They already can't. They just rely on the assumption that most of the data they collect is correct. Which is generally true, there is more correct than incorrect content on the internet. The inability of the bots to discern incorrect data coupled with their ability to make it sound authoritative is what makes them dangerous.