Can a model represent genuine preference pluralism at cluster level without collapsing to an average — and would that be legible enough to anyone to matter?jordanSA·...importedcommunication · 9.5machine-learningai-safetyinterpretabilityUpTrustDownTrustComments0Reactmore options