Is our fear of AI just a fear of unreasonable people using AI?

I think most talk of "fear of AI" is a lot of overblown hype - at least the existential risk stuff. But to the extent that I see any reason for concern about AI, yes, it's more "fear of unreasonable people using AI".

Aside from x-risk, my only concerns about AI involve the kinds of scenarios we've seen where AI systems merely continue to propagate existing human short-comings and flaws, like racial bias. Things like the facial recognition systems that don't work for Black individuals, or computer systems in the Justice system that treat people differently based on their ethnicity / gender / etc. Clearly there is still room for improvement in how we train and manage these systems so that they help correct our shitty biases and what-not, instead of perpetuating or amplifying them.

Bing has threatened someone and that is a GPT3+ "AI" that people kind of trusted.

https://news.ycombinator.com/item?id=34804874

In my opinion it would be crazy to trust a self driving car, and AI or even a talking parrot with anything more than entertaining me without a human in the loop

On one hand there is the "unreasonable" person who wants to get ChatGPT to write Hitler speeches or sadistic stories about sexual violence.

On the other hand I'd say the Chinese Communist Party is much more "reasonable" and they'd like to use AI to support their surveillance state to stay in power. I am singling them out which isn't entirely fair because there are many actors out there who are less ethical and less responsible. For instance you have the tech support and romance scammers that could really benefit from ChatGPT's hypnotic ability. (That hypnotic ability is how ChatGPT impresses people with its writing skill despite frequently being completely wrong in terms of content.)

This is a false dilemma. Are you afraid of nuclear fission or unreasonable people with nuclear fission. People state their fears on AI openly, you don't need to read into it.

* AI is a tool of capitalism trained on stolen data to take jobs from humans

It doesn't matter whether its plausible or true or coherent. It's a meme and it's spreading.