The Older I Get, the More I Think Most Learning Advice Is Bullshit

There is a wonderful feeling you get after reading three books about something: you feel clever. This feeling is extremely dangerous. I know because I am highly susceptible to it.

I like learning things. Computer science, mathematics, history, philosophy, languages. Give me an interesting rabbit hole and I can happily disappear into it until I've forgotten why I opened the laptop in the first place.

For most of my life I regarded this as an entirely good habit. Then I noticed something unpleasant: you can consume an enormous quantity of information without becoming correspondingly more capable.

This seems unfair.

The books don't warn you. You imagine knowledge entering the brain rather like furniture entering a house. Once the sofa is upstairs, the sofa is upstairs. Unfortunately the human brain appears to employ termites.

A few years later you return to something you supposedly learned and discover that your mind has preserved approximately three nouns and a vague sensation of familiarity.

"Oh yes," you say. "Eigenvectors."

Then nothing happens.

Recognition Is the Great Scam

One of the most deceptive sensations in learning is recognition. You read a paragraph and think, I know this.

Do you?

Close the book and explain it. Things become interesting very quickly.

I have had this happen with mathematics more than once. I had technically learned the material. I recognized the notation. I remembered the names. But did I actually possess the intuition underneath it? Sometimes, embarrassingly, no.

Take complex numbers. You can learn that multiplying by i corresponds to a ninety-degree rotation. You can pass a test on it. You can manipulate the equations. But there is a difference between knowing the rule and having the geometry sitting comfortably inside your head.

One is stored information. The other is a mental model.

Those are not the same thing.

I Think Understanding Has Levels

There is "I've heard of it." Then "I can explain it." Then "I can use it." Then "I can use it somewhere I haven't seen before." And finally: "I've forgotten half the terminology and I can reconstruct the idea anyway."

That last one interests me the most.

Deep understanding seems to involve compression. You stop carrying around thousands of isolated answers. You carry a smaller number of models capable of generating answers.

This is probably why experts occasionally give annoyingly short explanations. They are not necessarily skipping the reasoning. The reasoning has been compressed.

AI Makes the Problem Much Stranger

We now have machines that can explain nearly anything on demand. This is astonishing. It is also possibly the greatest recognition-producing machine ever invented.

Ask question. Receive beautiful explanation. Nod thoughtfully. Know nothing.

I have caught myself doing this. The answer arrives so smoothly that the brain mistakes the clarity of the explanation for understanding in the reader. Those are very different achievements.

So I've started using AI in a more annoying way. Explain it. Now explain it visually. Now remove the terminology. Now give me an example. Now ask me what happens next. Now tell me why my answer is wrong. Now make me reconstruct the concept from scratch.

The useful part is not always receiving the answer. Sometimes the useful part is arranging things so that I am forced to produce one.

Which is irritating.

I had hoped technology might eliminate this. Apparently the brain still insists on doing some of the work itself.