The AI adoption gap at the top of the house.

I asked a room of people leaders recently to score themselves, honestly, on how far AI has actually changed the way their function works. Not the pilots. Not the vendor demos. The everyday work. The average answer landed around a three out of ten. When I ask the same question of their peers in product, engineering, or finance, the number comes back a seven or an eight.

That gap is the story. AI has transformed how leaders talk about the future of work. It has not yet transformed how most companies operate, and nowhere is the distance between rhetoric and reality wider than in the function that owns the workforce.

Why the function that owns talent is behind on the tools that reshape it

It isn’t for lack of interest. It’s that the people function carries the most human-sensitive data in the building, the highest bar for fairness, and the least tolerance for a confident-but-wrong system. Add tooling that was built to be sold to IT rather than used by talent teams, and caution hardens into paralysis. Meanwhile the demands keep rising: boards now expect the CHRO to build AI fluency across the whole organization, while the function models the least of it.

The trap: reinventing the function instead of removing friction

The teams that stall are usually the ones trying to reimagine everything at once: a grand AI operating model, debated for a year. The teams that pull ahead start narrow. They point AI at the friction already in the work: the searches that take weeks, the market maps built by hand, the candidate signal buried across a dozen systems. The fastest value comes from doing what you already do (faster, deeper, with less guesswork), not from a moonshot.

What closing the gap actually looks like

• Start where the pain is measurable. Pick one repeated, high-stakes task (mapping a talent market, screening a slate, reading retention risk) and let AI compress it. Prove it, then widen.
• Treat talent intelligence as infrastructure, not a project. The read on who exists, where, and why they’d move should be always-on, not commissioned per search.
• Keep judgment human. AI is leverage on the decision, never the decision. At the top of the house, the last mile is still people reading people.
• Buy access, not overhead. Most people teams don’t need an enterprise AI contract and a new headcount to run it. They need the capability, on demand, without the platform tax.

The good news is that the gap is a choice, not a fate. The technology is ready. The question is whether the people function will lead its own transformation before it’s asked to lead everyone else’s, because that ask is already here.