2024 - present

Creativity support tools have long played a vital role in helping people produce creative work. Traditionally, however, these tools have been largely passive. For example, spell-checkers, version control systems, and template libraries reflect and refine a creator's intent rather than shape it. AI and especially generative AI fundamentally shifts this dynamic, introducing systems that can reason about creative content in their own right. This naturally gives rise to tools that actively suggest, iterate, and challenge a creator's assumptions, while also enabling new forms of expression and reimagined creative mediums. Under this research theme, we investigate two intertwined questions: how generative AI can be designed to uphold artists' agency, and how co-creating with generative AI can enable new types of creative expression. For instance, in one research paper, we leverage LLM-based multi-agent systems to help storywriters discover and develop a cast of characters in ways that would be difficult to achieve alone, while preserving their authorial agency over their final story. By studying these and related cases, we aim to advance a deeper understanding of how generative AI can be designed to uphold artists' creative voice, and how co-creating with generative AI can push the boundaries of what creators can express.