Firm Reading in an Era of AI Delirium

Happy Holidays, meatballs.

This piece from John Gallagher continues on our current “slop” theme. Read the whole thing.

“Here is the main point I want to make in this essay. We, as human readers until 2022, generally understand mechanics to be a strong indicator of conceptual sharpness. If periods, commas, semi-colons (hopefully), and other punctuation are all in the “right” spots, there was a general sense of trust that the ideas had a level of “hard work” put into them. To use a machine learning term, mechanics largely functioned as ground truth for a reader’s ability to understand the writer. Mechanics largely told me, as a reader, that there was something there in the writing. Maybe I didn’t agree with the writing. Mechanics didn’t inoculate the writing against criticism. But it provided a baseline for understanding something was there.”

“Now, though, we have something that was simply not possible a few years ago: mechanically perfect prose with complex layers of imprecise ideas. Beautifully written cues of mechanical correctness can now hide malformed ideas. If we expect fully formed ideas when encountering mechanically correct prose, there is a subtle expectation that there are, indeed, fully formed ideas already present in the text. It’s like eating a beautiful mass produced store-bought cookie: pleasing aesthetics hide cheap ingredients. Reading all of this synthetic, AI-driven text could actually be bad for us, cognitively and physically.”