Generativity is an increasingly popular and useful concept, referring to a machine’s ability to respond to user input with new constructions not foreseen by the programmer. Yet increasingly, people treat computational systems as unknowable black-box systems, writing off the possibility of forming mental models that allow a collaborative relationship between human insight and fast computation.

I argue for the efficacy of transparent, compositional semantics for collaborating with virtual agents and deriving insights from system models. Having built systems based on automated reasoning for linear logic and epistemic modal logic, we can formalize notions of belief, intention, and action, in order to create virtual agents that behave in ways that humans can reason about based on intuitions about goal-driven behavior. For example, some of Grice’s maxims of conversation can be seen as derivable consequences of these principles. Ongoing work includes applying these formalisms to the tasks of navigating unknown rule systems in virtual environments, social skills training, and generative storytelling.

Ph.D. Carnegie Mellon University, 2015

Postdoc UC Santa Cruz, 2015-2016

Assistant Professor, NC State University, 2016-present