They are usually nonsense and should be banned. The author should receive a warning and repeated offender should be banned from posting.
The rule is basically re-emphasising ban low quality content spam. We ban them because they are low quality, not only because they are AI generated. If someone can actually use AI to generate high quality content, then it is fine. But the bar will be high.
What distinguishes low quality AI generated content from low quality content?
I think we should generally ban any kind of spam, nonsense, off-topic, etc content.
The only difference will be AI generated one can be harder to distinguish compare to a super confused user post. So a warning first. Obvious spam doesn’t need a warning.
It’ll become repetitive sometimes, likely when quoting a human who writes succinctly. Our current example avoided this pitfall, likely because they’re quoting junk.
We should permit automatic translations which preserve semantics reasonably of course.
My one caveat is that I do know some non-Anglophone community members who use LLM tools to polish their writing, which I think is a valid use-case, but I’m not sure is a completely distinguishable one.
I think the core difference here is that AI is not generating the ideas in the post, but helping sculpt the words. This overall seems okay.
It seems in the recent example found on this forum, the ideas themselves were being sculpted by AI, which is of little value. This is not okay.
Likely this is just a rule that needs to be enforced with “I know when I see it”.
At least some of these come from, or claim to come from, translations from other langauge. In practice these LLM are always worse than google translate, so…
Does this forum software have multi-langauge features? Ideally special langauge tags for posts plus users selecting their langauge filters. Should we expressly support multiple lanauge here?
We’ve many Spanish, Russian, German, French, and Italian speakers at Parity and W3F, so really permitting those sounds straightforward. Chinese is harder, but useful.
If a super user comes along and assaults the forum with reasonable and well written posts containing nuance, is that a bad thing? Or is that awesome and heats up “real user” engagement and discussion? You could make a compelling argument that until it is impacting the real developer discussion in a negative manner -it would be beneficial to the ecosystem.
As Shawn pointed out - if there’s something with language models being done that doesn’t align with our community, there’s a great rule just for that…