A new post from me.

Everyone knows that science is drowning in AI slop. Publishers and conferences can’t keep up with a massive increase in AI-generated submissions. Some journals are publishing articles that start with “Certainly, here is a possible introduction for your topic.”

But here's the thing that puzzles me: why do scientists, who presumably entered the field because they loved it, feel the need to produce so much garbage?

The answer has surprisingly little to do with AI. It has to do with a revolution that began in the 1960s, with an “information crisis” in science and a retired legal publisher’s simple idea. It ended up reshaping the entire incentive structure around a single, seductive metric: citations.

That system made science more risk-averse, more incremental, and much more fragile. AI is breaking it apart. And now we’ll have to figure out what comes next.

Please enjoy!