Comment on Mom horrified by Character.AI chatbots posing as son who died by suicide - Ars Technica
Megaman_EXE@beehaw.org 1 week agoI’ve used one called PI which I’m assuming is some kind of branch off of chat gpt or something.
You don’t have to sign up or anything (for now) which is cool. But I assume they harvest all our data and information.
I tested to see if I could break it once, and from my brief tests, it seemed to never break out of character or tell me something bad or negative, which I thought was interesting(and good!)
Powderhorn@beehaw.org 1 week ago
I actually used Pi as my intro to generative LLMs. It was … I guess not encouraging self harm, but so fucking irritating that it led me to want to. Always with the irrelevant supportive words that I guess work if you’re a teen?
Megaman_EXE@beehaw.org 1 week ago
Lol yes, that was going to be the one downside I was going to mention. I wasn’t sure if it was just unique to my situation, but I found it would lead me down a logical path. It would ask me if I had tried various solutions.
Eventually, I would hit a point where it wouldn’t know where to go any further, and it would land on “here’s some things you can do” but those options would be things I was actively trying and failing with.
So that was fun. In a way, it was great at confirming that I had thought of all the logical options.