Reading a Team's Pushback on AI as an Early Warning
Three kinds of pushback on AI, what each one tells a leader, and how to respond
It’s six weeks after the company AI rollout, and there’s a pattern setting in. The same handful of enthusiasts are using the new tools every day, but everyone else has slowly drifted back to their old ways of working.
Ask why, and you get a familiar answer: “I’m just too busy.”
That may well be the whole explanation. Learning takes time, and the existing workload certainly hasn’t gone away. But it might also be the easiest answer to get an AI-pushing manager off your back. For now.
“I’m worried about my job security” is harder to say. So is “I’m not sure where my expertise fits in anymore.” And anyone who says “the outputs aren’t good enough” risks being labeled a stick-in-the-mud, desperately clinging to the old ways of doing things.
These concerns deserve attention, however, and they need different answers. Yet organizations often respond in the same way each time: more training, another demonstration, a stronger mandate.
We see it differently. Resistance is information a leader needs for the change to succeed.
Before trying to overcome it, find out what it’s telling you. Treated that way, pushback becomes an early warning: a signal of where the change needs attention while there’s still time to act.
The Rollout Is Part of the Problem
How a change is introduced affects how people respond. Announce a tool without explaining its purpose, leave people guessing about their jobs, or expect them to learn on top of a full workload, and you help create the hesitation you later encounter.
Research on organizational change has long challenged the assumption that resistance is entirely the responsibility of the people being asked to change. Leaders’ own actions and omissions can help produce it, and the resulting pushback can be a resource for improving the change (Ford, Ford, & D’Amelio, 2008).
Particularly in creative teams, we see three broad forms of resistance: threat to self (“Will I be OK?”), identity (“Where does my value lie?”), and what we call intelligent resistance (“This isn’t good enough”).
From the outside, they can look alike: low usage, polite delay and the familiar “too busy.” One person can experience all three, so treat them as starting points for a conversation rather than labels (Piderit, 2000).
Threat to Self: “Will I be OK?”
The first form is often the easiest to recognize.
Will my role exist next year? Does my experience still count? Who decided which tool I now have to use, and how I should work? If something goes wrong, who carries the blame?
David Rock’s (2008) SCARF framework offers a useful lens. It identifies five domains of social experience that can trigger a threat or reward response: status, certainty, autonomy, relatedness and fairness. An AI rollout can raise questions about every one.
Consider the message to “do more with AI.” A leader may intend it as encouragement to experiment without having thought through what it means for the people doing the work. An employee may hear that their expertise is becoming less valuable, their workload is about to increase, and decisions about their future have already been made.
Another demonstration won’t necessarily address any of those concerns, and a stronger mandate may secure compliance but it isn’t going to boost enthusiasm.
The response needs to fit the worry. If people lack certainty, explain what has been decided, what remains open, and when they’ll hear more. If they feel the change has been imposed on them, give them meaningful influence over how it happens. If they question its fairness, explain how the benefits, extra work and accountability will be shared.
Use SCARF to check the rollout as a whole. Have you explained where experience still matters? Have people had a say? Will they have colleagues to learn alongside? Who gets the time saved?
Then, in individual conversations, listen for the concern that matters most. You don’t need to guess what everyone is feeling. You need to make it possible for them to tell you, and be prepared to respond.
Identity: “Where Does My Value Lie?”
Some concerns go beyond what might happen to a particular job or role. They go to the heart of what the work means to the person doing it.
An illustrator who has spent twenty years developing a craft may watch a model generate an image in seconds and wonder where that leaves them. A copywriter may understand the efficiency argument perfectly and still dread a future spent correcting machine-written prose.
For someone in that position, being told to “focus on the higher-value work” can feel dismissive. They may have valued the work they were already doing.
It helps to distinguish skills, roles and professional identity.
A tool may change how someone produces a first draft or explores a visual direction. Their role may shift toward directing, selecting and editing. But their understanding of an audience, their taste, and their ability to recognize what works may remain valuable.
Give two people the same camera and the same scene, and the trained photographer’s picture will usually be far better, because of the eye behind it.
That doesn’t make the transition effortless. For some people, making the work themselves is part of the satisfaction. There may be a genuine loss to acknowledge.
A useful conversation starts with curiosity: Which parts of the work make you proud? What do colleagues rely on you for? What would you miss if this task changed?
Those answers help identify what needs preserving and where a future contribution might lie.
But leaders have to give that expertise somewhere to go. If judgment matters, what decisions will the person own? If the role is changing, what time and support will they have to develop? How will their contribution be recognized?
Reassurance becomes credible when people can see their contribution in the work.
There’s a responsibility on the individual side too: to examine which capabilities transfer, where new skills are needed, and what they’re willing to try. That’s easier when the organization offers a believable route forward.
Be specific about what is changing. Is the person using a different tool, taking on different responsibilities, or being asked to rethink their profession?
Intelligent Resistance: “This Isn’t Good Enough”
The third form is particularly easy to mistake for an attitude problem.
Someone says the copy doesn’t sound like the brand. An image looks polished but misses the brief. A proposal appears complete, yet the person responsible for delivering it can see how much remains unresolved.
These objections may be telling you something important about the work.
These patterns predate AI. Early in his career, Matt was tasked with spearheading the shift to digital within a global advertising agency. Creative directors pushed back in numerous ways, leading to frustration and friction all round. Over time, however, Matt came to see that at least some of the pushback was about protecting creative standards, particularly in areas where digital solutions were not yet good enough.
That distinction matters whenever a new capability arrives with enough excitement to make scrutiny seem obstructive.
Consider a hypothetical campaign draft. The copy is fluent, the structure looks right, and the benefits are clearly stated. But an experienced writer notices that it promises something the product can’t deliver, or uses a tone the audience will distrust.
The useful question is: “Show me where it falls short.”
We’ve also noticed a subtler problem in our own AI work. Evaluating what a tool proposes can feel like being firmly in control. After a long day, though, it becomes easier to stop questioning and simply hit enter.
You’re still reviewing the work, but you’re no longer giving it the same attention.
Research on automation complacency and bias (Parasuraman & Manzey, 2010) describes how people can become less vigilant when working with automated systems, including when their attention is divided. Expertise alone doesn’t remove that risk.
A colleague who challenges the output may therefore be doing something valuable for the whole team.
That doesn’t make every objection correct. Ask the person to make the concern specific. Where does the output fail the brief? What consequences do they anticipate? What would an acceptable result look like?
Then examine the evidence together. You may need better instructions, a different review process, or a clearer standard. You may conclude that the tool doesn’t help with this task.
High usage tells you people are using the tools, but you still need to know whether the work is any better.
Three Moves For Leaders
Name the kind. Treat the framework as a working hypothesis. Is the person worried about what they stand to lose? Struggling to see where their expertise fits? Identifying a problem with the work? Expect overlap, and let the conversation change your initial interpretation.
Ask where it points. Make the concern concrete. “What worries you most about using this here?” “Where would you try it first?” “What would give you confidence in the result?” The answer should help you identify something to investigate or change.
Keep the useful friction. When an expert objects, ask them to explain what they see. Assess the concern against the task and the required standard. A review that catches an unsupported promise or exposes a weak brief is doing useful work, even if it slows the process down.
People also need to know that raising a concern won’t damage their standing. If every objection is treated as evidence of a poor attitude, leaders are likely to hear less about the problems that remain (Edmondson, 1999; Detert & Burris, 2007).
At a larger scale, combine ways of listening. A short survey can reveal patterns; interviews and team discussions can help explain them. Feed back what you’ve heard, what you’ll change, and what you won’t. Where a decision remains uncertain, say so.
Start With “Too Busy”
Go back to the colleague who says they’re too busy.
They may need time to learn. They may be worried about what happens once they do. Or they may have spotted a problem nobody else has noticed.
Ask them: “What would you need to feel confident using this for the work?”
Then pay attention to what happens next. If their answer exposes a problem with the rollout, change something. People learn whether speaking up is worthwhile from what leaders do after they’ve listened.
References
Detert, J. R., & Burris, E. R. (2007). Leadership behavior and employee voice: Is the door really open? Academy of Management Journal, 50(4), 869–884. https://doi.org/10.5465/amj.2007.26279183
Edmondson, A. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350–383. https://doi.org/10.2307/2666999
Ford, J. D., Ford, L. W., & D'Amelio, A. (2008). Resistance to change: The rest of the story. Academy of Management Review, 33(2), 362–377. https://doi.org/10.5465/amr.2008.31193235
Parasuraman, R., & Manzey, D. H. (2010). Complacency and bias in human use of automation: An attentional integration. Human Factors, 52(3), 381–410. https://doi.org/10.1177/0018720810376055
Piderit, S. K. (2000). Rethinking resistance and recognizing ambivalence: A multidimensional view of attitudes toward an organizational change. Academy of Management Review, 25(4), 783–794. https://doi.org/10.2307/259206
Rock, D. (2008). SCARF: A brain-based model for collaborating with and influencing others. NeuroLeadership Journal, 1, 1–9.
Rob Cannon, PhD is an organizational psychologist and Co-Founder and Chief Product Officer of Average Robot. He helps organizations navigate the people side of AI adoption, with a particular interest in professional identity, creative work and how teams collaborate. Rob has taught at the University of Sydney, the University of the Pacific and the University of California, Berkeley. He studied at the University of Cambridge and earned his PhD at the University of Sydney. He is co-author, with Matt Perry, of the forthcoming book The Elephant in the Algorithm.
Matt Perry is Co-Founder and CEO of Average Robot. He has spent thirty years in marketing and creative, and has led creative businesses through 25-plus years of disruption, growing Genero's North American business to over $50 million with clients like Google and Meta. In recent years, Matt has focused on AI strategy and building AI solutions. He is co-author, with Rob Cannon, of The Elephant in the Algorithm.
