There is an intuitive way to read AI arriving in office work, and it is almost always wrong. The intuition says: AI does the knowledge work, so the people with the most years, the most expensive and the slowest with new tools, are the first to go. Sounds logical. The data says the opposite.

The thesis here is for anyone running a team: AI does not make your expert obsolete, it makes them your scarcest asset. And if you push them into early retirement to save payroll, you are not cutting a cost, you are giving away the one moat AI cannot hand your competitor.

The dilemma, as it arrives

Business Insider framed it as a personal dilemma: the senior professional weighing whether to learn AI or retire early. The real part of the dilemma is in the adoption numbers. A Pew survey found that, in early 2025, only a quarter of people aged 50 to 64 had used ChatGPT, against 58% of those under 30.

That gap exists and I will not downplay it. Plenty of experienced people look at the AI wave, count the years left to retirement, and decide the learning curve is not worth the fight. They leave. And every time one leaves, the company loses something that was never written down anywhere.

But the dilemma, framed only as “adapt or exit,” hides the most interesting finding in the whole story.

Why AI raises the value of experience

Here is the twist almost nobody saw coming, and it came from AARP, which knows a thing or two about older workers. Heather Tinsley-Fix, its senior adviser on employer engagement, put it this way: this is the first time she has seen a technological innovation benefit the older worker more than the younger one in job security.

Read that twice, because it breaks the script. Why would that be?

Because of how AI cuts the ladder. AI is excellent at the bottom-rung tasks, the entry-level work juniors traditionally did to learn the trade. Those rungs are exactly the ones automated first. What stays standing, and grows more valuable, is judgment: the ability to look at what AI produced and tell the correct result from the plausible filler it also generates. And that judgment does not download. It is built over years of watching things go right and go wrong.

This connects straight to something I wrote days ago: that review is the new bottleneck. AI made generating cheap and reviewing with judgment expensive. So who reviews with judgment? The person with twenty years in the craft who spots the subtle error at a glance, not the one who started yesterday. AI did not devalue experience. It moved all the value in the system toward exactly where experience lives.

Your senior people stopped being the ones who execute fastest. They became the ones who approve with authority. When anyone can generate volume, the rare thing is the person who can approve it with authority.

The mistake leaders make

With that data on the table, the mistake many companies make becomes obvious and expensive. They treat experienced people as the cost to cut when AI arrives, when they are the very asset AI makes more valuable. They push toward the door the one person able to tell them whether what the agent produced is any good.

It is rarely malice. It is the same trap I described when I wrote about how expert certainty blocks AI adoption, just seen from the other side. There the risk was the expert who blocks out of identity; here it is the leader who discards the expert out of age bias. Two ends of the same error: mistaking experience for a nuisance instead of reading it as the judgment the system needs most.

And there is a layer that gets lost that almost nobody counts: knowledge of how the processes actually work. That person knows why a given exception exists, which customer breaks if you change a step, where the gaps live that appear in no manual. That knowledge is the raw material of any automation that actually works. When you let it walk out through the early-retirement door, you do not automate better. You automate blind.

What we do at IQ Source

Our way of bringing AI into a company depends, literally, on the people many are pushing toward the exit.

The discovery phase of AI Maestro runs on the process knowledge only your seniors hold. To map how an operation truly works, we do not ask how it is supposed to work: we ask the person who has done it for years and knows every exception. That map is what later decides where AI adds value and where it would be an incident. Without those people, the map is a fiction.

So when we help a company adopt AI, the seniors are not the cost to cut. They are the architects of the change. AI absorbs the routine half, and they move up to the role the technology makes critical: setting the criteria, reviewing with authority, deciding what ships. It is the same logic I described yesterday when IKEA reskilled 8,500 people instead of firing them, applied to the most underrated asset you have.

Next time you look at your team wondering where to cut when AI lands, flip the question. Who in this room can look at what an agent produces and tell me, with authority, whether it is any good? That person is not your cost. They are your moat. Do not retire them. Promote them.

Turn your experienced people into your AI advantage