Mignon Fogarty on nostr

I've never been opposed to the word "hallucinating" for describing how AI makes mistakes ... until now. I just talked to someone who thought AI hallucinations would be obvious because it would be ob...

Mignon Fogarty @Mignon Fogarty 1780764410

The word "hallucination" isn't going away — it's a widely used industry term — but we need to explain it better for beginners:

"Hallucination" is just a fancy word for "confidently makes mistakes":

"Remember: AI hallucinates, and you need to confirm all facts" should be something like "Remember: AI confidently makes mistakes, and you need to confirm all facts" or "AI tells you things that are wrong in a way that sounds completely believable. Confirm all facts!"

James Gleick · 4w

nostr:nprofile1qy2hwumn8ghj7un9d3shjtnyd968gmewwp6kyqpqhvzjvzfmclu5l5zk65fxye5cgt7e0qvs6rcn60z72v9hy79mmpzqlfkkfs I don’t think we need to accept it just yet. The word is deceptive—intentionally so. What needs to be explained is this: chatbots and LLMs can't "hallucinate” because they have no ...

Merilee D. Karr, MD, MFA · 4w

nostr:nprofile1qy2hwumn8ghj7un9d3shjtnyd968gmewwp6kyqpqhvzjvzfmclu5l5zk65fxye5cgt7e0qvs6rcn60z72v9hy79mmpzqlfkkfs The industry named its own mistakes ‘hallucinations.’ Hallucinations is a forgiving term. ‘Delusions’ would be more accurate.

M. Grégoire · 4w

nostr:nprofile1qy2hwumn8ghj7un9d3shjtnyd968gmewwp6kyqpqhvzjvzfmclu5l5zk65fxye5cgt7e0qvs6rcn60z72v9hy79mmpzqlfkkfs Would "delusional" be more apt?

Alex Rosenberg · 4w

nostr:nprofile1qy2hwumn8ghj7un9d3shjtnyd968gmewwp6kyqpqhvzjvzfmclu5l5zk65fxye5cgt7e0qvs6rcn60z72v9hy79mmpzqlfkkfs It’s a fancy-sounding excuse for having precision issues with floating-point math, which they cover by adding randomness and calling it a feature (“temperature”).

Janet Grootebroeder · 4w

nostr:nprofile1qy2hwumn8ghj7un9d3shjtnyd968gmewwp6kyqpqhvzjvzfmclu5l5zk65fxye5cgt7e0qvs6rcn60z72v9hy79mmpzqlfkkfs In my opinion saying "confidently makes mistakes" is not a good way of describing it spewing nonsense, because it can make people believe the computer program can think or is conscious ...