Who Needs Expertise!
"AI to replace experts?" Okay… but are we even ready for expertise?
Consider this example.
"ChatGPT" I tell GPT that I’m interested in Beethoven. I want to start listening to him and build appreciation for the craft. Where should I start?
GPT takes the question and responds with a list of pieces, neatly sorted in order of suggested listening, all backed by solid reasoning. It also provides ways to build appreciation, with tips for each piece. Clean, efficient, impressive.
"Human expert" I ask the exact same question to a friend who’s been listening to Western classical music for years.
Instead of answering, he asks, “Why Beethoven?” Then follows a short conversation—more like an interview—through which he figures out that I’m not really interested in Beethoven. I’m fascinated by the feel and aura of classical Western music. Maybe I’m drawn to its grandeur. Since he knows me, he adds: this might be another case of me romanticizing the old because I haven’t explored what’s new and upcoming.
After all this, he gives me a list of pieces to start with. He tells me what to specifically listen to and what to try finding appreciation for. And adds: “If you enjoy these, I’ll share the next set.”
I ask, “What about Beethoven?” He says, “Not for a few months at least.”
This is what expertise looks like.
Because one of the ways in which experts stand apart is their craft of identifying the actual question.
They focus first on what is really being asked, before thinking of how to answer it. That’s probably also why they invest in long-term learning and practice, because that’s what lets them know what’s worth asking in the first place.
In doing so, they often go beneath the surface. That’s why talking to experts can be exhausting. You ask what you believe is a “simple” question, and instead, they ask you fifteen "difficult" ones.
So before asking whether AI can replace experts, maybe we should first ask: Are people even prepared to deal with what expertise actually demands?
Here's the thing: GPT can provide the kind of response my expert friend did. But for that, the user needs to prompt it like they are seeking an expert. And very few will do that. Most users will read the list GPT provided, map the appreciation points like it’s a competitive exam question, and then go on to quote it in conversations.
Will this apply to everyone? Of course not. Many will genuinely seek to understand, reflect, and build a deeper appreciation.
But in a world chasing speed over depth, it’s not that expertise is being replaced, it’s that it’s increasingly being replaced by "what can come across as expertise".
So while AI might join the list of experts, who or what will grow the number of people actually seeking expertise? Hopefully, AI will figure that out too.