Virtual experts : who's online?

There will be between 30 and 80 million online students in the world by 2025. Globally, school systems are investing huge resources in the development of online education programs in order to meet this demand. Related implications for online teaching have been largely ignored despite evidence of the greatly increased workload for online teachers and widespread student dissatisfaction with online leaming experiences. Opportunities for improving the online experience for both leamers and teachers revolve around minimizing procedural inefficiencies in dealing with large numbers of individual students, as opposed to a single class, and of enhancing students' social and cognitive engagement with leaming. Intelligent software agents that can automate many routine online tasks and some aspects of leamer interaction have enormous potential to facilitate this. These agents that can act as a personal online coach, mentor or tutor to increase the individualisation of learning. The development of evidence-based agent personas is essential if agents are to fulfil specific educational roles. CUITently there is little progress being made in this area because of the lack of an agentrole model that can be used to implement specific educational personas in agents. In this paper we discuss key foundational elements of the nature and basis for implementing elements of educational expertise in software and how this could be used in developing agent persona models for specific educational roles and a model for implementing pedagogical constructs in intelligent educational software.