Abstract:
Artificial Agents are increasingly interacting with humans on a daily basis. Whether they are personal assistants on our phones, customer service agents, or e-commerce bidding bots, these agents are becoming ubiquitous. Many of these tasks can be modeled as bilateral negotiations, and the agents that engage in them require advanced social intelligence. This talk explores an incremental roadmap to socially-aware agents, shows empirical results of human studies on human-agent interaction, and explores the horizons of this burgeoning research community.
Biography:
Johnathan Mell is part of the University of Southern California's Institute for Creative Technologies. Johnathan is a part of the ICT emotion group, where he designs more human-like computers for a variety of applications. Johnathan completed undergraduate work at the University of Pennsylvania, where he received his Bachelor's degrees in Economics (Entrepreneurial Management) from the Wharton School, and in Computer Engineering, from the School of Engineering and Applied Science in 2013. His undergraduate research was focused primarily on psychophysiological channels for game feedback mechanisms and on frustration in games and its effect on mood.
Johnathan’s current research focuses on the impact of social features of repeated negotiations with a computer partner. His work covers favor exchange, cross-cultural features, and temporal effects in an effort to make automated negotiators and emotive and realistic virtual characters. He is also interested in efficient designs for systems that are used by a non-AI "man behind the curtain", called "Wizard of Oz" systems. To investigate these questions, he has developed the IAGO platform, which serves as a framework for creating Virtual Agents that negotiate with humans.
Johnathan is published at AAMAS, IJCAI, AAAI, IVA, and ACII, and his IAGO platform was a finalist for Best Demonstration at AAMAS 2016. He is a member of the organizing committee for the Automated Negotiating Agents Competition (ANAC). He has previously worked on human interface and training platforms for Disney Engineering, and intends to continue working at the intersection of entertainment, computation, and human interfaces.
Read More