How many (currently unknown) problems for alignment on the same level as inner optimizers will we encounter in very advanced AI systems?
Hm this is not very reassuring guys
a Schelling point for those who seek one