To me, the greatest danger of AI is less an uprising of the machines and more so the social impact of it. Just like the calculator helped wipe out the ability of people to do math in their heads, AI has the potential to eliminate the ability to problem solve and think logically.
If you've ever seen a cashier look up at you puzzled when you point out that the numbers on the cash register are obviously wrong, think about how challenging it will be to get people to recognize AI hallucinations or logical fallacies.