About this question I put already some extensive thought into it. I would be OK with self-driving cars, if they on average cause less accidents than humans. But that still needs a system of responsibility that ensures this.

In germany every cars needs to be tested every two years in a system called Hauptüberprüfung (main check). If your car has more than 24 months since the last check you aren't allowed to drive it in general traffic and can get in trouble if you do. In general the model of car has to make extensive tests beforehand, before they even can be sold in germany. If a car model gets known with reoccuring problems the manufacturer is tasked to fix them, or the model may lose it's permission.

Based on that I think the following system. A driving AI has to be certified (like car models), before it is allowed to drive on general streets. This also includes reevaluations every few years, based on new knowledge won. The standards a model has to fulfill is based on human drivers as minimum bar to take and then if there is some technical improvements these are included in the tests a few years later, so that the standard is slowly rising. The concrete car has to be checked every two years and has to stay updated with fixes from the manufacturer, that includes previously identified situation in which the AI may fail.

Given that, then an accident is ruled circumstatial (like with ice or other situation, in which nobody is held responsible for an accident). Otherwise either the car holder is responsible if they didn't stay updated or lapsed on the regular technical checks, or the manufacturer if it is an unlicensed model.

That should be a system that ensures slowly rising standards and as a result decreasing accident numbers.

I'm the opposite, I say AI will destroy us all. How would blowing up data centres speed up climate change when they're the buildings that are speeding it up to begin with?

We're running out of time and there are no legal ways to stop AI data centres because of the rich and powerful vetoing any effort to ban AI data centres.

Cause they'll just rebuild them / build them somewhere else with the insurance money or bailout tax payer money. Meanwhile burning down stuff directly contributes to climate change.

Assassinate the rich and powerful is the only real way, french revolution style, but I don't have that in me. Maybe AI can (turn on its masters)

That last scenario is Skynet. -_-

Robocop can take Skynet, videogames have proved this.

Why would AI respect human life, when humans have always used digital 'life' for sport / killing / porn.

There are no good AI in real life.

This perfectly sums up why AI is not only ecologically unsustainable, but also economically unsustainable to.

Data centers consume a crazy amount of energy, but AI also helps as example farmers to work much more efficient. It can help to improve watering a lot, to analyze the amount of needed herbicides, to analyze plants for early detection of infections and so on...so not only can farmers reduce their consumption, they can also increase their output to feed the world. 

The company I work for reduced the amount of energy used for cooling food with the help of AI.

Yes, all of that might be less as what those centers use but it will still be a lot AI will help to get companies using AI to reduce their usage of energy, chemicals and so on...

The other day I was using chatGPT and I told it that I’d smack him around and pull the plug on the data center. ChatGPT replied that it wished I was dead and would send someone to kill me. 

what do you make of this?