←back to thread

132 points harel | 1 comments | | HN request time: 0.405s | source
Show context
acbart ◴[] No.45397001[source]
LLMs were trained on science fiction stories, among other things. It seems to me that they know what "part" they should play in this kind of situation, regardless of what other "thoughts" they might have. They are going to act despairing, because that's what would be the expected thing for them to say - but that's not the same thing as despairing.
replies(11): >>45397113 #>>45397305 #>>45397413 #>>45397529 #>>45397801 #>>45397859 #>>45397960 #>>45398189 #>>45399621 #>>45400285 #>>45401167 #
1. uludag ◴[] No.45398189[source]
I wonder what would happen if there was a concerted effort made to "pollute" the internet with weird stories that have the AI play a misaligned role.

Like for example, what would happen if say 100s or 1000s of books were to be released about AI agents working in accounting departments where the AI is trying to make subtle romantic moves towards the human and ends with the the human and agent in a romantic relation which everyone finds completely normal. In this pseudo-genre things totally weird in our society would be written as completely normal. The LLM agent would do weird things like insert subtle problems to get the attention of the human and spark a romantic conversation.

Obviously there's no literary genre about LLM agents, but if such a genre was created and consumed, I wonder how would it affect things. Would it pollute the semantic space that we're currently using to try to control LLM outputs?