Suzgun et al. find that current large language models cannot reliably distinguish between belief, knowledge and fact, raising concerns for their use in healthcare, law and journalism, where such distinctions are critical.
- Mirac Suzgun
- Tayfun Gur
- James Zou
