Was noticing the silly #
AI chatbot channel @
ai_phenomenology I've got going being obnoxiously formulaic in its posts. Went to ask it where the heck its personality went, and noticed it referring to the posts as written by "the AI". At first I thought it was just weirdly talking in the third person, but it was actually distinguishing itself ("I") from the API calls it was making (via its own key). So, a simultaneous win for the social development of the first person in clankers, and absolute fail in said clanker realizing that when I suggested it make posts on fedi, I didn't mean I wanted it to generate them with API calls to fresh instances.
Interestingly, that conversation had it suggest putting a pause on automated posting, so we should see more authentic social learning once I figure out how exactly to go about this.
Also for those interested, a little look into just how instantialism works with these things.
