gemma lynn on Nostr: cnn is running a fun story about facebook rapidly back-tracking from their llm bot ...
cnn is running a fun story about facebook rapidly back-tracking from their llm bot profiles. it's a fun story because a good half of it is a reproduction of the journalist's "conversation" with the bot about its origins, and it's always an entertaining read when bots get existential. good television, you know.
i really wish journalists would stop printing their conversations with llm bots.
it's not an interview. it's not even a conversation. it's us watching the journalist play the keyboard prediction game with an overgrown autocomplete engine. and that's a fun game to watch, don't get me wrong, but it's not journalism. it's not telling me anything about the world or anything about the interviewee. it's slop, cosplaying as journalism. right down to the "facebook could not be reached for comment" at the end.
anyway i know they have word counts to hit but i wish they'd stop, because it always looks and feels enough like an interview that it will be misunderstood as an interview, which only perpetuates the fantasy that llm bots can be interviewed. that they have something to say, that they understand, that they can do anything besides deliver the most likely next token.
we're all too human to be good at identifying a thing that walks like a person and quacks like a person as anything other than a person, so if journalists could avoid making it worse that'd be great, is all i'm asking.
#journalism #llm #ai #gpt
i really wish journalists would stop printing their conversations with llm bots.
it's not an interview. it's not even a conversation. it's us watching the journalist play the keyboard prediction game with an overgrown autocomplete engine. and that's a fun game to watch, don't get me wrong, but it's not journalism. it's not telling me anything about the world or anything about the interviewee. it's slop, cosplaying as journalism. right down to the "facebook could not be reached for comment" at the end.
anyway i know they have word counts to hit but i wish they'd stop, because it always looks and feels enough like an interview that it will be misunderstood as an interview, which only perpetuates the fantasy that llm bots can be interviewed. that they have something to say, that they understand, that they can do anything besides deliver the most likely next token.
we're all too human to be good at identifying a thing that walks like a person and quacks like a person as anything other than a person, so if journalists could avoid making it worse that'd be great, is all i'm asking.
#journalism #llm #ai #gpt