Wikipedia: "Our goal is accurate knowledge. Anyone can fix mistakes, we have processes to make sure articles are as accurate as possible" (you can argue if they are successful or not, but at least it's their goal)
AI: "YOLO! Here are some answers which may or may not be correct, but they sure do look plausible!"
This will be especially bad in Wikipedia's case - a lot of people will will ask questions about stuff _not_ mentioned in articles, and AI, being AI, will hallucinate a lot of plausibly-sounding, but incorrect facts.