Intelligent Entertainment: Shaping Policies on the Algorithmic Generation and Regulation of Creative Works

Abstract

The virtual spaces where intellectual property (IP) meets artificial intelligence (AI) are shaping the future of work. Creative industry firms employ AI to plan, edit, and remix artistic and literary works, in use cases ranging from computer games and graphics to short stories, news articles, songs, screenplays, films, and beyond. Policy in this area is driven in significant part by lobbyists' and lawyers' moves to implement AI to control the content of creative works and online communications and intergovernmental efforts to enhance Internet copyright enforcement amidst floods of digital speech.

In the far future, intellectual property law may become less relevant as advances in computing and AI contribute to an abundance of creative works. In the near term, the copyrightability of AI-generated works and the governance of human expression by AI systems are coming to the fore. The significance of human creators could fade as either the authors or the users of generative adversarial networks and other AI implementations create and own expressive output on a previously unimagined scale. Copyright law may inhibit this process by classifying works such as those generated by OpenAI's GPT-3 model as potentially unlawful to create or publish without the permission of the writers, recording artists, and camera operators with whose works the models are trained. Alternatively, copyright and intellectual property in general might excuse and empower AI programming that helps people create using "data" without substantially replicating a specific "work" or "works."

The use of AI to enforce legal or social norms represents a growing market and will raise public policy concerns. Courts and legislators know that the ease and speed of digital communications create gaps in legal enforcement. As a consequence, governments and industry conjure up AI solutions for detecting and limiting the spread of illegal digital content. While some may welcome reductions in the difference between legal obligations and legal compliance, others warn that unreasonable and overbroad application of legal norms will harm innocent parties and public discourse. This essay therefore considers the case for limiting AI's role in enforcing intellectual property rights to protect Internet users from the erroneous flagging and removal of their original commentary and creative works. It concludes with some observations concerning the importance of human lawyers, judges, and jurors in governing fruits of the imagination. The essay introduces a symposium on AI and the future of intellectual property with a focus on big data and algorithms. The contributors consider AI as the defendant in a lawsuit, AI as a means to control the proliferation of IP rights and to promote the fair use of entertaining or informative works, AI as an engine that mines texts and images for fuel, and AI as a threat to the traditional freedoms that humans have crafted.

Keywords: artificial intelligence, computer-generated works, AI-generated works, machine learning, neural networks, safe harbors, DMCA, E-Commerce Directive, automated decision systems, content moderation, platform liability, Digital Single Market Directive

JEL Classification: K00, K23, K42, O31, O33, O34