Post by Yale School of Engineering & Applied Science

AI models make mistakes. But there's an important distinction between a model that is misinformed, meaning it lacks the right knowledge, and one that is misaligned, subtly steering you toward an outcome that doesn't reflect your intentions. Understanding the difference matters enormously. Yale Engineering professors Yang Cai and Nicole Immorlica, along with colleagues from economics and statistics, are developing frameworks to tell the two apart. Yale News sat down with the researchers to find out how.