RobertM comments on Interpretability Will Not Reliably Find Deceptive AI

[RobertM]

Whoops, yes, thanks, edited.