Does employee resistance to AI explain low use?
Your rollout has stalled, and you're being asked whether employee resistance to AI adoption is the problem. Low use alone doesn't answer that question. Start by asking what people object to, what prevents an acceptable task, and what support they want.
Gallup's February 2026 survey of employed U.S. adults found that some non-users and infrequent users in organizations providing AI reported preferring their existing ways of working. These were reported reasons, not demonstrated causes of non-use.
We think the resistance label gets one thing right: willingness deserves attention. But it falls short when it turns every obstacle into an attitude problem; investigate the specific task before choosing a response.
Sources: AI in the Workplace: What Separates Adopters and Holdouts
How do I separate objections from obstacles?
Ask separately whether someone contests a proposed use and whether they can carry it out under acceptable conditions. An employee might oppose using AI for hiring decisions while also lacking access to an approved writing tool. Both explanations can coexist.
Use this suggested editorial sequence to investigate employee resistance to AI adoption. It's a conversation aid, not a validated diagnostic method; the questions below are suggestions.
- Check access: “Can you open the approved tool with your work account?”
- Clarify rules: “Which task or data are you unsure you're allowed to use?”
- Check time: “What work would need to move so you could practise?”
- Explore skills: “Would you like to try an approved task with someone supporting you?”
- Hear objections: “What specific risk or proposed use do you disagree with?”
When does concern need a different response?
Concern needs a conversation about its substance when the employee is challenging a risk, purpose or consequence. Don't automatically turn a job-impact objection into an invitation to another skills session.
AI Adoption Score's AI Anxiety Scale documentation describes expressed concern across learning comfort, job impact, AI safety and control, and human-like AI. Those categories don't establish refusal, diagnose individuals or demonstrate practical ability. The product's descriptive bands and index aren't validated severity measures.
In our view, disagreement deserves a specific answer before another demonstration. Ask what decision the employee wants explained or reconsidered, and be clear about what you can change.
Sources: AI Anxiety Scale (AIAS): how it works · AI Adoption Score
Who should own the practical response?
Assign an accountable owner to the reported obstacle, then confirm that person has authority to act. AI Adoption Score's AI Adoption Barriers documentation maps missing tools to IT, unclear rules to leadership, time constraints to managers and confidence support to training.
Treat that mapping as the product's suggested allocation, and adapt it locally. Its barrier statements aren't a validated scale, and rankings are relative even when reported barriers are low. We'd use the response log below to turn a survey answer into a checkable commitment.
| Reported obstacle | Follow-up check | Suggested owner | Proposed action | Review point |
|---|---|---|---|---|
| Missing access | Try work-account login | IT | Resolve access request | After access is granted |
| Unclear permission | Identify disputed task | Leadership | Clarify permitted use | After clarification |
| No practice time | Review workload | Manager | Agree protected time | After practice session |
| Low confidence | Offer supported task | Training | Arrange relevant support | After supported attempt |
Sources: AI Adoption Barriers: how it works · AI Adoption Score
How should I check what changed?
Review the practical commitment and employee responses separately. For employee resistance to AI adoption, our suggested review is to ask the same audience the same questions, record participation and changes in respondent composition, and compare reported obstacles separately from concerns.
Check whether the promised change actually happened: can the employee access the tool, identify the permitted task or use the agreed practice time? Describe any survey shift as a change following the action. Protect confidentiality by avoiding small-group breakdowns that could identify respondents.
Your next move is to replace a broad label with one specific issue, one owner and an agreed review date. Make the next action answer the actual problem.
Frequently asked questions
Who should coordinate this work across departments?
Choose one coordinator with access to the people who can approve changes. As editorial guidance, the Head of AI or transformation lead could maintain the response log, while People teams help shape sensitive conversations. Coordination should keep commitments visible without making one function responsible for every operational fix.
When should we send a follow-up survey?
Send it when employees have had a reasonable chance to encounter the promised change. As editorial guidance, agree that point before taking action rather than choosing an arbitrary recurring date. A permission clarification and a supported learning opportunity may need different review windows, depending on the work involved.
Should managers collect individual objections in writing?
Collect only the detail needed to understand and address the issue. As editorial guidance, explain who will read notes and how they'll be used, and offer a private route for sensitive concerns. Keep identifying details out of broadly shared action logs unless they're necessary and the employee understands the arrangement.
