COPYRIGHT AND CREATIVE ARTIFICIAL INTELLIGENCE (AI) SYSTEMS: A TWENTY-FIRST CENTURY APPROACH TO AUTHORSHIP OF Al-GENERATED WORKS IN THE UNITED STATES

Atilla Kasap
DOI: https://doi.org/10.31235/osf.io/gnyha
2021-06-28
Abstract:Complex Artificial Intelligence ("AI")-systems can generate original works as if they were human, yet with little or no human intervention. Nonetheless, the development of case law concerning copyright has implicitly centered on human creativity and originality, two qualities that a work must embody to be eligible for protection.' Moreover, copyright law has adapted over time, as technological innovations have confronted legislators and courts with new challenges in conception and application. Harmonizing the human role in copyright and technological innovations is therefore of the utmost importance if the United States' constitutional principle of furthering science is to be maintained. The difference between human creativity and computationalcreativity is the first and most crucial question that needs to be answered. This is because many Al processes are currently being employed to mimic human capacities. The state of the art in Al should be clarified to demonstrate the role that humans play in creative machines and the creative output of machines. Whether, as a matter of empirical fact, Al-systems are capable of all the creative capacities that humans possess is directly related to the following question: who is the author of Al-generated works? In other words, to whom are economic rights in authorship to be transferred to and who has standing before a court in the case of infringement issues and the like? This note is divided into five sections. The first section provides a brief definition, preliminary history, and summary of the latest developments of Al technology. This is followed by an in-depth investigation into the copyrightability of Al-generated works in the second section. Importantly, the second section will trace the locus of difference between human creativity and computational creativity and propose policy changes in the application of the notion of the originality which serves as a requisite for the allocation of authorship for copyright purposes. The third section will explore reasons to accept Al-generated works as copyrightable based on utilitarian grounds. Having thus established the grounds of copyright protection eligibility for Algenerated works, the fourth section will determine how authorship of these works is best attributed or recognized through a comprehensive analysis of the merits of seven plausible options: (i) the Al system itself as author; (ii) the programmer as author; (iii) the trainer as author; (iv) the user as author; (v) joint authorship; (vi) public domain or; (vii) the data proprietor. The final section proposes a solution to who should be accepted as the author of Al-generated works that better serves constitutional purposes and balances the many interests that are likely to arise between the various actors who develop creative machines or who lawfully obtain such machines.
What problem does this paper attempt to address?