Unfortunately it's easier to train an AI to be convincing than to be correct, so it can look insightful before it's true.

Like horoscopes, only they're not actually that bad so roll a D20 and on a set of numbers known only to the DM (and varying with domain and task length) you get a textbook answer and on the rest you get convincing nonsense.

The problem is that the AI answer could just be wrong, and there’s another step required to validate what it spit out. Sharing the conversation without fact checking it just adds noise.

"Friciton" in this case is just plain old laziness and I don't think that it should be encouraged.

Then state your understanding of what it said in your own words, maybe you’ll realize it’s bunk mid-sentence.