Our failure at "corporate alignment" makes it pretty clear that we're also going to fail at any version of "AI alignment"...