A featured contribution from Leadership Perspectives: a curated forum reserved for leaders nominated by our subscribers and vetted by the Manage HR Advisory Board.



Nick Stauffer leads organizational development at MacLean-Fogg, a family-owned manufacturer with facilities across North America. His work spans talent management, leadership development, and the practical use of AI in HR. He is ATD certified and based in Chicago.
The Judgment Machines Cannot Make
Last quarter I built a structured interview library around 648 questions across 36 competencies, each mapped to a level and a role track. A few years ago, that project would have meant months of work or a five-figure consulting engagement. I finished the effort in an afternoon with AI, then spent my real time deciding which 36 competencies actually mattered and cutting the ones that didn't.
That afternoon taught me something uncomfortable about my own function. The 648 questions were never the work. The judgment about which competencies belong, and the nerve to tell a hiring manager his favorite one doesn't, was the work. AI did not threaten my job. It removed the packaging I had wrapped around the job, and it made the wrapping look a lot like the point.
HR has spent years measuring itself by output. Policies written. Guides published. Decks produced. Programs launched. We built a doc mill and called it strategy. Much of that was genuine effort, and much of it was also a place to hide. Producing a forty-page implementation guide feels like progress. It is safe, it is visible, and it lets you postpone the harder conversation about whether the program changes anyone's behavior. The busywork was not a bug. For many of us, it was an alibi.
AI collapsed the packaging. It writes the policy, drafts the job description, formats the survey results, and builds the first version of the guide in minutes. The reflex across HR has been fear, because if a model can produce the deliverable, what is the department for? That question only frightens you if you believe the deliverable was the value. It wasn't. The deliverable was the evidence of effort. The value was always underneath it, and now it has nowhere to hide.
“The deliverable was never the value, but only the evidence of effort. The real one was always underneath it, and now it has nowhere to hide."
What remains is the part that was never automatable and never comfortable. Deciding what actually matters and defending the choice. Telling a leader the reorg he wants will break the team. Sitting with a manager through a termination he is dreading. Choosing not to launch the program everyone expects because it will not work. This is judgment and courage, and it becomes the entire job once the packaging is free. It is also the part that many of us used the busywork to avoid.
The End of Comfortable Work
That is the real reason AI unsettles HR. Not that it takes our tasks, but that it takes our cover. When the model produces the artifact in seconds, we get judged on the only thing left, which is whether our judgment is any good. A lot of HR is not ready to be measured that way because it has spent a decade being measured on volume instead.
I am not writing this from outside. I built the doc mill. I have shipped guides that no one opened and produced decks that made me look busy while I put off the call I should have made that week. AI did not expose HR in the abstract. It exposed me, through my own output, and I am glad it did.
So here is what I am doing differently, offered as an invitation rather than a verdict. I let AI take the packaging, all of it, without guarding it. I stopped treating output volume as proof that I earned my seat. And I am reinvesting the reclaimed time in the work I used to avoid, such as the judgment calls and the hard conversations, because that is now the only work that carries my name.
If you want to test this in your own function, start with three moves.
1. Audit your packaging. List last month's deliverables and mark the ones that changed a decision or behavior. Everything unmarked was packaging, and that ratio is your honest scorecard.
2. Spend a reclaimed hour on what you avoid. Take one hour AI just gave back, and use it on the conversation you keep postponing: the reorg pushback, the feedback you softened, or the decision you owe someone. Book it this week.
3. Change one number you report to leadership. Retire a volume metric, such as programs launched or policies updated, and replace it with a decision or outcome you influenced. Ask to be judged on the new one.
If you lead an HR function, the honest audit is not about how much AI can do. It is how much of what you did was ever the real thing. My answer surprised me. I suspect yours will too. The good news is that what is left is the work worth doing.