RoboBuddy in the Classroom: Exploring LLM-Powered Social Robots for Storytelling in Learning and Integration Activities

Synthesis: Tozadore, Ertug, Chaker, and Abderrahim (2025) present RoboBuddy, an intuitive interface that lets teachers create scenario-based storytelling activities from their regular curriculum using LLMs and social robots. The system addresses two practical classroom challenges: the significant planning time required to create improvised scenarios for content delivery (intensified when using complex technologies like social robots), and the need to embed multicultural integration into an already tight curriculum. The authors co-designed activity frameworks with four teachers and deployed the system in a week-long study with 27 students.

Key Findings

• RoboBuddy enables teachers to create scenario-based activities from their regular curriculum using LLMs and social robots, reducing the planning burden associated with improvisational content delivery and social-robot use.
• The system addresses multicultural integration within regular curriculum activities, tackling it as part of everyday learning rather than as a separate add-on.
• Co-design with four teachers shaped the activity frameworks; the system was deployed in a study with 27 students over one week.
• Findings validated the system's efficacy and highlighted the positive impact of integration policies perceived by children, and the importance of scenario-based activities for content engagement.

Study Design & Method

This is a design-based/co-design and deployment study. The researchers co-designed scenario-based activity frameworks with four teachers using an LLM-powered interface that generates storytelling activities from curriculum content for a social robot. The system was deployed in a classroom study with 27 students over one week, evaluating the system's efficacy in delivering scenario-based learning content and supporting multicultural integration. Data collection involved the classroom deployment and assessment of children's perceptions.

What this means for practice

• Instructors. Generate scenario-based activities from the curriculum content you already teach instead of authoring improvised scenarios from scratch: co-design with four teachers produced frameworks the teachers judged accurate and offered without extra planning time, which they named as decisive for adopting the system.
• Instructors. Deliver content through a storytelling frame rather than an expository explanation — students exposed to the scenario-based strategy reported higher self-reported enjoyment across both the activity and language sessions.
• Instructors. Keep a human approval step between the LLM output and the robot: the interface's validation stage screened content for sensitive or inaccurate material before it reached the class, which is how the study mitigated LLM pitfalls.
• Instructors. Embed integration elements inside ordinary activities rather than adding separate multicultural sessions, since most children identified the embedded integration elements in their drawings without being prompted.

Limitations

The deployment was a single-week study with 27 students, limiting generalizability and insight into long-term effects. The interface and activity frameworks were co-designed with four teachers, so the design reflects their context. The study focused on integration policies and scenario-based efficacy, with learning outcomes not comprehensively measured.

Citation

Tozadore, D., Ertug, N., Chaker, Y., & Abderrahim, M. (2025). RoboBuddy in the classroom: Exploring LLM-powered social robots for storytelling in learning and integration activities.