Original title: What Makes Creation Human? Authorship, Reasons, and Meaningful Human Control in Generative AI
Yuxi Cao · September 2026
Generative artificial intelligence helps people produce much more work, but producing more does not mean a person is truly the author. The researchers tackled this problem by asking what actually makes a human the creator when using AI tools. They found that human authorship does not come from manual effort, the level of automation, who had the first idea, or who makes the final choice. Instead, real authorship happens when human judgment and reasons guide how a project grows. The researchers explain that creative goals are not set completely in advance; they shift and evolve while working with the tool. To solve this, they created dynamic-reflexive trackingA method to ensure a person's changing reasons continuously shape AI creation, even as they evolve.. This framework shows how authorship works: a person's evolving reasons must be reflectively processed, directly change the creative path, and be able to reject the AI's default direction. Authorship depends on reasons shaping the work, not on counting manual steps.
Imagine you're cooking a meal with a friend. Normally, you decide the recipe together and stick to it. But suppose your friend starts changing ingredients based on tastes you haven't expressed yet, and you only get to approve the final dish. Here, the paper argues that true co-authorship requires that your evolving preferences—not just your initial recipe idea—actively shape each step of cooking. The AI is like the friend, and the paper's Dynamic-Reflexive Tracking ensures your voice influences every decision, not just the final outcome.
This matters because it challenges current legal and cultural assumptions about AI authorship. Artists and policymakers need a clearer standard for when AI-assisted work is truly human-authored. The paper proposes such a standard based on whether human reasoning genuinely shapes the creative process, not just the final output.
The abstract does not present experimental results or a concrete system for implementing DRT. It offers a conceptual framework but does not specify how to operationalize the four conditions in practice. The limitation is that the argument remains at the level of philosophical analysis, without addressing technical or legal implementation challenges.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
What Makes Creation Human? Authorship, Reasons, and Meaningful Human Control in Generative AI — Yuxi Cao
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.