Bio: Hua Shen is an Assistant Professor of Computer Science at NYU Shanghai, jointly affiliated with NYU Tandon. She received her Ph.D. from Penn State and was previously a postdoctoral scholar at the University of Washington and the University of Michigan. Her research sits at the intersection of Human–Computer Interaction and Artificial Intelligence, with a focus on bidirectional human–AI alignment—developing methods for humans to explain, evaluate, and collaborate with AI systems while integrating human values into AI design. Her work has been recognized with multiple awards, including a 2023 Rising Star in Data Science, Best Paper at AIED 2024, and Best Demo at CSCW 2023. More about her research is available at https://hua-shen.org.
Abstract: As large language models (LLMs) become embedded in daily life, ensuring their alignment with human values, and also understanding their influence on human opinions, has emerged as a central challenge. This talk clarifies the concept of human–AI alignment and presents insights from recent empirical studies. I will introduce methods for measuring value alignment, highlighting evidence of value misalignment in LLMs and their potential risks. Finally, I will present findings on bidirectional opinion dynamics, showing how interactions with LLMs influence not only human perspectives but also model responses. Together, these studies reveal both the challenges and opportunities of building human-centered, responsible, and safe AI systems.