Charlie Steiner comments on Why is pseudo-alignment “worse” than other ways ML can fail to generalize?

[Charlie Steiner]

)

[Ben Pace]

oh no