| Sumario: | This study developed and validated EmotionBridge, a GPT-based diagnostic tool for assessing empathic communication through real-time, scenario-based interactions. Using a Design and Development Research (DDR) approach, the system was grounded in a theoretical framework with two domains: empathic expression and empathic response. The prototype, built via GPT Builder, incorporated rubric-based evaluations. Expert reviews demonstrated high content validity (CVI = .95 and .97), and concurrent validity was supported by correlations between expert and self-report scores. Usability testing with 35 participants yielded a strong SUS score (79.64), and qualitative feedback informed iterative refinements. The results demonstrate the promise of generative AI in delivering scalable and personalized diagnostics in social-emotional learning, offering a replicable design model that links large language models with structured assessment principles.
|
|---|