Performance Management Needs New Metrics in the AI Era | Harvard Business Review - BRIAN HEGER

If you find value in content like this, sign up for my Talent Edge Weekly newsletter.

As AI reshapes how organizations execute work, many HR practitioners are rethinking how performance management may need to evolve, including how performance is evaluated. This new HBR article notes that many organizations still evaluate employees with pre-AI metrics like productivity and goal completion, creating a performance paradox: employees who rely heavily on AI look highly productive, while those who slow down to verify or correct outputs can look less efficient, when they are adding the most value. For example, one field experiment (published in Organization Science) on 750 knowledge workers using GPT-4 found workers were 25% faster and 12% more likely to succeed on tasks within the AI’s capability, with higher-quality output. But on a task just outside that capability, workers using AI were 19% less likely to produce a correct solution than those without it. The authors of the HBR article propose a three-layer measurement framework: 1) metrics for human contribution, 2) metrics for the AI system or AI agent, and 3) metrics for how well the two perform together, including a complementarity index measuring whether a person genuinely added value (e.g., catching an error or reframing a problem) beyond what AI produced. From my standpoint, the complementarity index is conceptually clear but would need to be operationalized, including how incremental value gets determined and defined, by whom, and at what standard. Is there a simplified version of this your organization could pilot? Regardless, the article offers various considerations for evolving performance management as AI itself continues to evolve.