By Ran Liu, Chief Data Scientist at MARi
One of the ultimate goals oaf education is to effectively prepare students for long-term success. Most existing intelligent systems in education focus on adaptive tutoring for specific academic subjects and deliver personalized learning on a relatively short-term scale. Delivering sustained personalization of learning/mentorship for long-term success and aligning education to a rapidly-changing workforce remain lesser explored issues in artificial intelligence. I will discuss some of the unique challenges presented by these goals. These challenges include (1) the ability to track and integrate data from many disparate sources, at multiple grain sizes, and over long periods of time, (2) the need to adapt personalized learning models to consider longitudinal, cross-discipline, whole-person context rather than based strictly on within-tutor or within-session data, (3) the need to adapt models of engagement and motivation to consider longer-term trajectories and broader categories of behavior (for example, school attendance and discipline trajectories in addition to momentary estimates of affect and engagement), and (4) the real-time alignment of educational goals to the skills and knowledge needed in a rapidly-changing workforce. I will discuss promising approaches that can help us solve each of these challenges and move us closer to building effective intelligent mentoring systems.