Research Focus

How can GenAI support collaboration between heterogeneous creators with different expertise, expressive modalities, and creative priorities?

Focusing on game narrative writers and visual designers, this study examines how these differences complicate building common ground. We identify opportunities for GenAI to mediate creative intent, shared context, and interpersonal communication, alongside concerns about trust and creative agency.

Research Approach

We conducted a multi-method study with 10 narrative writers and 10 visual designers in South Korea.

• Preliminary interviews: Examined individual workflows and collaboration challenges, informing a study designed around creators’ remote, asynchronous work practices.
• Creative prompting: Writers developed narratives with ChatGPT-4o. Designers asked the AI clarifying questions through shared conversation histories before creating concept art with Midjourney.
• Participatory design: Participants drew on these experiences and GenAI Capability Cards to propose collaboration tools and discuss their potential consequences.

Key Collaboration Challenges

• Knowledge and expression gaps: Writers struggled to communicate visual intent, while designers needed narrative and historical context that was often left implicit.
• Different creative priorities: Writers emphasized fidelity to narrative intent, while designers balanced that intent with visual originality and aesthetic coherence.
• Coordination and interpersonal barriers: Incomplete documentation, asynchronous schedules, hierarchy, and reluctance to question others’ expertise hindered clarification and feedback.

Key Design Opportunities

Participants envisioned GenAI supporting collaboration through:

• Translating creative intent: Generating visual references and alternative drafts to help creators communicate ideas and negotiate a shared direction.
• Maintaining shared context: Preserving narrative decisions, flagging inconsistencies, and answering contextual questions when collaborators are unavailable.
• Mediating communication: Clarifying terminology, adjusting message tone, and easing interpersonal tension. Designers’ questions could also reveal missing context, helping writers improve their explanations.
• Respecting different forms of expertise: Writers welcomed direct image generation, while designers preferred control over prompts, refinement, and project-specific styles—highlighting the need for role-sensitive support.

Critical Concerns

• Professional reputation and disclosure: Revealing AI use and early trial and error could undermine perceptions of competence, motivating selective sharing.
• Trust and accountability: Inaccurate AI outputs could misrepresent creative intent and damage relationships, while creators remained accountable for mistakes.
• Reduced human communication: Reliance on AI could displace conversations essential to mutual understanding, collective judgment, and creative negotiation.

Design Concept: Dual-Space GenAI

We proposed Dual-Space GenAI to connect individual creative work with collaboration while giving creators control over contextual sharing.

• Personal Studio: A private space for working with GenAI. Optional contextual memory captures creative reasoning, explored alternatives, and stylistic preferences.
• Shared Workshop: Creators share an artifact together with selected context. Collaborators can then ask the AI about the reasoning behind creative decisions.

This concept supports shared understanding across different areas of expertise while preserving control over disclosure and room for creative interpretation.