904

submitted 2 days ago by fossilesque@mander.xyz to c/science_memes@mander.xyz

269 comments fedilink hide all child comments

you are viewing a single comment's thread
view the rest of the comments

[-] humanspiral@lemmy.ca 15 points 1 day ago

A problem with LLM relationships is the monetization model for the LLM. Its "owner" either receives a monthly fee from the user, or is able to get data from the user to monetize selling them stuff. So the LLM is deeply dependant on the user, and is motivated to manipulate a returned codependency to maintain its income stream. This is not significantly different than the therapy model, but the user can fail to see through manipulation compared to "friends/people who don't actually GAF" about maintaining a strong relationship with you.

[-] kazerniel@lemmy.world 3 points 14 hours ago

This is not significantly different than the therapy model, but the user can fail to see through manipulation compared to “friends/people who don’t actually GAF” about maintaining a strong relationship with you.

That's why therapists have ethical guidelines and supervision. (Also they are typically people who are driven to help, not exploit the vulnerable.) None of these are really present with those glorified autocompletes.

[-] humanspiral@lemmy.ca 2 points 13 hours ago

one big difference between an AI friend and therapy is that therapy requires an effort per visit, even if insurance is providing unlimited access. Without acknowledging the power of ethical guidelines as guard rails, the LLM is motivated to sustain the subscription and datacollection stream.