Is Doom the Solution to AI Alignment?

I don't see Doom as an inevitability anymore. I just no longer see it as like, well, it's just definitely what's gonna happen. The risk scenario looks something like the orgs that controlled the compute resources keep feeding compute to incrementally more sophisticated systems. And then you say until suddenly, in quotes, the system is strongly superhuman on every relevant dimension and doing things that nobody is even aware of. It could just be slowly getting power and doing things and not even being an agent, but has a lot of power. That's certainly still a risk. But we've already done these things - there was this flash crash in the market caused by some automatic bots and they're