Disabling a suspected feature or circuit to see whether the behavior it's supposed to explain disappears.
Continue to AI University →