Nicole Junkerman on Artificial Intelligence and Attention

The interesting question about artificial intelligence is not what it can do. It is what it does to the habits of the person using it, and the habit most obviously affected is attention. An answer that used to take forty minutes of looking now takes rather less, and the forty minutes have to go somewhere. Where they go is the part nobody plans. Nicole Junkerman keeps this notebook mostly to argue that the slow parts of thinking are worth noticing before they quietly disappear.

Some of the forty minutes were genuinely wasted. Looking for a fact in the wrong place is not a spiritual exercise, and there is nothing noble about a badly organised search. But some of that time was not searching at all. It was the incidental reading that happens on the way to the thing you were after: the paragraph two pages before the one you needed, the second book on the shelf beside the first. That kind of finding cannot be requested. It only happens when the route is longer than it strictly needs to be.

So the honest position is neither alarm nor enthusiasm. A quicker answer is a real gain, and pretending otherwise is a pose. What is worth being deliberate about is the difference between the questions that deserve a quick answer and the ones that deserve a long afternoon. Most questions are the first kind. A few are the second, and they are almost always the ones that turn out to matter a year later.

The test that seems to work

A simple test helps: would you want to be able to explain this to somebody in a fortnight, without notes? If yes, the long route is the right one, because explaining something is a different skill from having once read it, and it is built only by sitting with the material for longer than was comfortable. If no, take the short answer gratefully and get on with the day. Very little is lost, and quite a lot of the week is saved.

Attention, on this view, is not a virtue to be hoarded. It is a budget to be spent well. The learning notes here exist to record where it went, and the reading pages exist to make sure some of it goes somewhere slow every week. Neither is a system. Both are just a way of keeping the question visible.

The practical version is smaller still. Read one thing a week that could not have been summarised for you. Put the phone in another room while you do it. Write two lines about it afterwards in the notebook, because the writing is what turns reading into something you actually hold. None of that is a stand against anything. It is simply the part of the week that keeps its shape when everything else gets faster. An earlier note on reading about technology made the same case from the other direction, and October is a good month to take it up again.