It's worse because they're often more confident in the AI output than they ever were of the google results, and the results are not-infrequently-enough so bad that no human would have made that error. When they do doubt, they can ask, and the AI will often defend its dumb position -- especially when they explicitly ask it to counter the rebuttal they received.

Skepticism also seems to be reduced because we're armored against people telling us lies in their own self interest and against ours, while AI will make stuff up that benefits no one. (And even where it could benefit someone, people assume the AI isn't trying to benefit itself).