Your People Are Asking AI What They're Afraid to Ask You

A new survey from Clutch puts a number on something many leaders suspect but rarely see. In AI Coworkers Are Answering the Questions We're Too Embarrassed to Ask, the B2B research and reviews firm surveyed 669 full-time U.S. workers who use AI. Seventy-six percent said they had asked AI a work question they would have been embarrassed to ask a colleague. AI is now the first place 45% of them turn with a question, compared with 16% who go to a coworker first and 9% who go to their manager.

Clutch frames this as a story about collaboration changing shape, not disappearing. Nine in 10 respondents say they collaborate as much as or more than before. But there are warning signs. Fifty-seven percent say AI is creating silos on their teams, 51% report fewer quick check-ins from teammates and 33% have noticed less mentoring of newer employees.

The Embarrassment Was There Before the AI

Most coverage of findings like these will focus on the technology: AI is fast, always available and never judges you. All true. But the most important number in this survey isn't about AI at all. Three out of four people were embarrassed to ask a colleague a question about their own work.

That embarrassment didn't arrive with the chatbot. It was already there. People have long avoided asking what an acronym means, how a process works or what their manager actually wants, because not knowing felt risky. They worried about looking unqualified, slowing someone down or confirming a doubt about whether they belonged. So they guessed, stayed quiet or waited until someone else asked. AI simply gave them a private way around a culture that made not knowing feel unsafe.

When people edit what they don't know, they edit who they are. They show up as the polished version of themselves that has every answer, and they hide the questions that would help them grow. Leaders lose the very signals that tell them where someone is stuck, what training is missing and which process confuses everyone.

I wrote about why admitting what you don't know takes confidence that has to be earned in The Confidence to Say I Don't Know Has to Be Earned. This survey shows what happens when a culture doesn't make room for that kind of honesty: people still find their answers, just somewhere their leaders can't see. The questions don't go away. They go underground.

What Gets Lost When the Questions Go Private

There is real value in what Clutch found. Workers who can get quick answers on their own become more independent, and the survey suggests many teams are spending their time together on bigger conversations. Consistent internal tools, trained on a company's own documentation, can help everyone work from the same standards. That is a genuine asset.

But a question asked out loud does something a private answer can't. It tells a colleague what you are working on. It invites a story about how the team handled the same problem last year. It gives a newer employee a reason to sit with someone more experienced and learn how that person thinks. Those small exchanges are how people get known at work, and how they learn the judgment no manual contains.

When the questions go private, the learning still happens, but the relationships don't. And conviction is built in exactly those relationships, through the reps of asking, being wrong, getting feedback and trying again. I made a related point in When AI Does Your Thinking, You Lose the Reps That Build Conviction.

Questions for Leaders This Week

Ask the question yourself first. In your next team meeting, admit something you didn't understand and ask someone to explain it. People take their cues on what is safe from what the leader is willing to show.

Notice how you respond to basic questions. If you sigh, rush or answer with a link, your team will take its questions to AI and you will never hear them again.

Pair every newer employee with a colleague whose explicit job includes answering the questions they are embarrassed to ask. Make it part of how you measure that colleague's contribution.

Look at what people are asking AI in aggregate, if your tools allow it. The most common questions are a map of where your training, processes and communication are falling short.

And for professionals: before you take a question to AI, ask yourself whether it is one a colleague would be glad to answer. Some of the best working relationships begin with a simple question asked out loud.

AI will answer whatever your people are afraid to ask you. The leadership question is why they are afraid to ask you in the first place.

When people feel safe not knowing in front of each other, they grow faster than any tool can teach them.

Want to explore these ideas further? Learn more about my work on leadership, identity and conviction at www.theglennllopis.com.