[happytoexplain]

This is a super common fallacy about how people "should" feel about AI reliability ("people are wrong too!").

People are wrong/lie in different ways from AI. We have highly developed personal heuristics about where to place human writing on a gradient from 0% to 100%, based on the source, the topic, and a hundred other variables we don't even realize we're ingesting. Even without a way to verify, we are comfortable with this state of affairs. AI lies in more randomly distributed, unpredictable, confident ways. Even giving the benefit of the doubt that these falsehoods are rarer than human falsehoods, it creates a constant background of cognitive stress (FUD) and a feeling of indignation.

Further, the "who is wrong more often" question is complicated because we ingest human-created and AI-created data in different contexts. But it seems both evident and intuitive that AI is wrong more often, as long as you accept "I don't know" as not being wrong. You can ask it anything, and it will much more rarely say "I don't know" than a human who also doesn't know would. For example, if you accidentally ask it a question that contains a false implication, it will more often than a human just assume your implication represents reality.

Also, nobody claimed 100%. It's a red flag to write in black-and-white like that.