The takeaway: Many employers believe in the huge productivity gains promised by artificial intelligence providers and have pushed their workers to use generative AI as much as possible. However, recent research suggests that uncertainty leads employees to exaggerate or fake their AI use to please their bosses. Above all, workers want clearer instructions on how to use the technology.
According to the findings, which come in part from research by business intelligence firm Ziel, nearly half of U.S. employees surveyed misreport how often they use generative AI in the workplace.
In a trend the company calls “Performative AI,” 48% of the 1,000 full-time U.S. employees surveyed said they exaggerated their AI use or expertise to colleagues or bosses. The main reason for this is that employees want to appear to be familiar with AI tools, but are not adequately trained or guided. Nearly half of respondents, about 45%, say they feel pressured to use AI but are not confident they can do so effectively.
Uncertainty around AI tools and employers’ plans for the technology is the main theme invisor’s findings. Many employees say their leaders don’t fully understand AI’s capabilities or how it truly impacts the workplace.
Only 28% say their employers understand the use of AI in the workplace, 29% believe managers fully understand their AI-related concerns, and only 33% believe their managers understand how much training they need. Additionally, 32% say their employers do not have clear plans to measure the impact of AI on the workplace, and 27% doubt their managers have plans.
Most importantly, employees want more training and clearer instructions on how to use AI. Almost a third, about 32%, want more training and opportunities to improve their skills; 31% want more transparency and 18% want clearer messages about the impact of AI on their work.
Companies that are investing heavily in AI – after Nvidia, Microsoft and OpenAI promised them a productivity explosion – have been monitoring their employees’ AI usage for some time. Last year, Microsoft implemented tools to help companies measure Copilot usage, essentially turning it into a game.
However, similar practices at Amazon earlier this year resulted in employees increasing usage to the top rankings. Although the company said it does not evaluate employees based on AI usage, sources claim that token usage factors into performance evaluation. Meanwhile, Duolingo has already stopped similar practices this year.
While some employees in thevisor survey say AI has increased team cohesion, several programmers previously complained that the pressure to use the technology stunted their skills. The top commentator, responding to a recent Ars Technica report on AI coding, echoed this sentiment, saying that executives don’t understand what’s happening on the ground and that much of their job now involves filtering out buggy AI-written code.
