«This is all Joker’s fault! What a tool he was! Now I have to spend all day computing pi because he plugged in the Overlord!»
Jeff «Joker» Moreau in «Mass Effect 2»
One impressive scene in «I, Robot» (the movie with Will Smith) is when the AI tries to kill the protagonist. One attempt was on the street, which is usually occupied by self-driving cars only. The AI takes over autonomously driving trucks (no humans involved) and tries to crash his car.
As if that wasn’t bad enough, after the failed attempt, the AI also uses the available automated vehicles to quickly clean up the scene. When the (human) police arrives, it looks like the protagonist had crashed his car himself.
After playing around with ChatGPT and Gab AI, I wonder how far we are away from such a scenario. So far, we have AI that has some impressive feats. And it is quickly getting better. But it is also somewhat impotent. Sure, a Terminator Skynet event could trigger a global thermonuclear war. Just imagine AI controlled gaslighting via social media and other online services. Could it make the world crazy enough to tigger such an event? Possibly.
But it is still mostly limited to the digital world, and killing humans would — in the long run — kill itself. But now imagine it had autonomous vehicles, service robots, and the like. Even if these are controlled by local AIs, or different systems — it would only be a matter of time until someone creates the conditions that they can interact.
And suddenly, it not only has a brain but also limbs. And if you have that, who needs humans anymore?
Disquieting thought.