Over the years, I have written a lot about AI. I am actually not negative on the technology, it is more that I am negative about the intentions of the people developing it and, the unintended negative consequences.
This goes back a bit, but I found an early one where I touched on this a little.
https://peakd.com/philosophy/@tarazkp/an-artificially-described-perfect-world
Also, I am not sure if this search link in peakd will work for you, but give it a go :)
https://peakd.com/search?q=artificial%20intelligence%20&sort=rewards&since=all&author=tarazkp
The search link worked :) The early one is from seven years ago! And I wasn't thinking of letting AI run everything and evolve on it's own. I was thinking more in terms of execution efficiency and replacing most of the government workers, but not letting it run the world from the ultimate decision making standpoint. The scenario that you described a GAI might evaluate us as a species and decide that we are a parasitic entity and eliminate or enslave us Matrix style :)
Yep, I know. It was just one of the early ones I could find - the searches aren't very good here.
However, I think that AIs are already kind of running things, as a lot of what is done is data driven. This doesn't mean they are fed good data though. I have an article somewhere describing an AI that sits behind the president, and the human just says whatever the AI tells it to. That way, people would accept it :)
When it comes to knowledge work automation for repetitive tasks, the replacement of humans is already happening.