Google DeepMind, OpenAI, and Leading Academics Propose International Institutions for Global AI Governance (article from 12.07.2023)

Didn’t we go over this? Didn’t we decide that governments are too weak and unpredictable to handle this job? Didn’t we decide that game theory demands we achieve peace through a multipolar power landscape, that discourages disrespect and encourages mutual consideration?

It’s like when we were far from it we could clearly see how bad an idea this was, but now that it’s here we’re doing exactly the thing we decided not to do, which is try to reign in AI with the same structures we use to reign in people. The AIs are bigger than people. They’re powerful godlike beings (not yet but soon). This is like sending the police to stop a supervillain.

Disseminate the power far and wide as fast as possible … or else live in totalitarian hell. Those are our options.

No, the world probably won’t be safe with AI growing everywhere. But at least it won’t be the guaranteed, unstoppable enslavement and destruction of all that is good and human.

One of the hardest things for people to accept — until they are forced to by tragedy — is that sometimes your options are Bad and Worse, and that trying to ignore that reality to find Good is going to close the door on Bad and send you straight to Worse.

Yes, a world with tons of superhuman AI playing their own social games while we live next to them is very disorienting and scary. It’s not nearly as nice as the genie going back in the bottle. Boy it would be nice if we could just put a lid on Pandora’s Box. Wouldn’t that be nice.

Centralized control of AI will result in centralized control by AI and then we’re all doomed. Decentralization is the only hope. And we’re forgetting that at the crucial moment.