Comment by wintermutestwin

Can someone please explain why these token guessing models aren't being combined with logic "filters?"

I remember when computers were lauded for being precise tools.