Language models cannot reliably distinguish belief from knowledge and fact

Science TLDR by Raymond Ruff