AI is not inevitable, because technological progress in general is not inevitable. It is shapeable by economic incentives just like everything else. It can be ground into powder by resource starvation.

We've long known that certain forms of financial bounties levied upon scientists working at the frontier of sciences we want to freeze in place work effectively with a minimum of policing and international cooperation. If a powerful country is willing to be a jerk (heavens!) and allow these kinds of bounties to be turned in even on extranationals, you don't need the international cooperation. But you do get a way to potentially kickstart a new Nash equilibrium that keeps itself going as soon as other countries adopt the same bounty-based policy.

This mechanism has been floating around for at least a decade now. It's not news. Even the most inevitable seeming scientific developments can be effectively rerouted around using it. The question is whether you genuinely, earnestly believe what lies beyond the frontier is too dangerous to be let out, and in almost all cases the answer to that should be no.

I post this mostly because inevitabilist arguments will always retain their power so long as you can come up with a coherent profit motive for something to be pursued. You don't get far with good-feeling spiels that amount to plaintive cries in a tornado. You need actual object level proposals on how to make the inevitable evitable.