How to respond to AI skeptics without polarizing your team
Respond to AI skeptics by acknowledging what is true in their concerns, sorting everything they raise into: practical, emotional, and distorted, and turning each real concern into a requirement for your rollout. Then give them a safe, documented way to try AI on their own timeline instead of yours.
I'm writing this for CEOs, CMOs and senior leaders at mid-market companies who are rolling out AI and hearing loud pushback from their own people. I've been on the receiving end of that pushback in speaking engagements and training rooms, and my advice is: stop treating skepticism as resistance to beat. Treat it as information you haven't mined yet.
The skeptics got louder
When I first started speaking and training on AI, the skeptics in my sessions were quiet. By the end of 2026 that has changed. They speak up, they speak with confidence, and they say plainly what they are worried about.
The themes repeat from room to room:
"We're going to get stupid."
Critical thinking will be lost.
The joy of doing things the analog way is going away.
Handwriting will disappear.
What may make you uncomfortable is the fact that some of this is simply true. I have a 15-year-old who doesn't use AI, and I can still see how differently the things we all spend our time on have changed. I can't wave away a worry I recognize at home.
So my goal in these rooms is not to convince everyone that AI is the answer to everyone's problems. It's not, and a skeptic who senses a sales pitch will dig in harder. My goal is a rollout that holds up under honest scrutiny.
Why speed of change makes this harder now
On an episode of my podcast, AI Literacy for Entrepreneurs, I sat down with Ian Alden Russell and we talked about a better approach. Ian is a strategist who started out as an archaeologist and anthropologist, and he facilitates a lot of rooms where people disagree about change. His rule is that everyone in the room is right about something.
Technology has always changed work. The speed is what's new. More generations now work side by side with very different experiences of technology: people who learned their craft before the internet, people who grew up on smartphones, and people entering the workforce with AI already in the toolbox. Each group carries a different picture of what "good work" looks like and what might be lost.
When change was slower, those pictures had time to blend. Now they collide inside one team meeting, and the collision sounds like a debate about AI while really being about what people value and fear losing. Leaders who hear only the technology debate answer the wrong question.
Step 1: Don't sort the room into right and wrong
The fastest way to polarize a team is to split it into believers and blockers. Once people have been labelled, they defend the label.
Start from a different assumption: everyone in the room is right about something they are seeing, even if they can't yet put it into words. The enthusiast sees real time savings. The skeptic sees real risks to skill and judgment. Both observations are accurate, and your rollout needs both. (If you're worried your own people will feel replaced, this piece on how to stop AI from replacing your team covers the growth side of that conversation.)
Note: I built the 4 AI Cliff Archetypes Quiz to help people understand how you are geared to meet this AI moment. Take the quiz here: https://www.tryinteract.com/share/quiz/69cd787bde5b5fd533f2a4d6
Step 2: Acknowledge what is true, out loud
Say it in the meeting: "It is possible this goes badly."
Leaders resist this because it feels like weakening the case. It does the opposite. A skeptic who hears you name the risk they were about to raise stops preparing the rebuttal and starts listening. You've also told the quiet people in the room that honesty is safe here.
Concede specifically. "Some skills may atrophy if we stop practising them" works better than a general "we know there are risks".
Step 3: Put everything on the table and sort it into three buckets
Ask for every concern, in the open, and write each one down without arguing. Then sort them into three buckets:
Practical and factual concerns. Accuracy, data security, who is accountable when the output is wrong, whether the tool fits the workflow.
Emotional concerns. Fear of being replaced, grief over craft, the sense that the part of the job they loved is being taken away.
Distortions. Beliefs that don't match the facts, such as "the tool is secretly scoring each of us". (This theme has come up in so much of our AI literacy work)
Don't skip the third bucket. A distortion that people feel still changes their behaviour. If a team believes AI is quietly measuring them, they will hide how they work, and no amount of correct information fixes behaviour driven by a belief nobody has addressed. Correct the facts, and treat the fear underneath as real.
Step 4: Turn each real concern into a design requirement
This is where skepticism becomes useful. Every real concern becomes a requirement your rollout has to meet. Here are three I'd write down:
"Critical thinking will erode." Requirement: a mandatory human review step on AI output, plus training on how to review it. Reviewing AI output is a skill, and the schools example in this post on AI governance shows what happens when no one teaches it.
"My job is being studied so it can be replaced." Requirement: a plain written statement of what the initiative is for. Our research on employee fear of replacement shows how quickly people fill a silence with the worst reading. Don't leave the silence.
"Nobody knows who is responsible when it goes wrong." Requirement: named owners across leadership, HR and IT. The Holy Trinity of AI governance lays out how those three functions split the work.
Then write the commitments down where people can read them. A one-page AI policy is a good home for them. When a skeptic's concern shows up as a line in the policy, they stop being a skeptic in the room and become a contributor to it.
Step 5: Don't ask skeptics to adopt on your timeline
Adoption on a leadership deadline produces compliance, and compliance produces quiet workarounds. People choose to use new tools for their own reasons and on their own schedules, which is the pattern behind why AI adoption fails at the individual choice level.
Give skeptics a named, safe way to try. Name one task, one tool and one place where a mistake costs nothing. Say who they can ask for help. Say plainly that trying it is an invitation, and that you want to hear what they find, including that it was worse than doing it by hand.
Be wary of the pressure that comes from outside the building. If the rollout exists because competitors announced something, your skeptics will smell it, and FOMO-led adoption is hard to defend in a room full of people paid to notice weak reasoning. Once people start reporting back what works, feed it into a loop that compounds. That's the idea behind the AI Flywheel.
What to say and what not to say
Managers carry most of this conversation, so give them language.
Say:
"You may be right that this goes badly in places. Here is what we're doing to catch it."
"Tell me every worry you have. I'll write them all down and we'll sort them together."
"Here is what this initiative is for." Then state it in one or two plain sentences.
"Try it on this one task if you'd like. Tell me what you find, good or bad."
Don't say:
"Everyone needs to get on board." It sorts the room into believers and blockers.
"That's just fear of change." It dismisses the emotional bucket and teaches people to hide it.
"AI won't affect anyone's job." Don't promise what you can't guarantee. Say what you know and what you're still working out.
"You'll get used to it." It tells the skeptic their judgment doesn't count.
The point of all this
A team that argues openly about AI is healthier than one that nods and quietly routes around the rollout. The skeptics in your company are doing unpaid risk analysis. Take it, sort it, and build it into the work. I go deeper on how people and organizations move from AI-curious to AI-literate in my book, Swan Dive Backwards.
Hear a deeper conversation on this
I explored how to meet naysayers in Episode 280 of my podcast, AI Literacy for Entrepreneurs, with Ian Alden Russell. Ian is a strategist who works at the intersection of culture, commerce, technology and leadership, and he trained as an archaeologist and anthropologist.
In the episode, we talk about:
AI as a "crystal clear mirror" of who you are and who your organization is.
Why US, European and Chinese cultures may build different AI and regulatory frameworks, based on how each views the individual and the group.
How shared stories, like science fiction, give executives and developers a common frame of reference.
How to meet naysayers by finding what is true in their objection. The three-bucket sorting in this post draws on the facilitation approach Ian describes.
Listen to Episode 280 on Spotify.
Find out where your own rollout stands
If skeptics are slowing your rollout, the NorthLight AI Readiness Audit shows where the real concerns sit and what your rollout needs to answer them. Start the AI Readiness Audit.
