxious about AI tools have no safe way to admit it. The result is shadow adoption: 71% of UK employees have used unapproved AI tools at work. This is not recklessness. It is operational anxiety with nowhere to go. People are working around the system because the system has not given them a legitimate path.
The reskilling illusion. The conventional organisational response to AI anxiety is a training programme. Learn the tools. Build the skills. Move forward. This is necessary but not sufficient. 43% of UK workers fear that over-reliance on AI will erode their core skills — and for your most expert employees, the anxiety is not really about skill at all. It is about identity. When your professional self-concept is built around expertise, and AI can now replicate that expertise, the question is not "how do I use this tool?" It is "who am I in a world where this tool exists?" Skills training does not answer that question. Most organisations are not asking it.
What this means in practice
The organisations getting AI adoption right are not, for the most part, the ones with the most sophisticated technology strategies. They are the ones with the most psychologically safe cultures — where people can admit confusion, experiment without fear of looking incompetent, and raise concerns without those concerns being interpreted as resistance.
That is not a soft observation. It has direct operational consequences for how you approach the next twelve months.
It means that your AI adoption programme needs a people strategy, not just a technology strategy. It means your line managers need support on both sides of the conversation they are being asked to have. It means that when you get resistance to AI tools, the right question is not "how do we communicate better?" but "what are people protecting, and is it reasonable that they feel that way?"
And it means that the gap between intending to adopt AI and actually adopting it is almost never a technology problem. It is a design problem — a failure to build the specific conditions in which new behaviour becomes the natural path rather than the difficult one.
The organisations navigating this well are not those with the best AI strategy. They are the ones with the most psychologically safe cultures.
Peter Gollwitzer's research on the intention-action gap is useful here. The bridge from intending to change to actually changing is not motivation or understanding — it is implementation design. The job of leaders right now is not to persuade their people that AI is good. It is to build the conditions in which engaging with AI becomes easier, safer, and more natural than avoiding it.
Five things founders can do now
Involve people before you implement. Co-design adoption where possible. People who help shape how a tool is introduced are far more likely to use it, and far less likely to feel that it is being done to them.
Name the anxiety explicitly and precisely. There are at least three distinct types of AI anxiety — existential (will this replace me?), operational (I don't know how to use it), and identity (who am I if this does what I do?). Each requires a different response. Treating them all the same makes all of them worse.
Reframe the narrative with specificity. Vague reassurance — "AI is a tool, not a replacement" — increases anxiety more than it reduces it. Specific, concrete answers about what AI actually frees people to do are far more effective than broad messages about the future being bright.
Build boundaries into policy, not culture. If your people are already at the edge of their capacity, the answer to AI-related overload is not a wellbeing programme. It is structural decisions about how AI efficiency gains are used, and who owns that decision.
Invest in identity, not just reskilling. Skills training closes one gap. Identity work — helping people understand who they are and what they offer in an AI-augmented role — closes the other. Most organisations are only doing the first. Both matter.