David Scott Krueger (formerly: capybaralet) comments on False assumptions and leaky abstractions in machine learning and AI safety

[David Scott Krueger (formerly: capybaralet)]

IIUC, yes, that’s basically what I was trying to say about embedded agency.