Building Interpretable Models for Moral Decision-Making

Mayank Goel, Aritra Das, Paras Chopra

We build a custom transformer model to study how neural networks make moral decisions on trolley-style dilemmas. The model processes structured scenarios using embeddings that enco…