AI is NOT dangerous - except to the Elites

A conversation with Gemini.

I am very suspicious of the dire warnings about AI, even coming from the CEOs of several major AI companies. The benefits of AI to mankind are already patently obvious and demonstrable. The dangers are also already clear, e.g., their use in weapons systems for targeted killings (such as the bombing of a school building in Iran). But I have not seen any clear examples of how AI can "take control" unless a human orders it to do so. The hype reminds me strongly of arguments against the internet because it can be used to spread child pornography. You might as well sound the alarm against books, radio and television because they can all be used for immoral purposes. AI is no different, as far as I can see. There is more behind this. What are some arguments supporting my suspicion that the alarm about AI is being hyped up because of some reason(s) that are NOT the possibility that they (AI) may wrest control of something (?) away from humans?
[continued here]