News

At the prestigious NeurIPS 2024 conference, Hannah Rose Kirk and Professor He He received the Best Paper Award in the Datasets & Benchmarks track for their remarkable work on the PRISM Alignment Dataset. This dataset provides critical insights into the subjective and multicultural alignment of Large Language Models (LLMs). With data from 75 countries and diverse demographics, their work benchmarks over 20 cutting-edge models and addresses challenges in Reinforcement Learning with Human Feedback (RLHF). Their research sets a new standard for advancing pluralism and understanding disagreements in AI alignment, solidifying their contributions to the field. Read more.