Visier survey finds almost half of US workers have overstated their AI use or skill

Derler's account of where the pressure comes from runs up the org chart as well as down. Leadership may feel pressure to show investors that AI is working for them, she said, while other employees feel it through fear of losing their jobs ^[7]. That fear has some footing: a June report from outplacement firm Challenger, Gray & Christmas cited AI as a major factor behind job cuts ^[8]. "People feel they have to constantly prove how great they are with AI and how much value they can create with AI," Derler said ^[5]. An executive who needs adoption figures for investors and an employee who wants to survive the next round of cuts both gain from the same high number ^[7].

Visier's finding is about what employees tell colleagues and leadership ^[1]. Business Insider's account does not give the survey's sample size or fielding dates. The finding applies most directly to self-reported adoption: pulse surveys, and the paragraph in a performance review where staff describe their AI work. Usage logs and licence counts are a different instrument, and they are where a company can check what staff say against what they do.

The same pressure shows up before people are hired. Employers asked Cristina Estupiñán, who spent years in software development and was job-hunting after a layoff, about her enthusiasm for AI in interviews ^[13]. "I have to make up something about what excites me about AI because, frankly, nothing really excites me about AI," she said ^[14].

A skeptic would answer that forced use still produces real users. Gallup's Jeffrey Jones said in a post that workers who use AI daily or weekly "are far more likely than irregular users to report AI benefits at work" ^[16]. Among regular users, 79 percent said AI helps them finish tasks faster, against 38 percent of other workers, and 59 percent said it improves quality, against 26 percent ^[17]. Those are gaps of 41 and 33 points ^[19][20].

They are also self-reports, from a January-to-March survey by Jobs for the Future, the Families & Workers Fund and Gallup ^[15], gathered in the same climate of performative AI that Visier describes ^[2]. I think the gap is good evidence that regular use helps some people, and weaker evidence about how many. The two surveys draw on different samples, so neither can be used to discount the other by a fixed amount. Jones said, "Continued measurement of AI usage at work, employee influence over technology implementation, and job quality will be important to understand how AI is augmenting, reshaping and replacing employee tasks" ^[18].

The trade-off is between speed and signal. A mandate backed by peer comparison produces visible adoption quickly. Derler said the performative part can be a waste of time because organisations do not learn how AI is actually effective and where it falls short ^[9]. Her alternative is slower: before setting requirements, executives and managers learn how AI can be most valuable to employees, and invite people to talk about what worked, what did not and where they made mistakes ^[12]. The Visier report said, "Performative AI use seems to line up with a trust gap: employees are unsure of the organization's AI plan, and unconvinced that leaders understand it themselves" ^[11].

A mandate set this quarter produces next quarter's adoption figures, and those figures go to the investors Derler says leaders are trying to reassure ^[7]. If part of what staff report is performance, the next round of AI spending will be approved on numbers employees shaped to protect their jobs ^[7].