Cleo McCormack

by Cleo McCormack

Co-Creation Director, Version1

This is the second in a series on AI-led change. The first post, ‘Navigating AI-led disruption: why people, not platforms, determine success’, made the case that genuine AI transformation depends on people, culture and leadership.  

This post examines what most organisations miss when they encounter resistance to AI and why learning to read it differently is one of the most powerful things a change programme can do. 

When people push back on an AI programme, fall silent in meetings, agree in the room and do nothing afterwards or quietly route around the new tool, the instinct is to treat it as a problem to overcome. A better communications plan, more training or clearer sponsorship from above. 

What if it is something more useful than that? What if resistance, read carefully, tells you almost everything you need to know about what the programme has not yet got right? 

In our experience working across AI adoption programmes, resistance is rarely irrational and almost never random. It tends to carry one of four types of insight — each one a signal worth understanding before you decide how to respond.

Four things resistance is usually telling you

The first is implementation insight: the practitioner is telling you, from direct experience, that the approach will not work. “We tried this before” or “our systems are not set up for this” sounds like obstruction. It is usually an accurate diagnosis. The question that surfaces it is simple: walk me through what you see breaking down.

The second is contextual insight: local knowledge about how things actually work that never made it into the programme plan. The culture, the informal systems, the invisible architecture of decision-making in a particular team. Programmes planned from the centre miss this almost by design. The people closest to the work rarely do.

The third is ethical insight: concerns about fairness, unintended consequences, about who benefits and who does not. “This doesn’t feel right” or “what about the impact on…” are not emotional reactions to be managed past. They are frequently the most important signals in the room, and the ones most likely to predict downstream problems if left unaddressed.

The fourth is identity insight: the values and sense of purpose at stake when a change asks people to do their job differently. An experienced developer, analyst, or clinician who says “I didn’t spend years learning this to hand it over to a tool” is not being obstructive. They are telling you what the change is costing them before the programme has offered them anything in return. That is information worth sitting with.

None of the above indicates a workforce that needs managing. They signal a programme that needs to listen differently. And when you do listen, resistance stops being an obstacle and starts becoming the criteria for what good adoption actually looks like. 

Why most programmes are not set up to hear it 

The readiness data helps explain why the gap is so persistent.  

  • Microsoft’s 2026 Work Trend Index found only 19 per cent of knowledge workers in what it calls the Frontier zone, where individual AI capability and organisational readiness are both high. 
  • Forrester’s AI Quotient  research put the figure at 16 per cent in 2025  
  • James Prochaska’s (1) research on behavioural change — developed over decades — helps explain why: at any given moment, roughly one in five people are genuinely ready to act on a change. The rest are somewhere between not yet understanding why it matters and understanding but not yet feeling safe enough to move 

This is not a failure of the workforce. It is how people respond to change that touches on identity, standing, and what it means to be good at a job someone has spent years building a skillset in.  

Research by Coombs (3) and colleagues in Electronic Markets identifies threatened identity, disrupted standing, and lost expertise as the three primary predictors of AI resistance. The most experienced people in an organisation are most exposed. That is not stubbornness. It is a rational response to a real threat the programme has not yet addressed. 

A 2025 paper by Gjerald (2) and colleagues in the Journal of Change Management names the broader problem as dead ideas — assumptions about how change works that persist even when the evidence has moved on. Top-down, linear, leader-driven change worked when disruption arrived in manageable waves. Applied to AI, where the change asks people to rethink not just their tools but their professional identity, it can accelerate exactly the resistance it was designed to prevent. 

Where listening must happen 

Karl Weick’s (4) work on sensemaking describes how shared meaning at work is built in conversation. Largely in the informal exchanges between formal communications, in what managers do and say day to day, in the small moments where people work out together what a change actually means for them. Meaning is not handed down from above. It is constructed, and the manager in the middle is where most of that construction happens. 

When organisations invest in senior sponsorship but leave the middle layer without the tools to surface and engage with resistance (and to feed it back into programme design) they create a gap between what leadership intends and what people experience. That gap, more than any technology shortfall, is what stalls AI programmes. It is also the gap that communications plans and training rollouts are structurally unable to close on their own. 

The organisations making genuine progress have accepted this. Accenture’s 2024 research across 2,000 executives and 15 industries found that those with the highest AI operational maturity achieve 2.5  times higher revenue growth than those with the lowest readiness. What separates them is not the technology stack, it’s whether the human system was treated as a design priority from the start.

group of professionals working in a bright office around computers

A different question to start with 

REACH, our change framework, was built from this evidence and tested in delivery across real programmes with real adoption challenges. It operates simultaneously at the individual and organisational level, treating managers as the primary sense makers and resistance as intelligence rather than obstruction. It gives teams a structured way to surface the four types of insight above so that the programme learns from what people are experiencing, rather than working against it. 

None of the above indicates a workforce that needs managing. They signal a programme that needs to listen differently. When you can identify which of the four is activated, you stop guessing. You know where the blockage sits, what it needs, and what kind of response will actually move things forward.

Talk to us about building change management that learns as it goes. 

References 

[1] Prochaska, J.O., DiClemente, C.C., & Norcross, J.C. (1992). In search of how people change. American Psychologist, 47(9), 1102–1114. 

[2] Gjerald, O., et al. (2025). Changing change: from heroic leadership to collective agency. Journal of Change Management. 

[3] Coombs, C., et al. (2020). Artificial intelligence and everyday work: Understanding employee responses. Electronic Markets. 

[4] Weick, K.E. (1995). Sensemaking in Organizations. Thousand Oaks: Sage.