Nearly half of U.S. employees in a new survey say they have exaggerated how much they use AI at work, or how good they are at it. Most leaders will read that as an honesty problem. I read it as a signal problem. Performative AI use is what happens when an organization asks people to show AI adoption before it has defined what good work with AI actually looks like.

We have seen this before. It used to be called looking busy.

Key takeaways

  • A Visier survey of 1,000 full-time U.S. workers found that 48% have exaggerated their AI usage or expertise to colleagues or leadership.
  • Nearly half of employees feel pressure to use AI, and most are unsure whether their employer has a clear AI plan.
  • Performative AI use is a new version of an old problem: when leaders reward visible signals, people learn to produce the signals.
  • The fix is clarity, not more monitoring: define where AI belongs in the work, what good work looks like, and how judgment will be evaluated.

What’s happening

On September 24, HR Dive reported on new Visier research into performative AI use at work. The analytics firm surveyed 1,000 full-time employees and conducted additional interviews. More than half said their roles had changed significantly because of AI in the past two years, and nearly 60% said they were unsure how their employer planned to address those changes. Nearly half reported feeling pressure to use AI, and a similar share admitted exaggerating their AI use and expertise. HR Dive also reported that 70% of respondents were concerned AI could hurt their careers or cost them their jobs.

Fast Company covered the same report on September 25 under the headline Workers are faking their interest in AI to satisfy bosses. It put the exaggeration figure at 48% and reported that 45% felt pressure to use AI without confidence that they could use it effectively. When asked what would help, employees pointed first to AI training and upskilling (32%) and greater transparency from leadership (31%). Fast Company noted that the report points to a lack of trust in leadership as an underlying driver. In its own research summary, Visier adds that only 28% of employees believe leaders fully understand how employees actually use AI.

A third piece, published by TechRadar on September 28, adds a revealing contrast. It reported on Adobe Acrobat research on AI use among UK adults at work: 99% of C-suite leaders said they use AI, compared with 41% of non-management employees. And 37% of C-suite leaders said AI saves them three to five hours a week, compared with 13% of knowledge workers.

Put those together and a pattern emerges. The people setting AI expectations are often having a very different AI experience than the people being asked to meet them. When the gap goes unnamed, employees fill it with performance.

Why performative AI use is a signal problem

In Chapter 7 of The Next Turn, “The Collapse of Effort,” I describe how effort used to work as a proof signal. Hours, drafts and visible struggle told a manager that someone was committed. AI broke the link between visible effort and visible value. A strong first draft can now appear in minutes.

Not the collapse of work. The collapse of effort as a signal.

The Next Turn

Here is what this week’s research adds. When an old signal collapses, organizations tend to reach for a new one. AI use itself is becoming that new signal. Are you using the tools? How often? Are you talking about it in meetings? Those are easy things to see. They are not the same as better work.

And when a signal is rewarded, people produce the signal. That was true of long hours, and it is true of AI. I make this point in Chapter 7 about leaders who respond to faster work with more check-ins and suspicion:

It teaches people that looking like you worked hard matters more than actually improving the work.

The Next Turn

Swap “worked hard” for “used AI” and you have a fair description of what Visier’s respondents are telling us. This is not a story about dishonest employees. It is a story about a system that made the appearance of adoption safer than an honest answer.

The book’s larger frame helps here. The structure of work, meaning how work actually flows and gets done, is changing quickly. More than half of Visier’s respondents say their roles have already shifted. But the definition of work, meaning the shared understanding of what good work is and what it asks of people, has not been rewritten to match. Employees are standing in the gap between the two.

Invisible work turns into suspicion

Chapter 9, “Invisible Work,” is about what happens when leaders increasingly see the product without seeing the path. AI makes the path harder to see. That becomes a trust issue, not just a workflow issue, and the research reflects it on both sides. Only 28% of employees in Visier’s research believe leaders fully understand how they actually use AI. Meanwhile, employees who are unsure what is expected of them are overstating their use.

If those definitions remain vague, invisible work does not stay neutral. It turns into suspicion.

The Next Turn

The wrong response is surveillance: tracking logins, prompt counts or tool usage and calling that adoption. That gives leaders more data about activity and no more clarity about judgment. It would likely make performative AI use worse, because it rewards exactly the signal people are already learning to fake.

The better response is to make thinking visible, even when the process is not. In my work with leadership teams, the most useful shift is often the simplest one: moving from asking whether someone used AI to asking them to walk you through how they decided.

Your next turn

If you lead a team that is being asked to adopt AI, try these moves this week:

  • Audit what you praise. In your last few team meetings, did you recognize AI use itself, or the quality of the outcome? People will follow the reward, not the memo.
  • Replace one usage question with a judgment question. Instead of “Are you using AI for this?” ask “What did AI do here, and what did you decide?”
  • Say where AI belongs. Name one or two tasks on your team where AI is clearly welcome, one where it is not, and what always needs a person’s review.
  • Close the experience gap. If your own AI use saves you hours, ask your team what their experience actually looks like. Don’t assume your workflow is theirs.
  • Make it safe to say “I’m still learning.” Pair any AI expectation with real time and support to learn. Pressure without training is how performance replaces practice.

Frequently asked questions

What is performative AI use at work?

Performative AI use is when employees overstate how much they use AI or how skilled they are with it, usually because they feel pressure to show adoption. In Visier’s 2026 survey of 1,000 U.S. full-time workers, 48% said they had exaggerated their AI usage or expertise to colleagues or leadership.

Why do employees exaggerate their AI use?

According to the Visier research as reported by HR Dive and Fast Company, many employees feel pressure to use AI, are unsure of their employer’s AI plans and worry about their careers. The report points to a lack of trust in leadership as a driver. When expectations are vague and visibility is rewarded, performing adoption feels safer than admitting uncertainty.

How should managers measure AI adoption without encouraging pretending?

Measure the work, not the tool use. Define what good work looks like in a given role, then ask people to explain their decisions, including what AI contributed and what they verified. Pair expectations with training and honest communication about where AI fits.

Standards, not theater

Every organization says it wants real AI adoption. But real adoption is quiet. It shows up as better decisions, fewer errors and time put to better use. It rarely shows up as someone announcing how much they used a tool.

If people are performing AI, it is worth asking what we taught them to perform.

So here is my question for you: on your team right now, what gets noticed more, using AI or using good judgment?

Sources