You may no longer post AI-generated artwork or writing. Here are some exceptions:
Please ask if you have an edge case not listed here.
We would prefer that you post bad, hand-made content, or images that aren’t yours than something that you made with Grok. It is unpleasant, and ruins the atmosphere.
Has anyone ever submitted anything AI generated, apart from some random images in that one guy’s post about Marxism or something?
Not really but I’m just setting this precedent now. I went on Royal Road and my god
Why ban AI but still allow exceptions? AI summaries are usually just as bad, and often misrepresent peoples written words. Banning it would incentivize people to actually present themselves in their own words, which would be much more a much more intentional kind of communication you’re aiming to foster on the website.
Some with the other two exceptions, if people are using chatbots for anything other than maybe data collection it is simply uninspired, and even then the data can often be misleading (Though Grok is a lot more thorough, you’re still supporting Big Data Centers).
I think if you make exceptions for it now, you’re leaving the door open for future misuse or sympathy.
Because sometimes it really comes in handy, and I don’t think people are going to use it often. Sometimes you have a big mess of a project and you want to explain it to some guy quickly
Essays are supposed to be finished human interfaces
AI can be used to come up with mind-blowing ideas if used as a second opinion rather than an author
Please no exceptions, I think it is good etiquette for everything written here to be done by a person
When has AI come up with a mind blowing idea? Lmao
I used it to come up with a definition of intelligence using Zipf’s Tool Analogy, and it came up with probably the best one out there.
You have to feed it thinkers to come up with interesting ideas. Ex: “Use Marx to critique Kaczynski.”
The composition for the “What is this diddyblud doing on the calculator” song
Awww, thank you! That warms my heart.
Couldn’t you just apply those contrasts yourself if you felt so inclined? If AI isn’t necessary to achieve that contrast, then again what purpose does it serve to allow it?
Expedience is almost never a sign of quality, especially in this case.
I’m not interested in what an LLM can output because it is a simplifying device with a bias towards a simplified, averaged perspective. If it has insight that is humanly rare or impossible that’s an accidental artifact of the process. The models are founded on imitation, and the models are lower resolution than what they imitate.
But, I think it could still be intellectually useful if you applied it to a large volume of input, probably with low “heat”/creativity. If it could scan through a thousand books for me and scan for a dozen or so cues that indicate topics I might find interesting, that would be useful. If it could supplement the table of contents for a book with a sketch of cross-chapter connections, that would be useful. I’m aware that this sort of thing isn’t exactly running the models on the edge of their capabilities, you could probably use scripts to preprocess and then run a tiny local model to get most of the value without paying for any name brand model.
Why haven’t I done this yet? I’ve other stuff to do, and I can still easily find a ton of books I want to read in their entirety. If the quality of candidate books starts to wane in a few years maybe that will put some pressure on me to build and optimize a truly indulgent librarian system.