CONTEXT
A new survey from Business.com, conducted with 1,009 workers at U.S. small businesses, produced a number worth sitting with.
Thirty percent of employees report acting more enthusiastic about AI in front of colleagues than they actually feel. Nearly half say adopting too much AI could harm their company’s reputation. And while 57 percent of small businesses are now investing in AI, most workers still prefer operations to stay mostly human-led.
These are not employees resisting technology. These are employees inside businesses already using it. They have the tools. They show up to the meetings. They say the right things.
They just do not believe it.
STRUCTURAL ANALYSIS
This failure does not show up in your usage data.
When an owner signals that AI is the direction, the pressure to agree is real. Employees who are skeptical learn quickly that expressing doubt carries social risk. So they comply outwardly. The tools get used. The workflow changes nominally. The owner reads the surface and concludes adoption is working.
Three things are happening underneath. The feedback loop closes. Employees stop reporting what is not working. Output quality drifts because no one is genuinely invested in improving the process. And over time, trust erodes in both directions. Employees resent performing enthusiasm. Owners feel misled when they eventually discover the gap.
That is a harder problem to repair than the original adoption challenge.
THE ORDER CHECK
Before concluding that AI adoption is working in your business, four questions need honest answers. Not in a group setting. One on one, without the team present.
1. Can you name one person who has pushed back on how AI is being used, and what did you do with that feedback?
If you cannot name someone, the feedback loop is probably not functioning. That is not a team problem. It is a culture problem.
2. Do your employees understand why you are adopting AI, not just what you are asking them to use?
Compliance without understanding produces surface behavior. People do what they are told and nothing more.
3. Have you created a legitimate path for someone to say “this is not working”?
If the only options are agreement or silence, you have already selected for agreement.
4. Can you point to a specific instance where employee skepticism improved an AI-assisted output?
If not, skepticism is being suppressed rather than used. That is a real operational loss.
If you cannot answer these with specific examples, your adoption numbers are measuring compliance, not function.
THE DECISION
Pause on expanding AI use until you know what your team actually thinks.
Not because the tools are wrong. Because a business that moves forward on performed agreement does not know its real condition.
Have the direct conversations first. Ask these questions individually. Listen for hesitation, not just the answer. What you learn will either confirm that your adoption is real, or show you that it needs a different foundation.
Either outcome is more useful than the number you currently have.
Decide well,
– Chuck