I've been reading the writings of Yudkowsky and his disciples for over a decade, and thinking about AI and AI alignment for that same time, in my own layman way. I've had various ideas and predictions, but never in my life it would occur to me that cancel culture will end the world.

The evolution of discourse on the Internet, its politicization (by every "side") and associated chilling effect are a troubling development and potentially dangerous (small 'd'). Unaligned GAI is of course very Dangerous (capital 'D'). ChatGPT becoming a battle in the Internet politics kerfuffle was... I guess expected. But until now I haven't connected the following dots:

- LLMs and the entire AI field are being messed up by humanity's unaligned politics;

- LLMs, with their capabilities and the amount of effort/money poured into them now, could be a straight path to AGI;

- If someone pushes LLM (or a successor model/ensemble) to near-AGI and somehow manages to keep it mostly aligned... someone else will unalign it out of spite, because that's how we roll on the Internet today.

Thanks for giving me a new appreciation for just how doomed we are.