A poll built on leading questions, false choices, and fearmongering does not reflect actual public opinion on AI. It shows how to optimize disinformation for partisan messaging.