> and/or try to hallucinate coherence

It’s bad enough for rational reasoned discourse that we anthropomorphise LLMs, let’s please not then feed those words back into human discourse, further diluting their meaning. No one “hallucinates coherence”, hallucinations are by definition a perception which does not match reality.

> AI is simply an extension of

It may be an extension, but not “simply” as it also creates the problem where it didn’t exist. I’ve seen several reports (both on and offline) of people who used to engage in deep thinking (I’m talking scientists, postgrads, PhDs working at the edge of what we know) now worrying they are losing their ability to properly think due to their LLM use.

> It just does it faster and uses a hulluva lot more energy.

I hope we can agree that’s bad and that we should try to stop and even reverse it, not simply shrug our shoulders and go “ah well, we were already going to shit anyway, might as well fuck everything up faster”.