Synthetic Dialogue Generation with Role-Playing LLMs I - Day 9

Explore synthetic dialogue generation techniques using role-playing large language models in this comprehensive lecture presentation from JSALT 2025. Learn from leading researchers including Ricard Marxer from University of Toulon, Sergio Burdisso from Idiap Research Institute, Alessa Carbo from Johns Hopkins University, Antonio Almudevar from University of Zaragoza, Severin Baroudi from University of Toulon, and Isabella Gidi from Harvard University as they delve into advanced methods for creating realistic conversational AI systems. Discover how role-playing capabilities in LLMs can be leveraged to generate high-quality synthetic dialogues, examine the underlying mechanisms that enable these models to adopt different personas and conversational styles, and understand the applications and implications of this technology in natural language processing and dialogue systems research. Gain insights into current methodologies, challenges, and future directions in the field of synthetic dialogue generation through detailed slides and expert analysis from this intensive academic workshop session.