Gilad Abiri (Peking U Transnational Law) has posted “ML-Mediated Creativity” (Harvard Art Law Review Musings, June 2025) on SSRN. Here is the abstract:
This essay examines how machine learning systems fundamentally alter the dynamics of cultural innovation. Using anime’s post-war evolution as a case study, it argues that genuine creativity emerges from productive friction—the collision of different cultural traditions, generations, and artistic approaches. However, ML systems trained on existing cultural works create statistical averages that eliminate this generative friction, replacing dynamic cultural processes with algorithmic optimization. Current intellectual property frameworks cannot address this transformation because they treat cultural works as discrete objects rather than materials for creative play. The essay proposes two interventions: preserving “friction spaces” in educational institutions and regulating ML architecture to maintain distinct cultural lineages rather than collapsing them into optimized averages.