Interesting – but isn't this starting with the assumption that there is a possible combination of words or actions that the AI could employ which would work on

Interesting – but isn't this starting with the assumption that there is a possible combination of words or actions that the AI could employ which would work on a human? What if no possible combination exists? I feel that the bigger and more immediate threat is the misuse of AI by human beings against other human beings. Governments are already abusing it.