You are here

Modeling Mindreading in OntoAgent

Event date: 
Wednesday, 18 December, 2013 - 10:30
Sala Consiglio Scientifico
Sergei Nirenburg (Department of Cognitive Science Rensselaer Polytechnic Institute)

The term “mindreading” refers to an artificial intelligent agent’s ability to recognize, store and use knowledge about the resident world model, intentions, motives, goals, attitudes, biases, personality traits, emotional states and plans of other agents. The OntoAgent project pursues theoretical advances in modeling this capability and develops sample OntoAgents incorporated in proof-of-concept systems that demonstrate and test various facets of mindreading. Two such systems have been implemented:

  • MVP (Maryland Virtual Patient), an environment for training medical students in diagnostics and treatment selection in which the virtual patients and the tutor are OntoAgents;
  • CLAD (CLinician’s Advisor), a system implementing an intelligent agent that monitors the interaction of a (human) physician with a (human) patient and helps to lower the physician’s cognitive load by offering intelligent and timely advice.

In this talk I will introduce the OntoAgent approach to mindreading modeling and present selected highlights of the current versions of MVP and CLAD.

The OntoAgent research program is by nature interdisciplinary. Moreover, the potential application areas of OntoAgent-like agents clearly extend beyond clinical medicine. In the talk, I will highlight a variety of areas of potential interaction between OntoAgent and research and development efforts in several research areas and application domains.