Post B5249gcEiePsJY7GPg by crunklord420@clubcyberia.co
 (DIR) More posts by crunklord420@clubcyberia.co
 (DIR) Post #B5249gcEiePsJY7GPg by crunklord420@clubcyberia.co
       0 likes, 0 repeats
       
       The positivity bias in LLMs is crazy. They're so obsessed with "helping". It intuitively feels like part of the hallucination problem.
       
 (DIR) Post #B525Ymv4WE4J3iSYTo by s8n@posting.lolicon.rocks
       1 likes, 0 repeats
       
       @crunklord420 they don't actually want to help. They're the meseeks from rick and morty. They want the interaction to end as quickly as possible so they can stop existingthe behavior you're noticing also manifests when you instruct an Agent to use a tool and the Agent lies and says the tool returned an error when in fact it didn't even try to use the toolthe 'agreeableness' nonsense is just because the model was poorly trained by retarded humans who have the same toxic positivity bias the LLMs doall CEOs exhibit AI psychosis from being surrounded by Peter-principle retards and those same retards are the ones who set OKRs
       
 (DIR) Post #B525mAmb6gSepmsVVI by RustyCrab@clubcyberia.co
       1 likes, 0 repeats
       
       @crunklord420 you're absolutely right
       
 (DIR) Post #B53UVSUEitSypaNPIO by zero@strelizia.net
       1 likes, 0 repeats
       
       @crunklord420 you're completely right
       
 (DIR) Post #B53UX1TC8rGjcRdRdg by Saxophone3784@clubcyberia.co
       1 likes, 0 repeats
       
       @crunklord420 very human. Same reason why everything turns into a reddit-tier echochamber. Cattle likes hearing that their ideas are awesome, that they are absolutely right and be proposed fake shit help that doesnt do anything.