AI tools hallucinate 100% of the time for me for the security related topics, in... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		betaby on May 31, 2024 \| parent \| context \| favorite \| on: Legal models hallucinate in 1 out of 6 (or more) b... AI tools hallucinate 100% of the time for me for the security related topics, in particular encryption.

kromem on May 31, 2024 [–]

Which models and how are you prompting/setting the context?

In some cases hallucinations are unavoidable, but 100% sounds like it might be a usage issue.

betaby on June 1, 2024 | [–]

Yes, I'm holding it wrong.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact