The Things You Used to Know by Heart

What leaning on AI to remember for you quietly does to your own thinking.

The Quiet Part is a short series on what living alongside AI quietly does to you. The others in the series are You Reach Before You Think, on judgement, and It Will Never Think Less of You, on being known.

You used to know your closest friend's phone number without reaching for it. You read a good book last month, cover to cover, and if someone asked you tonight what it actually argued, you would reach for the answer and come back empty.

The information has not gone anywhere. It has left your ready recall while staying well within your reach. The number is in your phone; the argument of the book is one question to a machine away. That is what makes the loss so hard to name: nothing is missing, and something is.

Because a thing you know and a thing you can find do different work.

Known, and merely findable

Ask the machine what the book argued and it hands you a whole, coherent account of the argument in a second. Search used to return documents you still had to think your own way through; the machine can hand back the thinking, already done. It is a strange kind of having, because you never did the building and you will not keep the result. You can call it up again tomorrow, always fluent, always gone the moment you close the tab: an understanding you can borrow forever and never once own.

What you actually know does the one thing the borrowed version cannot. It is already present when you meet the world, deciding what you notice, what strikes you as strange, what question even occurs to you. A thing you can look up waits, filed and patient, until you already know to go and ask for it. A thing you know is in the room before you reach for anything, quietly shaping the thought while it forms.

Stored, it waits to be fetched. Carried, it is already in the room.

The thing you could not have looked up

Say you know that your grandmother salted her cooking like she had a grudge against it. And say you know, too, that she grew up with nothing. For a long time those two facts never touch. Then one evening, as you stand over your own pot, they meet, and a question forms that you would never have typed: was the salt about the years of not having enough?

You may never know if you are right. Either way, you would never have gone looking for the question, because you did not know there was one to ask. The machine can connect anything brilliantly once you bring it the two things and the hunch that they belong together. What it cannot be is the place where the hunch first arrives, unbidden, because you happened to be carrying both halves at once.

A machine can answer anything you bring it, and connect whatever you hand it. What it cannot do is be already in you, shaping what you notice before you know there is a question to ask.

Carried, two facts can meet before you think to introduce them.

What stays with you

The machine is welcome to the phone numbers. Most of what you look up was never going to shape a thought, and it is right where it belongs. But somewhere in the pile is the other kind, the shape of a problem you are stuck on, or the way something you love actually works, or the argument of the book that changed your mind. Their whole use is to be in you already, uninvited, when something new walks past, and a thing you can call up is never quite that, however close to hand you keep it.

When every answer is free, the scarce thing is what you have taken in so deeply that the world can still reach in and remind you of it.

If an understanding can be borrowed this easily, the next question is what actually proves you can think.

I'm Harry, and this is The Durability Curve, where I work out where value goes as AI gets cheaper. The aim each time is to leave you with something you can take to your own work: a test to run, or a way of seeing it you did not have before. If you'd like the next one to find you, subscribe.

This essay is also on the site: durabilitycurve.com/blog/the-things-you-used-to-know-by-heart.