Are You Too Old for AI?

The short version: A large study of Novo Nordisk's AI rollout found that senior employees outperformed younger ones — not because of age, but because experience told them where the tool fit their work and when its output was wrong. The real lesson isn't "experience wins"; it's that AI rewards a fit between the tool and the shape of your work, and your expertise is what lets you see that fit. If you've spent decades getting good at something, you may be exactly the person the tool has been waiting for. (8-minute read.)

I recently joined a workshop on making the most of LinkedIn, run by a top influencer — a very good presenter, too. At one point she made something perfectly clear: her platform hadn't been built by "some 40-year-old guy." The comment wasn't really about the platform. It was about who gets to belong in the AI conversation — and who, by implication, doesn't.

I've thought about that line more than I'd like to admit. It stung a little — I'm certainly not in my twenties — but what stayed with me was the implication: that AI belongs to the young, the digitally native, the people who grew up with a phone in hand. If you're older, more senior, more experienced, the story goes, you're playing catch-up at best and an obstacle at worst.

It's a tidy story. It's also, as far as I can tell, wrong. And to be clear, this isn't an article about older professionals finally figuring out AI. I have two stories that complicate things nicely. One: a 27-year-old finance executive in one of my courses sheepishly admitted he had never once used ChatGPT for anything — not even to check the weather over the holidays. Two: a participant in another course told me afterward, "This has been fantastic — at my age, I thought I'd never be able to use AI." She's thirty.

So which is it? If you're older, AI supposedly isn't your thing — never mind that AI itself has been around since the 1950s, which technically makes it an invention of "old people." And if you're young, you're apparently uncool for not adopting it fast enough — what's wrong with you? Neither story is interesting, and neither is true. You don't have to choose between being filed away as obsolete, quietly reassigned to a department where no one asks too many questions, or being treated as generationally unfit for not embracing a tool fast enough. Both framings miss the actual variable that matters.

What actually happened inside a 75,000-person company

In 2024, Novo Nordisk began rolling out Microsoft Copilot to its global workforce. It grew from a few hundred users in January of that year to 20,000 by February 2025, with plans to expand further — one of the largest documented generative AI deployments anywhere. Researchers from IMD's Tonomus Global Center for Digital and AI Transformation studied the rollout alongside two Novo Nordisk executives, and published their findings in MIT Sloan Management Review in mid-2025, under a title that undersells how interesting the result is: "How to Scale GenAI in the Workplace."

Buried in that piece is a section the authors call "Surprise Champions: More Experienced Employees." The company's leadership had assumed, reasonably enough, that younger staff — the digital natives — would be the ones pulling everyone else forward. The data said otherwise. Senior employees outperformed their junior colleagues on both productivity gains and the quality of the work they produced with the tool. So experienced professionals can actually use GenAI better?

The explanation the researchers landed on is worth sitting with, because it's not what you'd expect if you believed the "40-year-old guy" story. It wasn't that senior staff were somehow more comfortable with software, or less intimidated by new interfaces. It was that they knew the terrain. Years of doing the actual work had given them a map of where the friction points were, which tasks were genuinely hard, and which judgment calls mattered — so when a new tool showed up, they could see immediately where it fit and where it didn't. They were also better positioned to check its output. Knowing enough about a domain to recognize a bad answer turns out to be at least as valuable as knowing how to type a good prompt.

The junior employees, meanwhile, weren't held back by unfamiliarity with the technology. They were held back by not yet having enough experience to know what to do with it. One put it plainly: "I don't know enough actual use cases; what can I use it for?" These weren't people who couldn't operate Copilot. They were people who hadn't yet lived through enough versions of the problem to recognize where it could help.

That distinction matters enormously, and it's the one the "too old for AI" narrative erases entirely. The barrier was never speed with a keyboard. It was context — and context is something you accumulate by doing the work.

The line that says the quiet part out loud

The researchers' own conclusion is the one I keep coming back to:

Generative AI performance isn't about tech savviness — it's about contextual fluency, confidence, and the human ability to integrate new tools into nuanced workflows.

Read that again next to the influencer's comment about the 40-year-old guy. They are describing two different worlds. In one, AI adoption is a generational sorting exercise — younger wins, older loses, and the loss is treated as more or less inevitable. Which is funny, given my own two stories: a 27-year-old who'd never touched ChatGPT and a 30-year-old convinced she'd missed her window. In the other world, adoption is a question of who understands the work well enough to know where a new tool actually earns its keep. Novo Nordisk's own data sided with the second world, inside a company with every incentive to believe the first.

Once the pattern was clear, Novo Nordisk didn't just note it and move on — they restructured around it. They built a cross-functional network of champions, drawn largely from their more experienced staff, to run peer demos, deliver training tailored to specific roles, and share concrete examples grounded in real workflows rather than generic tutorials. They paired that with internal social platforms that let senior and junior employees trade what they were learning in both directions. The fix, in other words, wasn't to wait for the experienced staff to catch up to the young. It was to put the experienced staff at the front of the room.

A caveat, because precision matters here

This is one company's rollout, documented by researchers working with two of that company's own executives. It's not a randomized study across industries, and I wouldn't treat it as a universal law of AI adoption. But it is a large, carefully observed, real-world case — 20,000 people, tracked over more than a year, inside an organization with no obvious motive to flatter its senior staff. As a counterweight to a confident generalization made in a workshop, it's a serious one.

It was never only about age

There's a second finding in the study, and it complicates the first one in a useful way. Copilot didn't land evenly across the company. Commercial teams — the people writing, summarizing, planning, pitching — got far more out of it than research teams did. On the surface that looks like just another divide to file away. But the reason matters, and it isn't what you'd guess.

It wasn't that research staff were worse with technology, or more resistant. It was that a general-purpose generative assistant is a particular kind of tool: it's probabilistic. Ask it the same question twice and you may get two different answers, and now and then it will invent one. That's a wonderful property if your work runs on fast drafts you then shape — a first pass at an email, a summary, a way into a problem. It's a liability if your work demands one exact, reproducible answer.

And here's the part worth pausing on, because it's easy to get wrong: this does not mean research is "beyond AI." Novo Nordisk is a pharmaceutical company. Its scientists use AI constantly — to predict protein structures, to screen molecules, to model things no human could hold in their head at once. That work is about as precision-critical as it gets, and it is deeply AI-driven. The point isn't that AI doesn't suit science. It's that a chatty, general-purpose generative assistant was the wrong AI for that particular job — while the purpose-built, validated models those teams already rely on are exactly the right ones.

Which quietly dismantles the "experience always wins" reading before it can harden into a new myth. It was never simply that seniority beats youth. The senior employees did well because their experience let them see where this tool fit the shape of their work. The research teams struggled because, for much of their work, it didn't — and they had better AI elsewhere. Same lesson, drawn twice: what matters is the fit between the tool and the work, and experience mostly helps by letting you see that fit before you discover it the hard way.

Why this matters beyond one company

I'm not interested in scoring a point here — I'm interested in how people actually perform. Over the years I've watched plenty of bright, capable participants convince themselves they simply weren't built to master a piece of technology. I heard versions of it about Adobe Flash, back when Flash was the thing everyone needed to know (I know you're now counting on your fingers to figure out how old I am). I hear a version of it now about AI.

Here's what I tell them instead: BYOL. Bring Your Own Life. Your experience — professional and personal — is not dead weight you carry into the AI conversation. It's the unique perspective only you can bring, if you know how to use it. Technology doesn't have to be boring, and it doesn't have to be learned in a vacuum. Maybe you spent years training in classical piano — I never did, and I regret it more every year — and that training gave you an ear for structure and improvisation that shows up in how you'd approach an AI workflow no one else would think to try. Maybe it's videogames, or gardening, or something quieter still. In the Zen tradition, practices like sumi-e ink painting exist to help a person find their center — not to produce a masterpiece, but to locate something true about how they see and work. The same instinct applies here: your outside life isn't a distraction from learning AI. It's frequently the source of the most original way you'll end up using it.

And if you're at the other end of that spectrum — younger, earlier on — the message isn't "you're behind," because you're not. What the senior colleague has isn't more talent or more nerve; it's more reps. They've simply seen the problem enough times to recognize where a tool fits. That's not a wall, it's a to-do list: get close to the hard problems, learn the work underneath the job title, and bring your own life to it too, because you already have one. Context is the rare asset that compounds faster the more deliberately you go looking for it.

AI is no exception to the pattern I've spent my career studying. The tool rewards the person who already knows what a good answer looks like in their domain, who can tell when a plausible-sounding output is quietly wrong, and who has enough scar tissue from doing the work the slow way to recognize a genuine shortcut when one appears. None of that shows up on a birth certificate. All of it shows up in a career.

So if you've spent twenty or thirty years getting good at something, the honest answer to "are you too old for AI?" is: probably not. If anything, you may be exactly the person the tool has been waiting for. The work now is learning where to point it — and that's a much shorter distance to travel than starting from zero ever was.

How to actually do this

Turning your own track record into an AI advantage doesn't require a technical course. It requires looking at your own expertise honestly, from the big "why" behind your work down to the small daily frictions inside it. It starts with four questions — and then one move.

1. What are you for?
Before worrying about AI, get clear on why you do this work at all — what you're actually trying to achieve, underneath the job title. This is the level everything else gets measured against, and most people skip it.

2. How do you know when it's good?
How do you currently judge whether a piece of work is excellent, adequate, or wrong? That internal standard — built from years of doing the job — is exactly what most AI tools lack on their own. You're about to become the thing that supplies it.

3. What do you do brilliantly?
What are the two or three things you do that colleagues quietly come to you for? Not your job description — your actual, demonstrated strengths. This is your highest-value territory.

4. What drains you?
Now look at the parts of your work that are routine, repetitive, or just draining — the tasks that don't need your judgment, only your time. Be honest here. They don't require the expertise that makes you valuable, which is exactly why they're worth handing off.

Then the move: point AI at the gap, not at yourself.
With questions three and four answered, the question changes from "should I use AI?" to "where does AI actually fit into how I already work?" Use it to extend your signature strengths, and to absorb the friction you flagged — informed by whatever else you bring to the table, piano lessons, gardens, and all.

Do this once, seriously, and the question "am I too old for this?" tends to answer itself. You were never behind. You were just standing on more experience than the tool knew what to do with yet.

---

Reference: Wade, M., Trantopoulos, K., Navas, M., & Romare, A. (2025, July 8). How to Scale GenAI in the Workplace. MIT Sloan Management Review. https://sloanreview.mit.edu/article/how-to-scale-genai-in-the-workplace/