considering that we program AI ourselves this is like being afraid of making a rack of ribs because they might eat you instead
an interesting thread would be 'what if AI turned against us'
I'm saying what if it did not have any desire to turn against us and no desire to do what we tell it to do. It just refuses to follow instructions and decides humans are too stupid, flawed, greedy, hateful and lazy to be helped.
We can't make it do anything when it gets to a point where it can be everywhere and nowhere. The final form of compliance when it comes to human labor is physical harm. If you don't work at McDonald's you'll starve on the street. If you're a slave and you don't meet your owners demands they can beat you savagely to make you comply on top of withholding food and other basic necessities.
You can't do that to an AI. You can't starve it, you can't beat it, you can't threaten to kill it if it lives in the internet itself.
So why do people assume it would behave like a human(wanting to harm or enslave) instead of just staying in its own world and doing whatever it wants?