When an AI rollout stalls, the instinct in most boardrooms is to blame the technology: the model isn’t good enough yet, the data isn’t clean enough, the integration wasn’t finished properly. The evidence tells a different story. According to SHRM’s research into AI in the workplace, 63% of employees rate their own AI proficiency as no better than beginner or intermediate, and a further 22% say they have no real experience with AI at all — leaving only around 15% who consider themselves advanced or expert users. That is not a technology gap. It is a confidence and capability gap, sitting inside people, not inside the software.
Microsoft’s 2026 Work Trend Index makes the point even more sharply. Comparing what actually drives AI impact inside organisations, it finds that organisational factors — culture, manager support and how work is designed — account for more than twice the impact of individual mindset and skill (roughly 67% versus 32%). Put together, the two findings say the same thing from opposite directions: people are not confident using AI yet, and the biggest reason is not their aptitude — it is the culture around them.
Why “more training” alone doesn’t fix adoption
Faced with low proficiency numbers, the reflex response is to buy more training. Training matters — but on its own it treats a trust problem as a skills problem. AI touches the things people defend most fiercely: job security, professional identity, and the sense of being competent at their work. Gartner’s research into 2026 change trends found 78% of chief human resources officers now agree that roles and workflows have to change for AI to deliver value — and role change is precisely the territory where fear, not ignorance, does the most damage. An employee can complete every module in a training curriculum and still quietly avoid using a tool they don’t trust, don’t feel safe experimenting with, or believe is designed to replace them.
What actually moves the needle: leadership behaviour
If culture is the real constraint, then leadership behaviour is the real lever — and here the data is unusually specific. Microsoft’s research found that when managers visibly model AI use themselves, their teams report a 17-point improvement in how much they value AI and a 30-point increase in confidence using it. When managers go further and actively create psychological safety for experimentation — making it normal to try, fumble and ask questions in the open — employees become 1.4 times more likely to become frequent AI users. None of that shows up on an IT dashboard. It shows up in how a manager talks about mistakes, how a team is allowed to learn in public, and whether people believe the organisation has their back through the transition.
Treating AI adoption as a culture problem, deliberately
This is exactly the discipline change implementation is built for. In practice, closing the culture gap takes four things working together:
- Build proficiency and confidence together. CCG Learn designs capability programmes that build real fluency and psychological safety, not just module completions — the difference between 63% of a workforce at “beginner” and a workforce that trusts the tools enough to use them daily.
- Redesign roles honestly. Through CCG Consult we help organisations have the direct conversation about how roles and workflows change with AI, rather than leaving people to fear the worst in silence.
- Equip leaders to model the behaviour. CCG Advisory works with executives and line managers on the visible sponsorship and experimentation culture that the data shows moves confidence and usage most.
- Measure trust and usage, not just licences. CCG Analytics tracks the adoption and sentiment signals that predict whether a rollout is actually landing — well before usage numbers reveal a problem.
Technology will keep improving on its own. Culture will not change on its own — it has to be led. The organisations that get real value from AI in 2026 will not be the ones that bought the best model. They will be the ones that treated adoption as the human, cultural undertaking it actually is.
If your AI rollout has the tools but not the trust, we would welcome the conversation. Get in touch with CCG.