Hacker News new | ask | show | jobs
by orbital-decay 1 hour ago
Every LLM is a classifier biased towards its own writing, but the bias is usually subtle and the naive way like this is not reliable.