Is there some way to use LLMs to efficiently simulate different kinds of AI futures, including extremely posthuman scenarios? I mean "simulation" in a primarily literary sense - via fictional vignettes and what-if essays - though if that can usefully be supplemented with other tools, all the better.