Research Summary: A humorous social-affective instructional style in generative AI pedagogical agents: effects on cognitive load, immediate learning performance, and learner perceptions

Research explored the impact of incorporating a humorous, social-affective instructional style into generative AI pedagogical agents on student learning. The study, involving Chinese university students, aimed to determine if such an approach influences cognitive load, immediate learning performance, and learner perceptions compared to a standard instructional style. The provided abstract is incomplete, so specific findings on the effects are not available, but the research design and intent are clear.

Why it matters

This research is strategically important as it explores methods to enhance engagement and effectiveness of AI in educational and training contexts. By examining instructional styles, it informs the development of more sophisticated and learner-centric AI agents, potentially improving knowledge transfer and user satisfaction in various applications.

Key insights

• Generative AI pedagogical agents offer adaptive instructional dialogue but can be perceived as impersonal or overly formal.
• The study investigated a humorous social-affective instructional style for AI agents, combining content-related, non-disparaging humor with colloquialisms and encouraging language.
• Key metrics for evaluation included learners' self-reported cognitive load, immediate post-test performance, and perceived agent value.
• A randomized experiment was conducted with 120 Chinese university students, manipulating instructional style and considering learner gender as a moderator.

Source

Educational Technology Research and Development — https://link.springer.com/article/10.1007/s11423-026-10715-x