AI Doesn’t Eliminate Bias—It Amplifies It: A Warning for Hiring Leaders
With the right questions and the right intentions, AI doesn’t threaten the quality of leadership. It can enhance it.
I was talking the other day with the Head of Talent at a large industrial company. 50 manufacturing sites around the world. And in her role as Talent Developer, always the same recurring question: “Are my site directors really the right ones for the job?” Usually, the answer revolves around a 360 review, some KPIs, reputation and gut feeling. This time, she decided to dig deeper. Using AI.
Here’s how she proceeded:
First step: have all their site directors take the Hogan assessment. (For context: Hogan is a robust personality and leadership test. We use it too at Robertson; we’re all certified.) Then things got interesting. She went to ChatGPT and asked: “Based on these Hogan results, can you create clusters of profiles?” Of course, the AI could. It identified three distinct clusters. When she compared them with the specific needs of each industrial site, she found that in one case out of three, the profiles didn’t match the actual needs.
Look for the next “director of directors”
She took it one step further. She asked the AI: “Who, based on the data, could become our next ‘director of directors’?” The AI generated a list of ten names. She then compared it to the internal shortlist. Not a single name overlapped. Fascinating, right? 😊
Who’s right—the algorithm or the humans? Hard to say. One thing is clear: AI can serve as an extraordinary mirror. Not to replace human judgment, but to challenge it, expose our cognitive biases, and open new ways of seeing talent. With the right questions and the right intentions, AI doesn’t threaten the quality of leadership. It can enhance it. 😊
Takeaways for talent leaders
- Start with solid data — Reliable assessments like Hogan give the AI a trustworthy foundation.
- Ask the right questions — “Which profiles fit which site?” or “Who could scale to a broader leadership role?” guide the model toward actionable insights.
- Blend AI output with human context — Use the algorithmic suggestions as a diagnostic lens, not a final verdict.
- Iterate and validate — Compare AI‑generated clusters with real‑world performance, adjust the model, and repeat.
