Nearly half of workers say they’re exaggerating their AI use at work

Sponsored by

Protect your process. Talent leaders are being asked to hire faster and protect against threats while delivering a top-notch candidate experience. With the right tools on your side, you can make it happen. Persona’s identity verification solution helps stop fraud at the top of the hiring funnel without impacting a qualified candidate’s experience.

Most teenagers are familiar with the ruse: Mom asks if you’ve cleaned your room. You reply, “yes,” because you did, indeed, toss some of your dirty laundry in the basket before logging onto Xbox. But it’s not the full truth; it’s mostly false, but to a teen, it feels like enough of the truth.

American workers have employed a similar tactic when it comes to AI use, according to new research from Visier.

The survey of 1,000 full-time US employees found that 48% exaggerate their AI usage or expertise to colleagues or company leadership at least some of the time. Visier has identified this as “performative AI”: overstating and overplaying AI use in order to meet employer expectations.

“Something I wanted to understand: How often do people actually feel they have to use AI because of all these mandates?” said Andrea Derler, Visier’s principal researcher and author of the report. “We discovered that almost half of employees actually do feel that pressure to use it for basically all their workflows, whether they make sense or whether they don’t make sense.”

This group of employees are participating, going through the motions, but they’re not meaningfully incorporating AI into their workflows, or they don’t understand how, she said. The findings articulate a potentially expensive problem for organizations working to extract productivity gains from the AI tools they’re investing in and developing.

“I don’t want to blame the employee here. That’s really important to me…That’s not the point. We all need to have a certain degree of impression management in our jobs. Everybody has to,” she said. “But in this case, because we have to learn so much, because it’s so critical that we get this right, because it’s so expensive not to get it right, that’s why it really matters.”

AI use as a measure of progress amid the AI transformation, may be more akin to the concept of tokenmaxxing as an indicator that employees were deeply engaged in working with the technology…meaning they’re actually not. AI adoption may simply measure adherence to a company’s AI-use mandate, rather than their engagement with the tool.

“We miss really important opportunities and learning opportunities and value creation opportunities if we give a goal like tokenmaxxing or in usage of AI, rather than investigation and critical thinking about where it could be helpful where it can’t, and that is not a convenient message,” she said.

Many leaders missed an opportunity to learn at the jump how their teams would be using AI and what they should be learning in order to best make use of the technology, she said, and workers are losing confidence even that their leaders and managers can lead on AI.

HR has an important role in helping managers create a culture of psychological safety and encourage a growth mindset, she said. HR leaders can help organizations build a culture that fosters more AI experimentation and creates space to talk about when things go wrong.

Some 45% of workers say they feel pressure to use AI even when they aren’t confident they know how to use it effectively, according to the report. What would raise employee confidence? Survey respondents pointed to AI training and upskilling (32%), transparency from leaders (31%), opportunities for career advancement (24%), and clearer communications (18%). Only 12% of respondents felt no intervention from their employer would move the needle when it comes to confidence.

“We’re delaying the learning. We could be much more efficient, faster if we allowed each other…[to] an experiment with AI. Let’s really deep dive into what works, what doesn’t work, versus having the tokenmaxxing or the AI-usage maxing, because you get the wrong behaviors,” she said.