Deeper learning through interactions with students in natural language

This chapter reviews work on building adaptive educational systems with natural language interaction, to note current challenges and discuss potential solutions based on recent advances in Artificial Intelligence. Interacting with students in natural language has many advantages. It encourages deeper, conceptual learning as students are required to explain their reasoning and reflect on their basic approach to solving a problem. It also provides deeper insights into the learning process by revealing students’ reasoning and mental model construction processes, including potential misconceptions, thereby enabling better detection of opportunities to scaffold students’ learning by providing immediate corrective feedback, which is most effective. Furthermore, natural language interactions give students the opportunity to develop the language of their target professional communities and therefore much-needed communication skills. When conversational AI systems for learning are augmented with virtual agents, they have affordances to social agency which allegedly leads to more engaging, satisfying, and effective learning experiences. This chapter primarily focuses on dialogue-based intelligent tutoring systems (ITSs) with one-on-one tutoring. We also make some conjectures about the length and density of interactive, intense learning sessions such as those students experience when interacting one-on-one with a dialogue-based ITS.