CIIR Talk Series: Beyond the Behavioral Turing Test: Initial Steps Towards an AI Behavioral Science
Abstract: Do Large Language Models (LLMs) possess personality traits, and do their behaviors differ from human ones? Exploring the preferences and behaviors of LLMs to ensure alignment with human values is crucial for facilitating effective human-AI collaboration and maximizing societal benefits. As a foundational step towards an AI behavioral science, we introduce a behavioral Turing test that contrasts the behavioral traits of LLMs against those of hundreds of thousands of human participants. In the second part of the talk, we explore a methodology teaching LLMs to generate content tailored to the specific context of a user, drawing on principles from writing education.
Bio: Qiaozhu Mei a professor in the School of Information and the College of Engineering at the University of Michigan, where he served as the founding director of the Master of Applied Data Science program. His research group develops novel methods in machine learning, data mining, information retrieval, and natural language processing and applies them to diverse domains, such as the Web, social media, healthcare, and education. His work has received multiple best paper awards at ICML, WWW, WSDM, KDD, and other major conferences in computing. Qiaozhu is an ACM Distinguished Member. He has served as the General Co-Chair of ACM SIGIR 2018 and currently serves on the editorial board of the Journal of Machine Learning Research, the ACM Transactions on the Web, and the IEEE Transactions on Big Data.
To obtain the passcode for this series, please see the event advertisement on the seminars email list or reach out to ataubman [at] cs.umass.edu (Alex Taubman).
For any questions about this event with the Center for Intelligent Information Retrieval, please contact please contact zamani [at] cs.umass.edu (Hamed Zamani).