AI Is Making Expertise More Valuable, Not Less

One of the more interesting things I've noticed over the past couple of years is that AI hasn't made me value expertise less. If anything, it's had the opposite effect.

When tools like ChatGPT first appeared, there was a lot of talk about expertise becoming less important. If anyone could generate decent code, write a legal document or produce a business case in a few minutes, surely the advantage of having years of experience would start to disappear. I can see why people thought that.

AI is incredibly good at getting you to a reasonable first draft. It can write, analyse, summarise and generate ideas faster than any of us. That's a genuine shift in how knowledge work gets done.

What I've found, though, is that getting to a first draft was never really the hard part. The hard part is knowing whether it's any good.

I've lost count of the number of times I've watched AI confidently produce something that looked perfectly sensible until someone with real experience read it. They spotted the hidden assumption, the edge case that had been missed, the legal risk, the customer impact or the technical constraint that completely changed the answer.

None of those things happened because they wrote better prompts. They happened because they'd spent years building intuition. That's the bit AI hasn't replaced. If anything, it's become more important.

The more capable AI becomes, the less time experts spend creating from scratch and the more time they spend doing the things only experts can really do. Framing the problem properly. Asking better questions. Challenging assumptions. Making judgement calls when there isn't a single right answer.

Those skills have always mattered. They're just becoming more visible because AI is taking so much of the routine work off people's desks.

I think we're going to see this play out in every profession.

A junior lawyer can produce a much better first draft than they could two years ago. That's fantastic. It also means the partner reviewing that draft becomes even more important because they're no longer correcting grammar or rewriting paragraphs. They're looking for the subtle things that could change the outcome of an entire matter.

The same thing is happening in engineering. I spend far less time worrying about whether AI can generate a piece of code. Most modern models are surprisingly good at that. What interests me is whether someone understands the broader system they're building. Have they thought about security? Operational risk? Failure modes? Customer behaviour? How the solution fits into the rest of the organisation?

Those questions don't disappear just because the code was generated more quickly. In fact, I'd argue they become even more important because we're able to build so much faster. That's why I don't think AI is lowering the value of expertise. I think it's quietly changing what expertise looks like.

For years, we rewarded people for producing work. The best engineer wrote the most code. The best lawyer drafted the strongest documents. The best consultant built the biggest slide deck.

Increasingly, I don't think that's where the value sits.

The value is in knowing what should be built in the first place. It's in spotting the flaw that everyone else missed. It's in knowing when the obvious answer isn't actually the right one. And perhaps most importantly, it's in understanding the business well enough to ask a better question before anyone starts looking for an answer.

That's one of the reasons I'm so optimistic about AI.

I don't think it's making human expertise obsolete. I think it's clearing away more of the routine work so expertise can finally be spent where it's always had the greatest impact: judgement, context and decision-making.