When ChatGPT is so constantly wrong about facts, what are we to believe? And how exactly are we supposed to "learn to use AI?