Answers are grounded in this paper and cite their sources.
We introduce a constructivist approach to AI-enabled malleable writing environments in which writers shape both their text and tools. This contrasts with one-size-fits-all writing software and instructivist UI-generation paradigms, which delegate dynamic interface creation to the system. Instead, our canvas-based interface allows users to construct, adapt, and reuse AI-enabled tools by working with nodes (notes, pages, buttons). Nodes can be reinterpreted across the roles of material and tool via a front-back metaphor: Text material appears on the front, while behavior and AI involvement are defined through editable prompts on the back. We deployed a functional prototype as a technology probe in two user studies (N = 16; N = 8 follow-up) and expert workshops (N = 12), showing how generative malleability supports externalizing goals and approaches, task decomposition, and deliberate AI scoping, while introducing metacognitive demands around tool construction and abstraction. This work advances AI writing tools and prompting-based generative UIs beyond efficiency and automation, toward user-constructed environments that evolve with writers’ thinking.
No takes yet. Share an insight, caveat, or question.
Zindulka et al. (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: