Personality-driven collaborative instructional design with digital humans: an empirical exploration based on Spanish public courses
人格驱动的数字人协同教学设计：基于西班牙语公共课的实证探索

摘要

随着人工智能生成内容（AIGC）和数字人技术的迅速发展，推动了教育领域教学模式和互动方式的深刻变革，传统公共课程在师资、课堂互动、激发学习动力等方面存在明显不足，“人格驱动”的数字人教学模式逐渐成为研究热点。本研究以西班牙语公共课程为例，采用实证分析方法设计多主体协同的教学框架，借助MIGUEL、ZENOBIA和西西三个具有人格特质的虚拟助教，建立基于个体差异的任务分配和协作机制，从而提升教学效果并提高学生参与度。经过对依照学生问卷调查的数据做统计显示，受访者能够正确识别出数字人持有的个性特点，认为数字人的性格会显著影响到自身学习态度的人占比达57.58%，另有72.73%的学员认同采用“数字人 +AI ”方式自学，同时有合理的安排数字人及其使用的频次会切实提高教学效率及提升自身的学习体验，在此次的调研当中既为教育数字人从功能性转向人格性转变在理论上给予证据支撑，又形成了如何形成高效互动式教学体系并加以完善的思路方案。

Abstract

With the swift advancement of AIGC and digital human technologies, the education sector is experiencing a profound transformation in teaching approaches and interaction patterns. Traditional public courses encounter various challenges in teacher assignment, classroom interaction, and student motivation. Leveraging digital technologies to optimize teaching has become a key issue worthy of exploration. This study examines a "personality-driven" digital human teaching model through an empirical investigation in a Spanish public course. A multi-role collaborative teaching design was developed, focusing on three digital characters with unique personality traits—MIGUEL, ZENOBIA, and Xixi—to establish a personality-adaptive task allocation and collaboration mechanism. Analysis of student questionnaire data shows that students can clearly perceive the distinct personalities of digital characters, and 57.58% think that a digital human's personality affects their reception of the teaching content. Moreover, 72.73% of students report being able to independently understand content with the aid of the "Digital Human + AI" combination, even without additional teacher explanation. Suitable frequency of use and role allocation also boost learning efficiency and teaching satisfaction. This study broadens the theoretical path from function-driven to personality-driven digital humans in education, offering practical, implementable teaching strategies and optimization suggestions.