Sometimes even I, as an ardent Luddite, wish AI companies would be slower in driving their own failure. Recently I've been working on a paper relating to faux-social LLMs - those which mimic relationships, therapy, care etc. In part of the paper I suggest viewing those 'relationships' built with LLMs as toxic or insecure in part because whatever a user can experience from them (which isn't a lot of things) that intimacy is always temporary. Behind it stands a tech company which, for whatever reasons, can issue an update or change the ToS in a way that completely breaks the nascent (if one sided) connection. That was only one point among many and a very brief summary of it at that but apparently Character.AI have gone all out on proving me right.Their latest update - PipSqueak 2 - is proving massively unpopular as users lament the passing of the bots they knew and, in some cases, professed to love. Along with ads, age verification and usage limits it's part of a raft of changes which…
No comments yet. Log in to reply on the Fediverse. Comments will appear here.