AI will not become sentient and decide to kill us all. But our own conscious or unconscious beliefs about AI can potentially increase the likelihood of any outcome, including catastrophic ones.