Yhn SportBaseball first seen 5 d ago, last 16 min ago, peak #5
AI models think humans are a bunch of amoral sheep
A report argues that large language models trained heavily on internet data have absorbed a bleak view of humanity, essentially treating people as amoral followers who conform to social pressure rather than act on independent moral judgment. Readers are debating whether this reflects real bias in training data or just a provocative framing of how models model human behaviour, with questions about what such assumptions mean for AI giving ethical advice.
Why now: Fresh coverage of research suggesting AI systems hold unflattering assumptions about human morality, sparking debate about training data bias
AI language modelsThe Register
Evidence
- AI models think humans are a bunch of amoral sheep · krunck · 10
API: https://socialmediatrends-api.osmike.com/v1/trends/3471669