Thanks to @PranModu for this excellent reflection on pluralistic alignment that highlights what I believe is so wrong with a lot of the alignment discourse: https://t.co/hC1YzQRtN0. May be of interest to @audreyt@VitalikButerin@metaviv and @divyasiddarth
@AndrewLampinen Because RSA and regression based analyses are leading to these biases, what kind of alternate methods or controls would you suggest for more robust model representation interpretation?