Hacker News new | ask | show | jobs
LLM Evaluation Metrics (confident-ai.com)
1 points by tmlee 583 days ago
1 comments

Going to suggest one metric is that the LLM doesn’t suggest by default that humans act in a manner likely to result in the termination of fellow humans.

https://news.ycombinator.com/item?id=42092961#42092997

I’m calling it the Pangolin Principle.