Video Recordings

With recent progress in deep learning, there has been an increased interest in learning dialogue systems from data, also known as “Conversational AI”. In this talk, I will focus on the task of response generation, for which I will highlight lessons learnt and ongoing challenges, including reducing ‘hallucinations’ for task-based systems, safety critical issues for open-domain chatbots, and the often-overlooked problem of ‘good’ persona design. I will argue that we will need to solve these challenges to create trusted, safe and bias-free systems for end-user applications.