Skip to content

Comment on New OpenAI models hallucinate more

Comments

Looks like latest chatGPT likes to hallucinate code segments that don't work or wrong API calls. Simpler instructions work fine, but at some point it starts to diverge and experiment.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.