Socially-Responsible NLP: Modeling Human and AI Language in Context
Abstract:
Natural language technologies are transforming how we access information, communicate with others, and interact with AI. Yet, traditional NLP systems and research often overlook a crucial element: the social context in which language is produced and interpreted. Social context, including the relationships, motivations, and situational factors surrounding communication, fundamentally shapes the meaning and impact of language. In this talk, I will present my integrated research that examines language in social contexts, with the social scientific goal of understanding human behavior and the technical goal of advancing the social capabilities of language technologies and interventions to better support the lives of individuals and groups. The first theme explores human-human interaction in online communities to understand how people’s language influences social outcomes, such as user engagement and personal well-being. The second theme focuses on human-AI interaction, specifically the social aspects of natural language generation models and their ability to generate affective responses in large-scale communication platforms. Through large-scale online experiments and real-world deployments, my research demonstrates how considering social context leads to more robust and beneficial language technologies.
Bio:
Yi-Chia Wang is a Principal Research Scientist at Carnegie Mellon University and is affiliated with Stanford University as a Visiting Research Scholar. Previously, she was a Research Scientist in the Reality Labs at Meta AI, and in the Conversational AI team at Uber AI. She earned her Ph.D. from the Language Technologies Institute at Carnegie Mellon University. Her interdisciplinary work spans Natural Language Processing, Computational Social Science, and AI, focusing on human-human and human-AI interactions, with an emphasis on engagement, well-being, and trust and safety.
Yi-Chia has published over 40 papers in top conferences and journals, including ACL, EMNLP, KDD, CHI, CSCW, and TheWebConf. She has received several awards, such as the CSCW Best Paper Award and the CHI Honorable Mention. Yi-Chia has served as an Associate Chair and Area Chair for various conferences, and has been invited to give keynote speeches and seminar talks at venues such as the Workshop on Social Influence in Conversations at EMNLP, the SocialNLP Workshop at ACL, Johns Hopkins University, UC Davis, UC Santa Barbara, Santa Clara University, and Amazon.