Abstract:

Cooperative interactive storytelling is recognized as a powerful tool for promoting children’s communication and collaboration skills, which are fundamental to their social and cognitive development. This work focuses on cooperative interactive stories in which two children co-create an episodic narrative, a scene-by-scene story structure where each scene advances based on user choices. Combining cooperation with interactive storytelling stimulates curiosity and engages children in game-like experiences, creating an effective medium for developing communication skills.

To support such activities, a diverse set of suitable stories is required. However, producing and maintaining a sufficiently broad pool of child-appropriate narratives could be challenging. Generative AI offers a promising solution by efficiently producing varied and engaging narratives, but the technological medium must go beyond simple text generation.

Providing an intuitive and engaging environment that facilitates joint decision-making becomes a necessity. Generative AI offers a promising solution for producing varied narratives, but the raw textual output can be insufficient for educational contexts without a structured layer of interaction.

This paper presents the design and evaluation of a dedicated Graphical User Interface (GUI) designed to orchestrate cooperative experiences created through generative AI. Our approach utilizes the generative capabilities of the Large Language Models (LLMs) as a backend engine, but focuses primarily on how these narratives are visualized, navigated, and managed by young users. The interface is designed to bridge the gap between complex AI generation and child-appropriate interaction, ensuring that the focus remains on the cooperative dynamic between peers rather than the technology itself.

The proposed system addresses the challenges of presenting episodic narratives by offering visual cues, simplified text presentation, and distinct mechanisms for communication and cooperation development. We explored how visual design choices, in showing story branches and proposing cooperation, can influence the engagement level and the quality of collaboration between children.

To assess the usefulness of the developed GUI in educational settings, a test was conducted with five educators, each with more than 10 years of experience in children's education. This study focused on assessing the GUI’s usability and engagement capabilities perceived by the testers, and its possible effectiveness as a pedagogical tool.
The paper discusses the challenges of designing interfaces for AI-generated cooperative interactive storytelling activities, details the specific features of our GUI, and presents the findings from the educators' validation.

Results indicate that a well-designed interface is crucial in transforming LLMs’ raw generative capabilities into a suitable educational cooperative storytelling tool that can improve users’ cooperation and communication. Thus, enabling educators to exploit advanced technology to promote diverse social learning activities.

Keywords:

Cooperative Interactive Storytelling, Children, Generative AI, Large Language Models.