It’s try or die for the AFF, anything else risks stifling innovation

The main claim of this paper is that current copyright law in the US provides a durable framework that can and should manage a Guetta-like case. However, such an accommodation does raise several novel issues and questions, first, about the relationship between humans and machines in the creative process and, secondly, about the shifting character of the network of relevant stakeholders implicated in this process. Our main argument here seeks to clarify the notion of an AI-generated work – as used in many provocative recent headlines. Our point is to emphasize that there is no such thing as purely AI-generated work, at least not yet, and that there is always a human/author in the loop of the creative. The idea of “AI-generated work” is overly simplistic and potentially misleading and does a disservice to those seeking to experiment creatively with generative AI systems. Even more so, the notion of “AI-generated work” implies that AI can act completely autonomously in generating AI content - something that is not factually true. As we try to show in the paper using the Guetta example, human/author involvement usually occurs either at the input (prompting) or output stage; and the creation of somewhat creative works usually requires human intervention and refinement of interim iterations of a work. As such, it is important that new concepts and terminology are developed that more accurately characterize the forms of creativity emerging in our rapidly evolving digital culture. We prefer the idea of “co-creation of works with AI tools” as this type of expression seems to capture the hybrid character of the creative process that is happening in the generative AI space right now. We believe that our effort to unfold the five dimensions of human authorship in creating works by using generative AI tools (i.e., conception, prompting, generation, refinement and deployment) can shed new light on the legal concept of human authorship which during the 19th and 20th centuries increasingly shifted focus from the author to the work.72 In this context, we might borrow Lawrence Lessig’s term “remix” and think about how it might be deployed in an age of generative AI. Crucially, this concept points to an essential feature of digital culture, namely the centrality of quotation and iteration, and our struggles with imagining the future and developing new artistic forms.73 Perhaps, the only or, at least, best response to a culture where we struggle to imagine the future is remixing and recycling – a self-conscious, often ironic, revisiting and repackaging of the past – and in that respect AI tools offer a contemporary and timely means to navigate the impasse of the current historical moment. From a legal perspective, our purpose here has been to suggest that the low threshold of creativity in US copyright law is easily met in a case such as David Guetta’s. Nevertheless, further research is needed to analyse more specific use cases involving music, voice, multiple generative art segments, as well as the use of specific applications across these different media. As such, we need to have an open and enlightened discussion about hard or fringe cases, where there may well be far less human creativity than discussed here, and a more restrictive approach may be justified. Our conclusion is that when it comes to genuine creators, who are using AI systems as innovative and unique tools to push the boundaries of their creativity, copyright law remains one of the primary mechanisms to facilitate them in this pursuit. Exaggerated claims about the role and capacity of AI in creation and possible flaws in the training data of such models should not be used as a mechanism or justification to curtail the discussion, stifle creativity, and limit opportunities to use AI tools in pushing digital culture forwards and in new interesting directions.