Ten years ago, if you needed a clean financial model, a well-argued strategy memo, or a passable first draft of anything, you hired someone for it, or you spent years becoming that someone yourself. Today a general-purpose AI model produces a rough version of all three in under a minute. The skill hasn't disappeared. It's been inflated — available to almost anyone, at almost no cost, almost instantly.
This isn't the first time this has happened. It's just the fastest.
The pattern is old, the speed is new
Photography once required years of darkroom craft; a smartphone now does exposure and focus automatically. Mental arithmetic was a professional asset until the pocket calculator made it a convenience. Spelling was a mark of education until autocorrect made it invisible. Each time, a skill that used to sort people by competence became something almost nobody needed to practice anymore.
What's less often discussed is that the underlying capability doesn't just become optional — it can quietly erode. The clearest evidence comes from GPS navigation. A 2020 study by Louisa Dahmani and Véronique Bohbot, published in Scientific Reports, tracked 50 regular drivers and found that people with heavier lifetime GPS use had measurably worse spatial memory when asked to navigate without it — and that the decline was dose-dependent and specific to spatial memory, not general cognition. The tool didn't just replace the skill. Over time, it thinned the capacity behind it.
That's the trade worth sitting with: inflation of access, but not automatically inflation of ability.
What the AI research actually shows
A 2023 NBER study by Erik Brynjolfsson, Danielle Li, and Lindsey Raymond (later published in the Quarterly Journal of Economics, 2025) followed 5,172 customer support agents after a generative AI assistant was rolled out. Average productivity rose 14–15%. But the gain was almost entirely concentrated among novice and lower-skilled agents, who improved by roughly a third. The most experienced agents barely moved — and in some measures, their output quality dipped slightly. The tool worked by capturing and distributing the tacit habits of the best performers to everyone else.
That's skill inflation in a controlled setting: the floor rises fast, the ceiling barely moves. What used to separate a strong performer from an average one narrows, because the average one now has access to the strong performer's patterns.
David Autor's more recent work on task automation adds a useful qualifier: automation raises wages where it removes tasks that didn't actually require deep expertise, but it can suppress them where it eats into tasks that did. The tasks that resist this longest are the ones built on tacit knowledge — judgment gained through lived experience, not documentable procedure.
What this changes at the micro level
In my coaching work, this shows up as a quiet identity question for leaders whose authority was built partly on technical command — the person who could always produce the sharpest analysis, the tightest write-up, the fastest read on a spreadsheet. That output is no longer scarce. What's left, and what was arguably always the harder part, is knowing which question to ask, which output to trust, and which trade-off actually matters here. That's judgment under uncertainty, not task execution — and it's exactly the category of tacit knowledge that hasn't inflated away.
What this changes at the macro level
For organizations, the shift is structural, not cosmetic. If the differentiator used to be "who on the team can do this," and now almost anyone can produce a version of it, the organization's real edge moves to whoever decides what "good" means, catches what the tool got confidently wrong, and takes responsibility for the outcome. That's a smaller set of people doing a different kind of work — less production, more accountability and integration. Organizations that don't consciously make that shift tend to keep rewarding the old scarce skill long after it stopped being scarce.
Which of your own skills used to be your edge — and what's actually still yours once anyone can produce the output?
If this is a question you're sitting with in your own leadership, I coach leaders through exactly these transitions — or you can start with my book on self-coaching.
Sources:
- Dahmani, L. & Bohbot, V.D. (2020). "Habitual use of GPS negatively impacts spatial memory during self-guided navigation." Scientific Reports, 10(1).
- Brynjolfsson, E., Li, D., & Raymond, L. (2023, rev. 2025). "Generative AI at Work." NBER Working Paper 31161; Quarterly Journal of Economics, 140(2), 889–942.
- Autor, D. (2022). "The Labor Market Impacts of Technological Change." NBER / MIT Shaping the Future of Work.
Want to explore this further?
Start with a free 30-minute discovery session — no commitment required.
Book a Discovery Call