Rewarding Judgement in AI Use

AI is changing which employee skills create value and how to incentivize these skills. EY recently announced a $100 million employee rewards program aimed at recognizing workers who develop “future-focused” capabilities and use technology to produce better outcomes. The interesting part is what EY has chosen to reward. Along with technology adoption, the firm emphasizes business acumen, judgment, adaptability, experimentation, collaboration, and innovation. As AI makes some technical capabilities more abundant, human judgment can become increasingly scarce and valuable. EY is not simply paying employees to use more AI; it is rewarding them for developing skills that make AI more productive.

That changes the incentive problem. If AI substitutes for routine research, analysis, or coding, producing more of those activities becomes a less useful measure of employee performance. Knowing which questions to ask, recognizing when an AI answer is wrong, exercising judgment when information is ambiguous, and finding new ways to combine AI with expertise can become more valuable. But these qualities are also difficult to measure. Counting hours, completed reports, or even AI usage is relatively easy; determining whether an employee exercised good judgment is not. EY's new program therefore allows employees at all levels to recognize colleagues, while other EY initiatives use simulations, coaching, and assessments of decision-making and collaboration to identify these less tangible skills.

Moving from objective output measures toward judgment, adaptability, and collaboration also changes the role of supervisors. Managers must have more discretion to decide whose contributions actually created value, making the quality and credibility of those evaluations more important. EY has argued that AI-era performance management should place greater weight on peer feedback, coaching, adaptability, and cross-functional impact, rather than relying simply on traditional manager ratings. It is also investing in more structured ways of assessing these skills: its new Career Residency program uses workplace simulations, coaching, and client situations to assess decision-making, collaboration, learning, and adaptation. As easily measured tasks become easier to automate, performance evaluation may consequently depend more on subjective judgment, but also on multiple evaluators and better ways of evaluating the evaluators.