Between Copyright and Computer Science: The Law and Ethics of Generative AI (2403.14653v2)

Abstract: Copyright and computer science continue to intersect and clash, but they can coexist. The advent of new technologies such as digitization of visual and aural creations, sharing technologies, search engines, social media offerings, and more challenge copyright-based industries and reopen questions about the reach of copyright law. Breakthroughs in artificial intelligence research, especially LLMs that leverage copyrighted material as part of training models, are the latest examples of the ongoing tension between copyright and computer science. The exuberance, rush-to-market, and edge problem cases created by a few misguided companies now raises challenges to core legal doctrines and may shift Open Internet practices for the worse. That result does not have to be, and should not be, the outcome. This Article shows that, contrary to some scholars' views, fair use law does not bless all ways that someone can gain access to copyrighted material even when the purpose is fair use. Nonetheless, the scientific need for more data to advance AI research means access to large book corpora and the Open Internet is vital for the future of that research. The copyright industry claims, however, that almost all uses of copyrighted material must be compensated, even for non-expressive uses. The Article's solution accepts that both sides need to change. It is one that forces the computer science world to discipline its behaviors and, in some cases, pay for copyrighted material. It also requires the copyright industry to abandon its belief that all uses must be compensated or restricted to uses sanctioned by the copyright industry. As part of this re-balancing, the Article addresses a problem that has grown out of this clash and under theorized.

Between Copyright and Computer Science: The Law and Ethics of Generative AI

The paper "Between Copyright and Computer Science: The Law and Ethics of Generative AI" by Deven R. Desai and Mark Riedl explores the intricate interaction between the legal construct of copyright and the technological progress of generative AI, particularly focusing on the deployment and development of LLMs. As advancements in AI increasingly rely on vast datasets, often comprising copyrighted material, this intersection raises complex legal and ethical challenges.

Core Argument

The authors assert a nuanced stance on fair use, contending that access to copyrighted materials is not universally sanctioned for AI training purposes, especially if obtained under questionable conditions. Nevertheless, the scientific imperative for accessible, large datasets is integral to AI progress. Thus, Desai and Riedl propose a balanced approach requiring both the computer science community and copyright holders to adapt their expectations and practices.

Legal and Ethical Tensions

The paper critically assesses how recent AI developments, particularly LLMs, have heightened the tension between copyright laws and the need for large-scale data. It challenges the assumption that fair use law automatically permits unrestricted access to copyrighted materials for AI training and highlights the inadequacy of legal doctrines in addressing non-consensual data scraping and the consequent model outputs. The discussion around OpenAI's practices and its commercial pivot underscores the disparity between academic and commercial AI deployment, exacerbating the legal scrutiny over the data's provenance.

Practical Solutions and Implications

Desai and Riedl propose a multidimensional solution to harmonize these competing interests:

1. Licensing and Access Framework: They suggest rethinking access to copyrighted materials via more structured frameworks that permit fair use for AI development, potentially envisioning centralized repositories managed under collaborative public-private partnerships. Libraries and existing digital projects like Google Books could serve as secure data sources for academic research.
2. Differentiated Treatment for Academic and Commercial Entities: Recognizing the difference in usage intent, they argue for delineating clearer boundaries between academic and commercial AI activities. This calls for a careful balancing act where academic pursuits are protected under fair use doctrines, whereas commercial entities might be held to stricter standards, particularly concerning model outputs.
3. Output Regulation and Filtering: AI systems need to incorporate mechanisms for filtering and limiting possibly infringing outputs. The paper emphasizes the importance of distinguishing between the input datasets' legality and the resulting potential for infringing outputs, advocating that responsibility be placed on the latter rather than the former.
4. Policy Reform and Fair Compensation: The authors discuss the establishment of compensatory schemes akin to those in music licensing to address copyright holder concerns and ensure fair remuneration without stifling innovation.

Theoretical and Future Considerations

The paper provides a meticulous analysis of the theoretical implications of AI research practices within the framework of copyright law, warning against an overly restrictive legal atmosphere that could hinder scientific progress. It suggests the necessity for updated legal interpretations that reflect the realities of modern AI technologies. Additionally, the proposal of a centralized data access system demonstrates forward-thinking in balancing data needs with intellectual property rights.

Conclusion

"Between Copyright and Computer Science: The Law and Ethics of Generative AI" navigates the complex terrain between technological advancement and legal frameworks, presenting a balanced view that encourages mutual adaptation and enhanced ethical standards. As AI continues to evolve, this paper offers critical insights into developing sustainable legal infrastructures that support innovation while respecting intellectual property rights, setting a precedent for future discourse in AI ethics and law.