The Amplifier Paradox

A year-long study on cancer specialists found that using AI improved their output metrics while quietly eroding the judgment that made those metrics meaningful. Researchers called it "intuition rust." (totally stealing this one) The operational gains were real. So was the slow hollowing out underneath them.

This is the version of the AI-and-expertise question that doesn't get asked enough in consulting circles. We talk constantly about whether AI will replace us. The more uncomfortable question is what happens when it doesn't: when you keep your job, keep your title, keep your client relationships, but the repetition that built your judgment has been replaced by a tool that handles it faster and adequately. Adequate is not the same as good. The gap between those two words is what clients are actually paying for.

More than 11 million gateway jobs disappeared in 18 months. The entry-level work that felt like busywork (the analysis that took a junior analyst two days, the documentation that nobody wanted to write) was actually where pattern recognition got built. That's not just a labor market story. It's a training infrastructure story. We removed the reps.

A 2026 coding study found that individuals who relied heavily on AI tools scored 17% lower on manual-mastery tasks. That's the kind of drift you don't catch until you need the skill and it isn't there. I think about this on my own teams. The consultants who are most deeply involved in AI-assisted delivery are producing more, faster. I'm proud of that. I'm also watching for the moment when we hand them a problem the tool can't solve, and they have to fall back on the judgment they built before the tool existed. That moment will come. It always does.

The AI-as-amplifier story is true. It is also incomplete. The same tool that multiplies your output can quietly atrophy the capabilities beneath it if you're not deliberate about what you practice without assistance.

What are the judgment calls you've stopped making because the tool makes them adequately? Adequate and good aren't the same thing, and the gap between them is exactly what you're being paid to close.