Who Is The Creator? Rethinking Authorship In The Age Of Generative AI

1. Introduction

Generative artificial intelligence now produces articles, images, music, video, software code, and designs at a scale few people expected a decade ago. The usual way of describing this material is to call it "AI-generated," and more and more often "AI-created." Publishing platforms, detection tools and public commentary use these labels as if they settle the question of where a work came from. This article argues that they do not.

The label is accurate at one level. A model does produce the final sequence of words, pixels or sounds. What the label leaves out is everything that made that production possible. The languages a model writes in, the knowledge it draws on, the examples it learned from, the architecture it runs on and the purposes it serves were all supplied by human beings. The central claim of this article is that every output of generative AI is best understood as human-made, even when an AI system is the immediate mechanism that produces it.