Learning with Chatbots

AP Dr Nantha Kumar Subramaniam,
Head Centre for Learning Technology, OUM



A chatbot is a computer programme that, through voice or text, simulates how a human would behave as a conversational partner.


Introduction: Artificial Intelligence in Online Learning


Advancements in the field of Artificial Intelligence (AI) has had a clear spill-over effect on higher education, particularly online learning. In this article, I will be looking at an AI derivation to support "student-interface interaction", a relatively new form of interaction in online learning that complements the conventional student-student, student-instructor, and student-content interactions. There are various ways through which student-interface interaction may be realised. For the purpose of this short article, I will focus on the deployment of intelligent chat robots, or chatbots, as a way of enabling student-interface interaction. And I will try to show that chatbots have the potential to make online teaching and learning much more engaging and effective.


Chatbots to the Fore


Student-interface interaction is defined as the interaction between the learner and the tools needed to perform the required learning task. In the student-interface dyad, the former in most cases requires active participation, while the latter may take the form of the chatbot. A chatbot is a computer programme that, through voice or text, simulates how a human would behave as a conversational partner As part of my research, seven instructional-based chatbots were built to test how they might help students learn the Java programming language. A programming course was chosen because learning programming like Java is difficult. Most computing educators would be familiar with the struggles of their students as they battle in vain to understand the basics of programming concepts and constructs.


The chatbots were designed to turn the commonplace lectures into a series of messages that look like a chat conversation. Each chatbot focuses on selected concepts covered in the course. All of them require students to learn-by-doing in a series of problem-solving steps. As well, the chatbots possess the following unique features:


i. They are self-contained yet interconnected to each other. As well, they are able to initiate learning based onselected learning outcomes and to provide feedback to students as they work through the problems.

ii. They are able to engage students on a one-to-one basis, conversing with them in a problem-solving process for more than an hour.

iii. They support immersive learning via simulation of realistic scenarios and environments, providing students with an opportunity to acquire set skills while interacting with the automated tutor.

iv. They support all the major learning domains, namely cognitive, psychomotor, and affective.


The chatbots acquire their intelligence through a hybrid approach that combines pattern-matching, machine learning algorithm, and the retrieval-based model in order to formulate their responses and to propel learning. They support real-time human-like dialogue to help students learn Java in an interactive and engaging way via an automated tutor.


AI holds great promise in enriching online learning, not only for programming courses but also for non-technical courses


In addition, the chatbots are able to carry a running conversation, complete with probing questions, positive and negative feedback, follow-up questions, and requests for explanation as to why something is correct or incorrect. As well, they are able to push interactive exercises based on the behaviour-recording approach. Pedagogically speaking, the chatbots built for my research to teach programming sought to promote self-managed learning, covering student self-awareness and self-evaluation of the learning process, independence, and self-direction. The chatbots also promote critical thinking and analysis, as well as problem-based learning.


Conclusion: Widening Deployment


Over a span of about three years involving over 150 programming students who filled survey forms and sat for pre- and post-tests, my research into the educational application of AI, specifically intelligent chatbots, yields sufficient grounds for me to claim that AI holds great promise in enriching online learning, not only for programming courses but also for non-technical courses. Deploying chatbots to drive online learning is immensely cost-effective, cutting across, as it does, time-zones and geographical limits. A push for its wider deployment is certainly something I would advocate.