Reproduce “Creative Ownership in the Age of AI” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Annie Liang, Jay Lu
Researchers from Northwestern University and UCLA developed a formal economic model to define creative ownership and infringement for generative AI, proposing a "counterfactual dependence" criterion to determine if an AI output relies on a specific copyrighted work. Their analysis suggests that infringement risk may vanish in creative markets with gradual innovation but persist in markets driven by "superstar" creations.
Key insights
- The "permissible set" of non-infringing creations is stable and expands or remains constant as the corpus grows, implying that combinations of permissible works remain permissible.
- Adding a work that previously caused an infringement can actually expand the permissible set by creating alternative generative paths, thereby reducing dependence on other works.
- The nature of creative innovation—whether it follows a light-tailed distribution (incremental) or a heavy-tailed distribution (paradigm-shifting)—critically determines the long-term persistence of infringement risk.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.