All the AITA, Two Hot takes, ask reddit etc story based subreddits are something like 95% AI
They often take advantage of the “throwaway” angle, but they all share the same type of name+name+number like Asphalt-Chameleon4593 instead of the classic “throwaway + topic” etc type names genuine people used to make when posting real stories.
Since about 4 months the AI bots have been replying to comments and act more authentic.
There’s often edits to add context a normal person probably would have included, or add things like “we have 9 kids and breaking up is not an option, because we would get deported” or something like that.
Very obvious LLM stuff trying to figure out what people consider normal, or where social boundaries are for things like cheating, stealing, lying, etc.
Very obvious LLM stuff trying to figure out what people consider normal, or where social boundaries are for things like cheating, stealing, lying, etc.
Oh god, I hadn’t thought of that angle. We’re not just teaching bots to detect lies, we’re training them to lie successfully 😱
All the AITA, Two Hot takes, ask reddit etc story based subreddits are something like 95% AI
They often take advantage of the “throwaway” angle, but they all share the same type of name+name+number like Asphalt-Chameleon4593 instead of the classic “throwaway + topic” etc type names genuine people used to make when posting real stories.
Since about 4 months the AI bots have been replying to comments and act more authentic.
There’s often edits to add context a normal person probably would have included, or add things like “we have 9 kids and breaking up is not an option, because we would get deported” or something like that.
Very obvious LLM stuff trying to figure out what people consider normal, or where social boundaries are for things like cheating, stealing, lying, etc.
Oh god, I hadn’t thought of that angle. We’re not just teaching bots to detect lies, we’re training them to lie successfully 😱
Ok who let the sarcasm in ?
Creepy man Crazy