Abstract

Increasingly complex and advanced AI systems are performing an increasing number of tasks in our day to day lives. Some of these tasks are hidden, such as maintaining the electrical grid, and others are more overt, such as self-driving taxis or direct communications with chatbots. A desire of many researchers is to use AI systems as teammates rather than tools, in a way that benefits the performance and well-being of their human users. In this talk, we present some architectural requirements for AI systems to become team players, and discuss how Large Language Models satisfy these requirements and implications for their use when they do not.

About the Speaker

Thomas Eskridge is a Professor of Computer Science and Human-centered Design at Florida Tech, and a researcher at the L3Harris Institute for Assured Information.