A Framework for AI-Powered Decision Making in Developing Adaptive e-Learning Systems to Impact Learners' Emotional Responses

This paper proposes and implements an AI-based decision framework for e-learning systems that can assess learners' emotions and adjust the learning activity in order to improve learning performance. An evolutionary genetic algorithm is proposed to identify suitable micro-brake activities to alter learners' emotions in the event learners feel emotions that are not optimal for learning, such as anxiety or sadness, so they can concentrate on learning more effectively. To test the proposed framework, a case study was conducted with English as second language learners over one semester. Fourteen participants were recruited from a gifted school and were randomly allocated to two groups, each consisting of seven participants. The first group was provided access to the system and its break activities, whereas the second group had no access to this system. The results showed that students with access to the system with break activities performed significantly better. This confirmed that micro-break activities, selected based on students' sentiments and inclinations, can have a positive effect on learning performance. The results have also confirmed that the proposed framework was successful in selecting and offering students effective activities that could improve student's mood. The results of this study hold significant practical implications for the design of adaptive e-learning systems and learning management platforms. Additionally, they contribute to the theoretical landscape of AI and learner emotions by proposing an innovative method for identifying, classifying, and providing personalized learning paths tailored to the unique needs of individual learners.