[Paper Review] Art and the science of generative AI: A deeper dive

Ziv Epstein, Aaron Hertzmann+12 more|arXiv (Cornell University)|Jun 7, 2023

Law, AI, and Intellectual Property14 citations

TL;DR

A multidisciplinary white paper examining how generative AI reshapes aesthetics, law, labor, and the media ecosystem, arguing it’s a new art medium with distinct affordances and policy implications.

ABSTRACT

A new class of tools, colloquially called generative AI, can produce high-quality artistic media for visual arts, concept art, music, fiction, literature, video, and animation. The generative capabilities of these tools are likely to fundamentally alter the creative processes by which creators formulate ideas and put them into production. As creativity is reimagined, so too may be many sectors of society. Understanding the impact of generative AI - and making policy decisions around it - requires new interdisciplinary scientific inquiry into culture, economics, law, algorithms, and the interaction of technology and creativity. We argue that generative AI is not the harbinger of art's demise, but rather is a new medium with its own distinct affordances. In this vein, we consider the impacts of this new medium on creators across four themes: aesthetics and culture, legal questions of ownership and credit, the future of creative work, and impacts on the contemporary media ecosystem. Across these themes, we highlight key research questions and directions to inform policy and beneficial uses of the technology.

Motivation & Objective

• Understand how language and AI framing shape perceptions of generative AI in art and policy.
• Identify distinct affordances of generative AI as a new artistic medium.
• Explore cultural, legal, economic, and media ecosystem implications of AI-generated art.
• Propose research questions and directions to guide policy and responsible use.

Proposed method

• Synthesis of interdisciplinary perspectives from culture, law, economics, and media studies.
• Critical analysis of perceptions, authorship, and meaning in AI-generated art.
• Examination of data sourcing, ownership, and credit in legal contexts.
• Assessment of labor market impacts and the labor economics of creative work.
• Discussion of media ecosystem risks and trust, including provenance, watermarking, and misinformation.

Experimental results

Research questions

• RQ1 What are the linguistic and conceptual traps around AI that shape perception and credit in AI-generated art?
• RQ2 How should meaningful human control be defined and achieved in generative AI art systems?
• RQ3 What distinct cultural aesthetics and labor implications arise from generative AI as a new art medium?
• RQ4 What legal frameworks govern training data and outputs for ownership and compensation?
• RQ5 How do provenance, watermarking, and platform dynamics influence trust in AI-generated media?

Key findings

• Generative AI is a new medium with its own affordances, not an outright end to art.
• Perceptions of authorship and credit are shaped by human control, intent, and disclosure.
• Training data origin and output ownership present complex copyright and ethical questions.
• Generative tools can both threaten and complement creative labor, potentially changing employment and productivity.
• The media ecosystem faces challenges like impersonation, misinformation, and trust, mitigated by provenance, authentication, and forensic methods.