Theory of Mind and Intent in Procedural Agents: Learning Lessons from Improvisational Theater and Text-based Adventure Games


Speaker:Jonathan May (USC)

Date and Time: 12pm CT, July 12

Place: TBD

Abstract:

I might be able to have some fun chit-chatting with the bot on the other end of the line and I might be able to book a flight too, but I'm under no impression that the bot really understands me, nor do I expect the bot to actively work with me to try and solve a problem. What are the key improvements we need to make to get from Siri and Alexa to Jarvis and Samantha? In the first part of this talk I'll discuss the importance of focusing on grounding acts as a way to indicate shared theory of mind, and how we can learn lessons (and mine data) from improvisational theater. In the second part I'll present work on reinforcement learning to make decisions and take action given a language signal, even in an unfamiliar domain.

Bio:

Jonathan May is a Research Assistant Professor in the Computer Science Department of the Viterbi School of Engineering at the University of Southern California, as well as a Research Lead with USC’s Information Sciences Institute, where he received his PhD in 2010. He has previously worked at BBN Technologies and at Language Weaver. His research interests include machine translation, generation, semantics, and machine learning. He was a co-organizer of the International Workshop on Semantic Evaluation (SemEval) and is the current treasurer of the North American Chapter of the Association for Computational Linguistics (NAACL). He has received a research award from ISI, an outstanding paper award from NAACL, and a best demo paper award from ACL.