A new TalentLMS survey of 1,200 U.S. employees finds that 41 percent say their role has evolved faster than their company’s training can keep pace, and nearly 60 percent now use AI tools at least sometimes to complete tasks they were never formally trained on. The gap is not showing up in performance reviews because employees are quietly compensating for it: 62 percent say they use workarounds to cover skill deficiencies, and 47 percent stay silent about the gaps rather than flag them to a manager.
The more striking number is the 29 percent who say they have delivered work they could not fully explain if questioned on it, a sign that AI is not just filling skill gaps but actively concealing them. “AI is blurring the line between learning and doing,” said Dimitris Tsignos, CEO of Epignosis, TalentLMS’s parent company. TalentLMS researchers put it more bluntly: “The result is a workplace where performance doesn’t reflect capability.” That blur is exactly what makes the risk hard for HR to see, even as demand for AI skills keeps climbing in job postings while the underlying competence behind AI-assisted output goes unmeasured.
The original insight: performance metrics built for a pre-AI workplace now reward the appearance of capability rather than capability itself, which means people-analytics teams may be tracking the wrong signal entirely. A learning function that only trains employees on tools, not on the judgment to know when AI output is wrong, is financing a debt it cannot see on the balance sheet until a fast-changing role breaks under real pressure, and the survey suggests that moment is closer than most L&D catalogs are built for.
