Generative AI for Analyzing Real Teamwork Interactions and Identifying Socioemotional Competencies

Socioemotional competencies are a key priority in higher education. However, their systematic integration and assessment remain challenging. This study examines how a structured teamwork context, based on the Comprehensive Training Model of the Teamwork Competence, generates interactional evidence that allows socioemotional competencies to be observed using Generative Artificial Intelligence. The study was conducted with eight teams of university students, focusing on the analysis of interaction threads. Seven socioemotional competencies were assessed using a rubric that was automatically applied by two Generative Artificial Intelligence systems: ChatGPT and DeepSeek. The results were visualized using heat maps representing scores by team and by competency. Both systems identified cooperation, responsibility, flexibility/adaptability, and communication as the most developed competencies, while empathy and support/help were the least developed. In addition, the relative ranking of competencies within teams remained consistent across both models, although differences were observed in the absolute values generated. Overall, the findings indicate that Generative Artificial Intelligence is useful for the descriptive analysis of socioemotional trends in teamwork contexts, but further calibration is required before it can be reliably applied to quantitative individual or group assessment within teams.