EdoArad comments on My personal cruxes for working on AI safety

[EdoArad]

Jaime Sevilla wrote a long (albeit preliminary) and interesting report on the topic