Systematic Characterization of the Effectiveness of Alignment in Large Language Models for Categorical Decisions

Isaac Kohane

As large language models (LLMs) are deployed in high-stakes domains like healthcare, understanding how well their decision-making aligns with human preferences and values becomes c…