Designing Emotionally Intelligent Generative AI: Enhancing Self-Regulated Learning and Socio-Emotional Outcomes in Education

While generative AI (GenAI) tools are emerging as powerful educational resources, general-purpose applications like ChatGPT are not explicitly designed to support the metacognitive and socio-emotional dimensions of learning, often failing to address students’ emotional states or promote independent thinking. In response to these challenges, the present study aims to investigate the impact of a learner-centred GenAI chatbot design that combines self-regulated learning (SRL) scaffolding and emotional support. In a pre-registered experiment conducted in an authentic educational setting, 144 high school students were randomly assigned to one of three conditions: (1) ChatGPT, (2) SRLbot (GenAI with SRL scaffolding), (3) Emotional SRLbot (GenAI with SRL and emotional support). Outcomes included learning achievement (conceptual knowledge and transfer), engagement in SRL phases (forethought, performance, self-reflection), and socio-emotional states (enjoyment, boredom, hopelessness, self-efficacy, and trust) assessed at pre-, post-, and follow-up tests. All chatbot conditions significantly improved conceptual knowledge and transfer from pre- to post-test, and only the ChatGPT group showed a significant decline at follow-up. Both SRLbot and Emotional SRLbot outperformed ChatGPT in promoting performance-phase SRL processes. Yet, only the Emotional SRLbot, significantly outperformed ChatGPT in increasing enjoyment and reducing boredom. These findings offer novel insights into how GenAI agents integrating SRL and emotional scaffolding can more effectively foster productive emotional states and self-regulation than general-purpose GenAI. This advances the design of emotionally intelligent and inclusive GenAI tools that humanize and personalize the learning experience. notReviewed other