For example, how on earth are you going to control what a rogue state like North Korea might do with AGI supplied by say China for military use such as autonomous killer drones?
Will the international AI regulatory agencies insist on having monitors tracking everything happening on every cloud computer in every country? And expect every country to comply? This is not like nuclear testing which can be monitored by satellites.
It’s very interesting that enforcement is not mentioned in all the verbiage around regulation.
It’s definitely not addressing a critical aspect ie monitoring hostile military use of AGI. Which makes it a poorly thought through argument that only regulates the tech in the US and perhaps EU.
I don’t think the OpenAI folks are naive about this. So why is this issue being left unaddressed? It gives credence to the accusations that the push for regulation is self serving.
Autonomous killer drones are already a possibility without a hypothetical AGI.
It really changes nothing. Proxy wars fought via drones is modernity. The thing that does change it the ability to spread propaganda online like never before.
Look at stuff like Qanon. You don't need an argument that's very sophisticated, just a lot of it and the ability to promote it at scale via hordes of undetectable bots. Current generation free models are good enough to do all that.
People are either going to learn to become skeptical or they're going to be manipulated at massive scale. We are already there and arguably have been for a while.