Research Overview: SaGE - Quantifying Moral Consistency in LLMs

This lecture from NPTEL-NOC IITM provides a comprehensive research overview of SaGE (Systematic Assessment of Generative Ethics), a framework for quantifying moral consistency in Large Language Models (LLMs). During the 26-minute presentation, explore how researchers evaluate ethical reasoning and moral consistency in AI systems, with particular focus on measuring how consistently LLMs apply ethical principles across different scenarios. Learn about the methodologies used to assess whether language models maintain coherent moral frameworks when faced with varying ethical dilemmas. The lecture includes supplementary materials available through a provided Google Drive link for deeper understanding of this important aspect of AI alignment and safety research.