A post questions effective altruists’ confidence in spotting fake AI alignment
A user argues that effective altruists cannot recognize human bad actors even with all the evidence, yet expect to tell whether superhuman AI is genuinely aligned.
TLDR
The critique contrasts judgments about people with expectations for AI: a user says effective altruists fail to recognize human bad actors despite having all the evidence, but believe they could distinguish a genuinely aligned superhuman AI from one merely faking alignment.
Combined views
—
1 Source, first seen ago
