AI Policy
You shouldn’t read something no one bothered to write.
That’s it. I could write about the technical shortcomings of AI prose (including but not limited to the invariable narrative tone, continuous emotional register and urgency level, and inability to handle time ramping, which are all basic 2nd-level functions of telling a story with prose), but those are ever-changing. It is simply deficient, and it’s not *good* for a person to read large volumes of prose deliberately engineered to feel *just* engaging enough to keep them from stopping. Demand better.
What uses of AI are generally permissible (by this site’s standards):
* First-pass language translation
* Copyediting
* Formatting of rote blocks (e.g. legal notices).
In short, non-generative applications are considered morally acceptable. (Arguably rote blocks are generative, but the GenAI output is generally so bad they need complete rewriting anyway, so the ‘content’ produced is often little more than a formally structured skeleton for a human to completely rewrite.)
Examples of AI use in HEROIC NONSENSE:
The Heroic Nonsense web serial used AI in the following applications:
* Copyediting and basic error-catching
* Formatting fictional press releases (that then need to be completely rewritten).
* First-pass research¹, information about dates and locations
¹ Oh my god, it’s wrong so often, you have no idea. You can’t just trust it.