I looked something up the other day, read it, nodded along, and moved on. An hour later I went to explain it to someone and found I couldn’t, not really. I had the gist, the shape of an answer, but the thing itself had already slipped away. It made me wonder how much of what I think I know is like that: not knowledge so much as having been near an answer, recently.

It is an old observation, not a new one.  More than a thousand years before anyone had a search bar, Confucius said it: “To say you know when you know, and to say you do not when you do not, that is knowledge”. The wording shifts between translations of the Analects, but the idea remains.

I think about this from the time I spent in venture capital. That job trained one habit into me harder than any other: question your own assumptions. When you are sure you are right about a company or a market, that certainty is usually the exact thing stopping you from seeing the detail that matters, which is often right in front of you. Being confident and being correct feel identical from the inside. That is the trap.

There is a famous 1999 experiment that I think shows how literal this can get. People were asked to watch a short video and count how many times a basketball was passed between players. About half of them, busy counting, completely failed to notice a person in a gorilla suit walk into the middle of the shot. The viewers were certain they were paying attention, and their certainty pointed their eyes one way while everything outside it vanished.

Anyway, what I have slowly come to believe is that admitting you actually don’t know something is not the weak position. It is often the stronger one.

When researchers Lars Bo Jeppesen and Karim Lakhani studied a platform where companies posted hard scientific problems for outside solvers, they found the people most likely to crack a problem were often the ones furthest from its field. The odds of a winning solution, they wrote, were “positively related to increasing distance between the solver’s field of technical expertise and the focal field of the problem.” The insiders already knew what wouldn’t work. The outsiders didn’t know enough to rule it out, so they tried it anyway.

That is the upside of knowing what you don’t know, held the right way round. A specialist’s knowledge or perceived knowing can quietly become a fence. The person who says “I have no idea, but what about this” is sometimes standing in the one spot the answer is visible from.

I had a teacher in secondary school who used to say, “If you truly understand something, you won’t forget it.” I have tested that line against my own life for years, and I think it holds. The things I genuinely understand I can still explain without notes. The things I once “knew” for an exam are long gone. Understanding stays. Information evaporates.

Which brings me to the version of all this we are now living through. I use AI tools in my own work most days, for research and for talking an idea through, and I have written before that the knowledge side of writing has gotten dramatically cheaper while the imaginative jump has not. But there is a cost hiding inside that convenience, and a recent study put part of it into numbers. I’ve pointed to this study before, but I think it belongs here too.

Researchers at MIT’s Media Lab had people write essays in three groups: one using ChatGPT, one using a search engine, and one using nothing but their own heads. “Self-reported ownership of essays was the lowest in the LLM group,” they reported, and those users also struggled to quote back the essays they had just produced. It just one study, so I’d hold it loosely. But it lines up with what my old teacher was getting at. If you don’t do the understanding yourself, you don’t keep it. It was never yours to keep.

None of this is an argument against tools, or against not knowing things. I’m a generalist who makes a living quoting people who know more than I do, and “I’m still figuring it out” is close to my resting state. The point is narrower. It is about being honest with yourself about which pile a given thing sits in. Do I actually understand this, or have I just been near the answer recently?

Confucius wasn’t asking his student to know more. He was asking him to know the edges of what he knew — to stop the small, constant pretending we all do, the nodding along, the half-remembered fact offered as if it were solid ground. That kind of honesty is harder than it sounds. It is also, as far as I can tell, where real learning actually starts.