@Duretum@NASASpox@NASA I really like the way you put that. Kinda similar to what happens in the series "For all Humanity", it finally feels like the space race is back on.
Language models naturally generate outputs towards the "center" of the data distribution they are trained on, neglecting rare/tail perspectives.
This paper argues that while useful, overreliance on recursive AI systems could lead to a "knowledge collapse" phenomenon where public knowledge gets narrower over time.
I like the recommendations, and I think this is also a strong argument for model multiplicity, encouraging creativity, and trusting users with more choice and customization.
AI should reflect the rich diversity and weirdness of human experience, not just weird corporate marketing/HR culture. https://t.co/jxCLCwrVM9