alt Hacker NewsYou are not wrong.
I wonder how it manages to find the middle ground in controversial topics. My own extreme views have become somewhat moderated thanks to reading AI-LLM opinions and reasonings on the topics.
You are not wrong.
I wonder how it manages to find the middle ground in controversial topics. My own extreme views have become somewhat moderated thanks to reading AI-LLM opinions and reasonings on the topics.
My understanding is generated words are determined first by the corpus (apparently the whole internet - eek!), then by training in certain areas, then influenced a little by the final context.
Would welcome correction but it does seem to inevitably lead to a sea of slop if GAI is trained on GAI output, particularly areas like this where there is little data and no training.