Personality analysis in education enables educators to tailor their teaching methods by offering insights into students’ personality traits, creating more personalized and effective learning environments. While traditional assessments often fail to capture personality dynamics, computational personality analysis using Machine Learning (ML) offers a more nuanced approach. However, ensuring the interpretability of these ML models remains a challenge. This research addresses the gap by exploring how to enhance the interpretability of Multi-Task Learning (MTL) models in educational contexts, aiming to provide educators with actionable AI-driven insights for personalized learning. This paper applies a Multi-Gate Mixture of Experts model based on MTL for Computational Personality Analysis (MTLCP), and it examines various analytical approaches using different computational models. The MTLCP enhances the trustworthiness of personality model interpretations by employing Model-Agnostic Explanations (ME) such as Local Interpretable Model-Agnostic Explanations (LIME) and conducting model explanation similarity assessments. The results indicate that MTL achieves a cosine similarity higher and more consistent than that obtained with the Multi-Label Learning (MLL) technique. Finally, the application of MTL in the explanatory stage adds to the degree of integration and consistency, which translates into better quality analyses that are useful for educational research.