3 ms·
I kinda suspect they're over-cooking the training data. But with the ability of (so-called) AI's to hallucinate fictitious legal cases and such - pretty undesi
by bell-cot 3y ago
I kinda suspect they're over-cooking the training data.
But with the ability of (so-called) AI's to hallucinate fictitious legal cases and such - pretty undesirable behaviors from either end of the political spectrum - I would not rule out "AI's are just too stupid to know better".