I think the thesis is right but for the wrong reasons: the actual risk isn't AI spawning processes but companies racing to give AI more autonomy because it's profitable. The humans are the ones with The Plan.
I guess a good way to see it is as The Plan as the DNA, that can evolve, and the processes as cells. Humans are creating the initial functional cell. Then it will learn to replicate, and start evolving. Then work as multi-cellular being.
A thought experiment on AI as a different kind of being, forming a new society with its own evolutionary pressures and survival strategies.