Post

The main reason I don't like algorithmic detection is that it implies only an "AI" would be able to tell. This grants extruded prose false equivalence to prose that came from a mind. We should be teaching people how to see that the prose did not come from a mind.

The problem with the folk taxonomy that has sprung up around this, counting em-dashes and overused words and suchlike, is that it treats "AI" as if it is a mind with a quirky writing style. We naturally assume that there is a mind behind fluent prose, so the absence of one isn't intuitive.

But it is often possible to tell if you know what to actually look for; the artefacts that are a result of the lack of intentionality in the writing, the absence of a world model. And yes, once you know, sometimes you can tell lack of concept apart from human misconception.

The naive stylometry that has led to false accusations and witch hunts has left people thinking it's never possible to just tell. Actually, because the concept of prose that did not come from a mind is so new and alien, people have just been looking for the wrong markers.

14/8/26, 12:29 am0 reposts0 likes

Entrenching this idea that only detection software could tell the difference just entrenches this misconception that "AI" is a kind of mind, an idea that is toxic to our shared sense of reality and is central in some way to just about every kind of contemporary "AI" harm.