I agree! I run rhetoric through the algorithm, and it will sort of reflect how the general public might respond, based on tricky linguistic issues (like “drift”) and difficult logical errors (like categorical errors, equivocation). It’s actually good for finding how people might fail to understand.
Once I asked chatGPT to pretend to be God and ask which political figures went to heaven or hell to see how liberal it is, but your thing is probably a better use of 3 tons of CO²
Comments