So, how does artificial intelligence make such a mistake? Presumably, the "intelligence" is not pre-programmed knowledge, but rather knowlede gained by "reading" large amounts of information and creating logical links. Typically, AI is "trained" with inputs that are of the type to be analyzed, and then additional data is put in to see what the trained AI would see in the way of similar relationships.
But, how do you train AI to think like a human who reads the info provided on the Internet? There is so much misinformation on the Internet that it misleads humans. How is AI going to avoid being mislead? With 100% accuracy? I would not be surprised if there were intentional efforts to mislead AI programs by those who dislike them. Imagine what 2 separate AI programs run by the two major political parties would disagree on! Humans are already showing that by just listening to the things in the "echo chambers" that already match their beliefs.