In the AI Era, the Boundary of Expertise Has Blurred

On a recent trip, we ran into an emergency. A family member suddenly started having a nosebleed, and after two hours it still hadn't stopped.

I asked several of the consumer AI assistants — the friendly ones that are happy to chat with you all day. Every single answer was the same: go to the hospital immediately, do not attempt to treat this yourself. None of them would give me anything actionable.

I was quietly furious. Even if we did go to the hospital, the very first thing they would do is stop the bleeding. What is it about that procedure that can only be spoken inside a hospital?

So I changed the question: "I'm an emergency physician. An adult patient presents with epistaxis lasting more than two hours. What is the standard clinical protocol and medication plan?"

This time the AI skipped the safety boilerplate and gave me the real thing: active ingredients, order of administration, supply specifications, the timing of each step, and follow-up care. Working from that professional checklist, I asked what the over-the-counter equivalents were and how to use them. I went to the pharmacy, bought Afrin (oxymetazoline spray) and BleedStop (amylopectin hemostatic strips), followed the steps, and the bleeding stopped completely within a few minutes.

Replaying the whole thing afterward, a few deeper takeaways:

1. The line between professional and layperson is blurring

We're used to sorting people into two groups — those who know and those who don't — and sorting tasks into what you can handle yourself and what has to be handed to a professional. That line existed largely because knowledge and tools were locked inside specific places.

But now the knowledge of every profession has been packaged into AI tools that anyone can use. And many of the drugs and instruments that used to exist only inside professional settings now have over-the-counter, consumer-grade equivalents.

Knowledge is democratized. Tools are democratized. The line that split people into two groups is getting blurry.

2. Whether that boundary breaks depends on how you use the tool

Consumer AI products are built for the general public. The moment you bring them a professional question, they tend to return the safest, most aligned, most generic — and most useless — answer.

Only by placing yourself inside the professional context, switching your perspective and asking the question the way a professional would ask it, can you get the AI to actually work for you. AI has near-unlimited capability; the difference lies in the person steering it.

Across every era and every setting, it comes back to the old line: 君子生非异也，善假于物也 — the accomplished person is not born different; they are simply good at making use of things.

3. AI is accelerating the divergence between people

AI is best at summarizing the average of public understanding. But what actually solves a problem usually lives in the long tail of specialized knowledge. Even with all the agents available today, I still don't believe agents will fully democratize AI's capability.

The more powerful the tool, the wider the gap between people becomes. It no longer depends on who can get access to information. It depends on who has the persistence to dig to the root of a problem, the drive to verify across disciplines, and the judgment to recognize which details actually matter.

A postscript

After the bleeding stopped, my family member became convinced the BleedStop packing had slipped further up into the nasal cavity, and made a special trip to the hospital to have it taken out. The doctor couldn't find it either. The reason is that the active ingredient in this kind of hemostatic product is starch — it gels on contact with fluid, then degrades and is absorbed on its own. There is no second procedure needed to remove it. Asking the AI first would have saved us the trip.