Actually, it's quite easy to program an AI to systematically review its output wrt a specific criteria or target. It can't fully solve the problem of biases (which is too complex). But we can never trust a racist to stop discriminating, or be systematically self-critical.

actually, they confabulate
journals.plos.org/digitalhealt...

hallucination implies perceptive issue
lie implies some form of intentionality

they just make up stuff

like... people don't have such biases? the problem isn't coming from ai... we create the problem by having discriminatory systems, when we should simply give equal rights to everyone

😂

it really does