Abstract:

AI-driven educational tools can be effective for children’s learning experience, and especially conversational AI can have a positive impact on a child’s learning journey. Using conversational AI, with generative AI language models, personalised study partners can be made for children that can learn with them, motivate them, teach them, and gamify their learning experience using one-to-one conversational interactions. In this study, TeddyAI, a conversational AI learning app, tests this concept. It is a conversational Teddy bear, powered completely by AI, that supports learners between the ages of 4 and 7. From a teaching perspective, it helps them learn alphabets, and topics like general geography, history, animals in a zoo. Apart from these, TeddyAI is also able to teach two subjects from Key Stage 1 and 2 of the UK national curriculum. The objective of this research is to understand the user experience and overall usability of TeddyAI to offer a personalised approach to learning by delivering tailored experiences through the study partner; Teddy. To understand the user experience of TeddyAI, 35 primary school children interacted with TeddyAI, each for 12-15 minutes. During their interaction, their real-time engagement was measured, and they were observed. The System Usability Scale (SUS) (Brooke, 1995) was used to gather user feedback and was followed by a discussion to explore the experiences of the children in detail. Results showed that the SUS score was 72.6 indicating a score above the average of 68. This suggests that TeddyAI’s usability is good, the children found it easy to play TeddyAI, and loved their interactions with him. This study provides insights for future improvements for AI study partners, focusing primarily on user interaction and engagement.

Keywords:

Artificial Intelligence, Conversational AI, Education, Personalised learning, Generative AI.