A joint field study from KPMG and the University of Texas at Austin’s McCombs School of Business, published July 23, 2026, tracked 523 U.S. early-career professionals doing real client-style work alongside the same AI agent. Workers split into three distinct profiles: “AI Amplifiers” (50.1%) who outperformed an AI-only baseline by directing workflows, framing problems, and iteratively refining outputs; “AI Delegators” (25.8%) whose results matched the AI baseline because they accepted outputs with little scrutiny; and “AI Apprentices” (24.1%) who performed below the AI baseline despite strong underlying skills, because their attempts to critique the AI’s work sometimes steered it in the wrong direction.
Why it matters for HR: the study’s central claim is that performance gaps between early-career employees with nearly identical knowledge and skill are now explained by how they work with AI, not what they know. UT Austin professor Ashish Agarwal put it directly: researchers “weren’t simply looking for people who knew how to use AI,” but for what “enables individuals to consistently create value beyond what AI produces.” That reframes two HR functions at once: campus and early-career hiring assessments built around technical knowledge tests are measuring the wrong thing, and learning-and-development programs built around tool tutorials are training the wrong skill.
The original insight here is that this behavioral gap is not hypothetical, it is already visible in productivity data. It lines up with what we found in our earlier reporting on workers losing a full day a week to unproductive AI oversight: the “Apprentice” pattern of ineffective, time-consuming AI supervision described in the KPMG study is very likely a meaningful share of that lost time. KPMG’s own response, a firmwide program called “You Can with AI: Next Level Learning,” treats AI-direction behavior as coachable rather than innate, which is the model other employers will need to copy if they want the productivity gains AI promised without the hidden oversight tax.
Source: KPMG
