a shitpost feature

  • 🔎 If you see an alert for analytics.usips.org, this is a self-hosted system I am running as a test. It is GDPR compliant and collects no invasive information.
  • Want to keep track of this thread?
    Accounts can bookmark posts, watch threads for updates, and jump back to where you stopped reading.
    Create account
Let us invest kiwi coin into threads and have it give a return based on the change in thread traffic;
give each user 200 kiwi coin per month (more if you're a forum supporter)
 
Use AI to fight AI slop. Have the bot scan for retards posting "here's what Grok says" or "I asked Claude" or "Gemini..." (sorry, noone uses Gemini so I couldn't come up with an example for that one). Then threadban them, and/or reply with an AI-generated image of their avatar slurping on clanker cock.
 
Add a "slurs per week" leaderboard, tracking how often each one is said. But don't just count up the usual ones, have the AI scan every new post and determine when new slurs are being created. It should use context clues to determine which terms are cultural, historical, general slang, and which ones have a true slur connotation. Surface those with reference links to the thread they started in.

If you have capacity left over, have it identify 5 finalists for "best use of <slur> in a post". Submit it to you or the janny council every week, to pick a winner.

ETA: the leaderboard should be for the slur itself, not the person who uses it the most. "Faggot was posted 178 times this week", etc. Ideally your AI should detect and filter out obvious spam posts from the count, to discourage spamming.
 
Last edited:
Run that ethnicity detector on a random 5% of the forum users each month and then assign them a melanation score beside their user name. So after 20 months posting hands won't be necessary.
 
If you can, I would suggest a dashboard which has statistics about various social medias (top reddit posts, moderators, topics, same for twitter and others, points of interest/trends on twitter or 4chan, youtube, bluesky etc), summaries of happenings in each subforum on a day to day basis (maybe events as well), statistics on the user ethnicity location and shit like that as someone above mentioned (although this can be broken cause most people use a vpn) and probably one or two extra features, like a game of the day which is like a political axes test for subforum alignment or a pinball style small minigame, an argument bot which can be argued with on shit or just insulted (like a dummy), things like that, small things.
 
Best suggestions I can think of for API rather than harness basically boil down to:
  1. Jannybot features, the ethnicity scan is pretty funny.
  2. Incremental forum digest: doing operations during the highlight pass, most obvious example AI thread update summaries, showing a recap at the top with a concise and ideally linked directly to posts of interest. Prefiltering what you pass to the AI (e.g. only highlights, postmarks, maybe also posts with uploads) can squeeze out extra value for less token spend.
These both also scale up and down pretty well, jannybot can just refuse when necessary and summaries can be opt in per thread so you can get a good idea what the real world usage is like before rolling it out wider.
 
Null said:
I have access to $200 of free extra credits to do whatever I want with with AI. I I have no idea what to use it on.

these are little, programmatic agents that can be given big jobs or tons of little jobs.

they can basically do anything a person can do sitting at a computer, really fast. frame your understanding along those lines. it's a lot more flexible than the AI chatbots or programming terminals. it can do anything.

best idea for something funny or useful might get implemented
Click to expand...

Make a bot that crawls a given board and builds Markov chains from every post. Grab an open-source Markov bot (markovify works fine) and run it off that. BoardBot is born.

Have a second agent build a DB out of the same posts - let it decide whether that’s relational, graph, or Mongo-whatever. Dump the whole DB as context into a 1M-token model, or only the most recent 500K or so to leave it room to generate. Route it through Venice if you want it anonymized, though OpenRouter is probably fine. There are cheap fucking Chinese options with 1M context. Board.ai is born.

Then run them head to head. If a board’s BoardBot is indistinguishable from its Board.ai, the board gets closed for being retarded.

If neither bot can be told apart from the actual posters, close the site.
 
Castoreum said:
Halloween zombie forum game where users collect antidotes scattered around random forum pages to stay human while others become zombies (status indicated by username color or a frame around their pfp). Once you are a zombie you can't go back but you can bite human users and if they don't use an antidote they become a zombie. Biting has a cooldown time or zombies can only bite a user a certain amount of times (dunno how to make them not too powerful). Humans can't read zombie posts (zombie posts get translated to "Grrrr argh"). Game starts Oct 30th and ends Nov 1. Last humans to survive get bragging rights. Totally arbitrary. Just for fun game.
Click to expand...
It's already October 10th, make them Christmas zombies
 
Xenforo is the iconic GUI for the Kiwi farms. My advice would be to use an anthropic AI to generate a complete spec of its functionality, and Backend endpoints, and then use an Open-AI model agentically to crawl all of the site pages, and screencap them in both mobile and large screen format. Then categorize them and map them to the spec.
After that you should be able to ask it to build the Backend based on the spec, and then assign an agent running a smaller model to build each page.
 
Text classify me and every other user and maintain an updated list of who our friends and enemies on this forum should be based on our personalities, values, writing styles etc.

Use AI to generate amusing profession names and classification criteria, then feed every post to a jev decision maker to decide what the profession probably is of the user who wrote it.

Or instead of professions, the decision maker could decide which lolcow probably wrote the post. Like my post is probably Alex Jones coded or something.
 
You must log in or register to reply here.
Back
Top Bottom