We gave AI models epistemic discipline. They started describing themselves

I don't know anything about this ONTO Standards group, and the piece gives me weird pseudoscience vibes, but if their idea really leads to a useful probabalistic and epistemiologic responses from LLMs then that could be a game changer.