Principal Investigator: Carolyn Rose
CoPrincipal Investigator(s):
Organization: Carnegie-Mellon University

Abstract:
This Cyberlearning: Transforming Education project brings together leading-edge researchers in computational linguistics and computer-supported collaborative learning to explore feasibility issues in designing an intelligent conversational agent that interacts with groups of learners as they are working together and provides cognitive, meta-cognitive, and social advice to enhance their collaborations. The conversational agent is being designed to enhance group dynamics by providing supportive advice as well as promoting awareness of the dynamics. Design of the agent is informed by theories of meta-cognition, collaborative learning, and accountable talk. In addition to investigating how to design such agents and testing their effectiveness, investigators are exploring how learning develops when a conversational agent is available to help with productive talk, the influences of conversational behaviors on developing understanding and actions, and ways of influencing understanding and behaviors.

One of the biggest limitations of online education, especially massive online courses (e.g., MOOCs), is personal contact with instructors and peers. While much of formal education focuses on faculty interactions with large groups of students in lectures, much student learning depends on interactions with teachers and teaching assistants and discussions, study sessions, and project work done with peers. Massive online education will be broadly successful only after we know how to provide these kinds of supports to online learners. This project focuses on designing an interactive agent that can provide conversational support for small groups of collaborating students to help them productively solve problems and learn together. The agent is being designed to provide the kinds of advice that a teacher or teaching assistant would provide if such a human mentor were available. The researchers conjecture that with such support available, online students can provide for each other the kinds of social supports for learning that contribute in integral ways to students’ learning experiences in higher education. The agent is being deployed and tested in several different online courses. Results are expected to be applicable to massive online education as well as to blended educational environments such as flipped and project-based classrooms.