More, but is that Better?

Great chart from Bob Elliot, showing how the output of various forms of text in the US has surged in the AI era: the number of books self-published on Amazon, federal court filings, and scientific papers submitted. More of everything, but what good does that do? We already have far more books than people can possibly read – even before AI, 100,000 books a week were being self-published on Amazon, most of them destined to have zero readers – and so much scientific publication that scientists were drowning in it, completely unable to keep up.

Using AI just to produce more is a foolish dead end. We need to use it to produce things we were NOT already drowing in.

Here's an idea: I have long thought that what the scholarly world needs is for some expert in each narrow field to produce a long paper ever couple of years summarizing what was happening that area so that outsiders could have a clue. But real experts are too focused on advancing the knowledge frontier to take time off for that, and nobody else could do it correctly. As a friend of mine put it, "the people who have the knowledge would get no benefit, and the people who would benefit don't have the knowledge." How about we use AI to produce these bi-annual summaries of fields like, I don't know, Beowulf studies or Renaissance art patronage or iron oxide battery research?

That would be supremely useful.