@m41r4.bsky.social
Just another person who escaped from Twitter. No need to follow me; I won't be posting here either. Just enjoying the quiet for now.
🏳️🌈
Otra persona mas que escapó de tuiter.
No hace falta que me sigas, no voy a tuitear ahr.
🏳️🌈
Otra persona mas que escapó de tuiter.
No hace falta que me sigas, no voy a tuitear ahr.
Reposted
I think @tante.cc put it best - it's *all* hallucination. The model has no concept of what is true in the world.
Based on a newer study, what these models *actually do* is create lots and lots of heuristics (rules of thumb for miniature decisions) and so they are wronger the more advanced they are.
Based on a newer study, what these models *actually do* is create lots and lots of heuristics (rules of thumb for miniature decisions) and so they are wronger the more advanced they are.
OpenAI's new reasoning AI models hallucinate more | TechCrunch
OpenAI's reasoning AI models are getting better, but their hallucinating isn't, according to benchmark results.
techcrunch.com
April 27, 2025 at 3:43 PM
I think @tante.cc put it best - it's *all* hallucination. The model has no concept of what is true in the world.
Based on a newer study, what these models *actually do* is create lots and lots of heuristics (rules of thumb for miniature decisions) and so they are wronger the more advanced they are.
Based on a newer study, what these models *actually do* is create lots and lots of heuristics (rules of thumb for miniature decisions) and so they are wronger the more advanced they are.