I know this isn't popular to say on here but I'm not universally opposed to LLMs and Stable Diffusion. I work in online safety, and these technologies have helped us develop better classifiers to catch all kinds of badness you don't want on your screen.

But they're neither intelligent nor alive.

Again (and realizing this should really be a blog) I'm sympathetic and consider myself part of Dunkleman's project of making democratic governments responsive and functional. But you can't hand wave the accountability stuff. The argument can't be a "unitary exec theory - but for good!"

Are they awesome?