The system was built. It was tested. It was demonstrably faster and more accurate than the manual process it replaced. It was announced to the team with a slide deck and a launch email.
Six months later, three people used it. Everyone else had found reasons to keep doing things the old way.
This is not a technology story. It is the most common AI implementation failure mode – and the one that gets the least attention in technical planning. According to Writer’s 2026 Enterprise AI Adoption report, 79% of organisations face significant challenges in AI adoption – a double-digit increase from the year before. ManpowerGroup’s research found that while AI usage across workforces jumped 13% in 2025, confidence in using AI tools dropped 18% over the same period. The gap between deployment and adoption is widening.
Why People Do Not Use AI Tools They Have Been Given
The assumption most organisations make is that resistance to AI is about fear of job loss. That is part of it – 75% of employees worry AI could eliminate jobs, according to workplace research. But in practice, the deeper barrier to day-to-day adoption is usually confidence, not fear. Most employees do not feel equipped to judge when AI output is reliable and when it is not. They cannot tell a good AI result from a plausible-but-wrong one. So they default to what they know – the manual process – because that is the method they can trust.
Only 13% of workers have received any AI training, according to ManpowerGroup’s 2025 workforce data. You cannot build confidence without capability. And you cannot build capability without training that goes beyond ‘here is the login.’
The Manager Problem Nobody Talks About
The most common silent killer of AI adoption is middle management. Not because managers are opposed to AI in principle – but because AI often surfaces information that disrupts the way a manager has always run their function. An AI that shows which sales activities actually correlate with revenue is a threat to a sales manager who has built their playbook on a different instinct. An AI that identifies process bottlenecks makes visible inefficiencies the manager has explained away for years.
When managers feel threatened by AI outputs, they do not block the rollout. They simply do not champion it. And without management advocacy, adoption fails quietly – not dramatically.
The Three Interventions That Actually Move Adoption
1. Train for judgement, not just usage
Effective AI training does not just show employees how to use a tool. It teaches them how to evaluate the output – when to trust it, when to question it, and what to do when they are not sure. This is the capability gap that low adoption rates actually reflect. Workers who can make these judgements use AI tools; workers who cannot, avoid them.
2. Make the benefit visible at the individual level
Organisation-wide AI benefits – efficiency gains, cost savings, competitive advantage – do not motivate individual behaviour change. What motivates individuals is a clear, specific answer to the question: what does this make easier for me, in my actual job, today? The most successful AI rollouts identify two or three specific tasks the AI removes or simplifies for each role, demonstrate that value in training, and measure it publicly. Individual benefit, not aggregate impact, drives adoption.
3. Address manager concerns directly and early
Before rollout, work with managers one-on-one to understand what the AI reveals about their function, what concerns that raises, and how the organisation will use that information. Managers who understand that AI output will be used to improve their function – not to expose or replace them – are far more likely to champion adoption. Managers who are left to assume the worst will not.
Where We Come In
At DoSystems, every AI implementation engagement includes a change management workstream – because a system nobody uses is not an asset, it is a liability. We work with the people side of deployment in parallel with the technical build, not as an afterthought once the system is ready to launch. DoSystemsInc.com
Frequently Asked Questions
Why do employees resist AI tools?
The primary barrier is usually confidence, not fear. Most employees have not been trained to evaluate AI output reliability, so they default to familiar manual processes. Only 13% of workers have received any AI training (ManpowerGroup 2025), leaving a significant capability gap that training needs to close.
What is the most common AI change management mistake?
Treating change management as a communication exercise – announcing the AI, sending a launch email, and assuming adoption follows. Effective AI change management is a behaviour change programme that addresses capability, individual benefit, and management advocacy specifically.
How do you get managers to support AI adoption?
Address manager concerns directly before rollout. Managers need to understand how AI output will be used – specifically, that it is a tool for improving their function, not exposing or replacing them. Managers who understand this become adoption champions; those left to assume the worst become passive blockers.
How long does AI adoption take in an organisation?
Meaningful adoption – where the majority of intended users are reliably using the AI in their workflow – typically takes 3–6 months after launch for a well-managed rollout. Organisations without structured change management programmes often see adoption plateau at 20–30% of intended users.
#AIChangeManagement #AIAdoption #AIStrategy #DoSystems #DigitalTransformation #PeopleAI #AIImplementation
Comments are closed