 | As the sub has grown (and as AI based tools have gotten better) with over 1M weekly visitors, we've seen a marked increase in slop, spam etc. This has been on the mod team's mind for a while + there have been many threads started by users on this topic garnering lots of upvotes/comments. We're thus happy to announce the first set of rule updates! We believe these simple changes will have a sizable impact. We will monitor how these changes help and appropriately plan future updates. Changes
See the attached slides for details. FAQ Q: How does this prevent LLM Bots that post slop/spam? A: For fresh bots, the minimum karma requirements will stop them. Unfortunately most of the bots that are getting through reddit wide defenses are from older reddit accounts with lots of karma. These wont be stopped and is a site wide problem with even bot bouncer being unable to detect them. Often times, humans (mods and users) on the sub struggle to detect LLM based bots. We are looking into options on how to better detect these programmatically. Q: This is an AI sub so why don't you allow AI to post or allow AI written posts? A: The sub is meant for human posters, commenters and readers, not AI. Regardless, posting LLM written content without disclosure is deceitful and betrays the implicit trust in the community. It will long term result in erosion of participation and goodwill. And generally, it merely falls into Rule 3 - Low effort. Prompting an LLM and simply copy-pasting its outputs does not require much effort. This is specifically different to thoughtful use of LLMs, validating/filtering/verifying outputs etc. [link] [comments] |