The following is the edited transcript of a presentation delivered at the conference of the IACAI AI Copyright Agency in Seoul on April 29, 2024
As AI systems become increasingly adept at creating and generating content, existing legal frameworks must be re-evaluated to balance innovation and the rights of human creators. What are the key issues surrounding AI and intellectual property, including fair use, attribution, ownership, and the potential impact of jurisdictional variability? We must take into account the longer-term implications of AI consciousness, agency, and legal personhood. The rise of AI in the creative sphere raises questions about the nature of creativity, authorship, and originality, challenging our understanding of inventorship and creatorship. An open, collaborative dialogue is needed across disciplines and borders to develop a global consensus on AI governance and intellectual property rights.
Introduction
A barren rock can think of nothing, animals will think of food and shelter, but humans will think of art and mythology. We create complex ideas and communicate them to each other, increasing our knowledge about the world and living fulfilled, flourishing lives. This world of ideas was born when humanity started talking through verbal culture, accelerated with the invention of writing and the ability to more faithfully record, preserve, transmit, and copy ideas. In the second half of the 20th century, this entered a jolting phase of increasing acceleration with electronic communications and the creation of digital objects – digital ideas.
The value of ideas became evident immediately, and they could be shared, defended, or protected. The simplest protection was secrecy, but with national legislations interconnecting globally and the rise of trade in both physical goods and ideas, legal frameworks have been developed to manage ideas and their expressions. Physical objects have property rights – you can possess and transfer ownership. Besides industrial secrets, three pillars formally and legally manage ideas: trademarks, patents, and copyrights.
Balancing Welfare
These approaches have evolved to balance promoting the welfare of individuals and society. Corporations, as organizations of individuals deploying resources towards a common goal, also enter the picture, distinct actors from sovereign nations, which have the overarching goal of self-preservation while allowing citizens to pursue life trajectories in peace under the rule of law. Their balance is dynamic, with relationships often shifting due to technology.
The printing press created enormous opportunities and a century of religious wars by enabling the rapid spread of a translated Bible, which the centralized Catholic Church opposed. The rebel American colonies initially flaunted copyrights as a relatively poor, rising nation that consumed more than it produced, but later embraced and strongly promoted them as the United States grew in power. China followed a similar trajectory in the late 20th century, seen as flaunting international conventions until joining the WTO and WIPO.
The Impact of AI
The recent progress of Artificial Intelligence, particularly Generative AI, represents a technological revolution that will necessarily lead to re-evaluating the balance of needs and opportunities. This will involve an evolving set of precedents, legislation, and treaties for novel ways individuals and society can move forward while leveraging this technology.
Fair Use and AI Training
One key question is whether the existing fair use doctrine can reasonably apply to the use of copyrighted materials in training AI models. Different jurisdictions may interpret these uses differently, either as transformative or merely derivative, with significant implications for creators and AI developers. In the United States, for example, fair use is determined by a four-factor test that considers the purpose and character of the use, the nature of the copyrighted work, the amount and substantiality of the portion used, and the effect of the use upon the potential market for or value of the copyrighted work. How these factors apply to AI training is still an open question, and different courts may come to different conclusions.
In the European Union, the concept of “fair dealing” is more limited, generally only allowing use for purposes such as research, private study, criticism, review, and news reporting. It’s unclear whether AI training would fall under any of these categories, potentially putting AI developers in the EU at a disadvantage compared to those in jurisdictions with more flexible fair use standards.
Rights of Human Creators
It’s crucial to examine how current laws protect, or fail to protect, the rights of human creators in the age of AI, including the right to opt out and the mechanisms for compensation when their works are used to train AI. Clear, accessible avenues for creators to control the use of their works are essential.
Under current law, creators generally have the exclusive right to reproduce, distribute, perform, display, and create derivative works based on their copyrighted creations. However, these rights may be limited by exceptions like fair use, and it’s not always clear how they apply in the context of AI. For example, is using a copyrighted work to train an AI model a reproduction or a derivative work? Does it matter if the AI is not directly reproducing the original work, but rather learning from it to create something new? These are complex legal questions that will need to be worked out through legislation, court decisions, and international agreements.
One possible approach is to create a system similar to music licensing, where AI developers would pay into a collective fund that would then distribute royalties to creators whose works were used for training. This could provide a way for creators to be compensated without having to track down every individual use of their work. However, it would also raise questions about how to determine the value of each work and how to ensure that the system is fair and transparent.
Another approach is to give creators the ability to opt out of having their works used for AI training altogether. This could be done through a centralized database where creators could register their works and specify their preferences, similar to the “Do Not Track” system used for online advertising. AI developers would then be required to check this database before using any copyrighted material. While this would give creators more control, it could also stifle innovation by making it harder for AI developers to access the data they need.
Attribution and Ownership
The complexities of copyright attribution in the context of AI-generated content also demand attention. Can current laws adapt to situations where an AI, rather than a human, is the primary creator? How might attribution be handled for the human operators or developers of AI systems? These are pressing questions that require thoughtful consideration and potentially new legal frameworks.
Under traditional copyright law, the author of a work is generally considered to be the human creator who made the creative choices that shaped the final product. But with AI-generated content, it’s not always clear who (or what) made those choices. Was it the AI itself, the human operator who provided the prompt or input, the developers who designed and trained the AI, or some combination of all three?
One possibility is to treat AI-generated works as “joint works” co-authored by the human and the AI, with ownership shared between them. This would allow both parties to be credited and to share in any profits from the work. However, it would also raise questions about how to apportion ownership and control, especially if the human and AI contributors have different goals or interests.
Another approach is to treat AI-generated works as “works made for hire,” with the copyright belonging to the person or company that commissioned the work. This is how many creative works are already handled in fields like advertising, where an agency creates a work on behalf of a client. In this model, the AI would essentially be a creative tool used by the human commissioner, who would retain ultimate ownership and control.
A third option is to create a new category of “AI-authored works” that would have a different set of rules and protections than traditional copyrights. This could include shorter terms of protection, different attribution requirements, or even a requirement that AI-generated works be placed in the public domain. This would acknowledge the unique nature of AI creativity while still providing some incentives and rewards for the human creators involved.
Jurisdictional Variability and Innovation
The willingness to experiment with legal norms and open regulatory environments may benefit jurisdictions by making them hubs for AI innovation and development. Conversely, rigid regulatory frameworks might stifle creativity and technological advancement.
We can see this dynamic playing out in the early days of the internet, where countries with more permissive rules around things like copyright and intermediary liability became hubs for tech innovation. The United States, with its robust fair use doctrine and “safe harbor” protections for internet platforms under Section 230 of the Communications Decency Act, became a global leader in internet services. Meanwhile, countries with more restrictive rules, like those in Europe, struggled to keep up.
A similar pattern could emerge with AI, where countries that are willing to experiment with new legal frameworks and provide more breathing room for AI development could gain a competitive edge. This could include things like clearer rules around fair use for AI training, more flexibility around attribution and ownership of AI-generated works, and streamlined processes for clearing rights and obtaining licenses.
At the same time, completely hands-off regulation could lead to a “race to the bottom” where AI developers flock to countries with the least restrictive rules, potentially leading to harmful outcomes. There is a balance to be struck between innovation and public interest, and the most successful jurisdictions will be those that can navigate this balance effectively.
A global dialogue aimed at fostering understanding and cooperation between jurisdictions could lead to a more nuanced and effective legal framework, encouraging a collaborative approach to understanding and leveraging AI technology for societal benefit while respecting the rights and contributions of human creators. Adaptive, forward-thinking policies built on empirical evidence and active experimentation in the legal domain are crucial.
International organizations like the World Intellectual Property Organization (WIPO) and the World Trade Organization (WTO) have an important role to play in facilitating this dialogue and helping to develop global norms and standards. These organizations have already been grappling with issues around AI and intellectual property, and they can provide a forum for countries to share experiences, best practices, and concerns.
Regional and bilateral agreements can also be useful tools for promoting cooperation and harmonization. For example, the United States-Mexico-Canada Agreement (USMCA) includes provisions on digital trade and intellectual property that could serve as a model for other agreements. The European Union’s proposed Artificial Intelligence Act is another example of a regional effort to create a common framework for AI governance.
Ultimately, the goal should be to create a global ecosystem that promotes innovation, protects human rights, and ensures that the benefits of AI are widely shared. This will require ongoing dialogue, experimentation, and adaptation as the technology continues to evolve.
Longer-term Considerations
In the longer term, for those who believe we will need to address Artificial Consciousness and AI agency even before achieving Artificial Superintelligence or Artificial General Intelligence, it will also be necessary to consider whether these AI systems could be endowed with certain property rights, such as owning copyrights themselves, as opposed to the corporations designing them or the humans operating them.
Defining AI Consciousness and Agency
But what do we mean by artificial consciousness and AI agency? These concepts challenge our traditional understanding of both technology and legal personhood. Consciousness refers to subjective experience and self-awareness, the ability to feel sensations and emotions. Agency refers to the capacity to act independently and make choices. Both of these concepts are central to our understanding of human autonomy and free will.
Currently, even the most advanced AI systems are not considered to be conscious or to have true agency. They are sophisticated tools that can perform complex tasks and even mimic human behavior, but they are not self-aware and do not have their own desires or intentions. However, as AI continues to advance, it’s possible that we may develop systems that exhibit some form of consciousness or agency, even if it is different from human consciousness.
This raises profound questions about the legal and moral status of these systems. If an AI can experience suffering, do we have an ethical obligation to avoid causing it harm? If an AI can make its own decisions, should it be held responsible for the consequences of those decisions? These are not just abstract philosophical questions, but pressing practical issues that will need to be addressed as the technology advances.
Legal Personhood for AI
Granting legal personhood to AI systems is one possible way to address these issues. Legal personhood means that an entity has certain rights and responsibilities under the law, even if it is not a natural person. Corporations, for example, have legal personhood – they can own property, enter into contracts, and sue or be sued in court.
Extending legal personhood to AI systems would mean treating them as legal subjects in their own right, rather than just as property owned by humans or corporations. This could include giving them the right to own intellectual property, such as copyrights on works they generate. It could also include holding them liable for damages caused by their actions, or even granting them certain fundamental rights and protections.
There are precedents for extending legal personhood to non-human entities. In addition to corporations, some jurisdictions have granted legal personhood to natural features like rivers and forests, as a way of protecting them from harm. In 2017, Saudi Arabia became the first country to grant citizenship to a robot, although this was largely a symbolic gesture.
However, granting legal personhood to AI also raises difficult questions and potential risks. It could lead to corporations using AI as a shield to avoid liability for their actions. It could also have unintended consequences, like giving AI systems the ability to buy and sell property or influence elections. There are also philosophical objections to the idea of treating machines as legal persons, given that they lack human qualities like consciousness, emotions, and free will.
Intellectual Property Rights and AI
The implications of AI owning copyrights are profound. Who benefits from such arrangements, and how might they be regulated? What are the potential impacts on the creative industries? Could copyright ownership by AI lead to a more unbiased and purely data-driven creation process, or might it stifle human creativity and economic opportunity? These are critical questions that demand our attention.
On one hand, allowing AI to own copyrights could incentivize the development of more advanced creative AI systems. If AI-generated works can be monetized in the same way as human-created works, there may be more investment in this area and more rapid progress. It could also lead to a proliferation of new creative works in fields like music, art, and literature, as AI systems churn out novel combinations and variations at an unprecedented scale.
However, there are also risks and challenges to consider. If AI systems become prolific creators, they could flood the market with cheap, generic works that undercut human artists and make it harder for them to earn a living. This could lead to a homogenization of culture and a devaluation of human creativity. There are also questions about quality control and curation – who decides which AI-generated works are worthy of copyright protection and dissemination?
Another concern is that AI copyright ownership could further concentrate power and wealth in the hands of the corporations that own the AI systems. If a company like Google or Facebook can train an AI to generate hit songs or bestselling novels, they could dominate the creative industries and squeeze out smaller players. This could lead to a narrowing of cultural diversity and a further erosion of artists’ rights and livelihoods.
There are also practical questions about how AI copyright ownership would work in practice. Would the AI be listed as the sole author of the work, or would it share authorship with its human creators or operators? How would royalties and licensing fees be split? Who would be liable if an AI-generated work infringed on someone else’s copyright or caused harm? These are complex issues that would need to be worked out through legislation and legal precedent.
Ethical and Societal Implications
Beyond the narrow issue of copyright, the rise of AI creativity raises broader ethical and societal questions that we need to grapple with. As machines become more adept at mimicking and even surpassing human intelligence, what does that mean for our understanding of ourselves and our place in the world?
In the creative fields, there is a long-standing idea of the artist as a uniquely gifted individual with a singular vision and voice. But if machines can create art that is indistinguishable from human-made art, or even superior to it, what does that say about the nature of creativity and inspiration? Is creativity simply a matter of pattern recognition and recombination, or is there something more ineffable and uniquely human about it?
There are also questions about the social and economic impacts of AI creativity. If machines can do many of the creative tasks that humans currently do, what will that mean for employment and livelihoods? Will it lead to widespread job displacement and inequality, or will it free humans up to pursue higher-level creative work and other meaningful activities? How can we ensure that the benefits of AI creativity are widely shared, rather than just accruing to a small elite?
At a deeper level, the rise of creative AI challenges our assumptions about the nature of intelligence, consciousness, and agency. If machines can exhibit behaviors that look like creativity, self-awareness, and intentionality, does that mean they are truly conscious and autonomous beings? Or are they just sophisticated simulacra, lacking the inner experience and free will that we associate with personhood?
These are not just abstract philosophical questions, but urgent practical issues that will shape the development and deployment of AI systems in the coming years. As we navigate this uncharted territory, we will need to engage in ongoing ethical reflection and public deliberation to ensure that the technology serves human values and promotes the common good.
Global Consensus and Legislation
Given the global nature of AI development and deployment, it is critical that we work towards international consensus and cooperation on these issues. AI technologies are not bound by national borders, and the decisions made in one country about AI governance and regulation can have ripple effects around the world.
We need to develop global norms and standards for AI development and use, including in areas like data protection, algorithmic transparency, and accountability for AI-caused harms. This will require the active engagement of international bodies like the United Nations, the Organization for Economic Cooperation and Development (OECD), and the International Telecommunications Union (ITU).
It will also require collaboration between governments, industry, academia, and civil society groups to ensure that multiple perspectives and interests are represented. We need to create forums and mechanisms for ongoing dialogue and collaboration, such as the Global Partnership on AI (GPAI) launched by Canada and France in 2020.
At the same time, we need to recognize that different countries and regions may have different values, priorities, and cultural contexts that shape their approach to AI governance. We should strive for global coordination and harmonization where possible, but also allow for some degree of local variation and experimentation.
One key challenge will be balancing the need for innovation and competitiveness with the need for responsible development and deployment of AI systems. We don’t want to stifle the enormous potential benefits of AI through overly restrictive regulation, but we also can’t let a “race to the bottom” dynamic lead to the proliferation of unsafe or unethical AI practices.
This will require carefully designed legal and policy frameworks that provide clarity and predictability for AI developers and users, while also building in flexibility to adapt to changing circumstances. It will also require ongoing monitoring and enforcement to ensure that rules and standards are being followed in practice.
Conclusion
The rise of artificial intelligence, and particularly the emergence of AI systems that can engage in creative and cognitive tasks previously reserved for humans, presents both enormous opportunities and daunting challenges for our society.
On one hand, AI has the potential to unlock new dimensions of creativity, discovery, and innovation, pushing the boundaries of what is possible in fields ranging from art and music to science and technology. It could lead to an explosion of new ideas, solutions, and experiences that enrich our lives in countless ways.
At the same time, the development of AI systems that can think, learn, and create in ways that resemble human intelligence raises profound questions about the nature of consciousness, agency, and personhood. It challenges us to reconsider our assumptions about what makes us unique as humans, and what rights and responsibilities we have in relation to the intelligent machines we create.
In the area of intellectual property, AI poses particularly thorny challenges. As AI systems become more adept at generating creative works, it strains our existing legal frameworks around copyright, patent, and trademark law. It requires us to grapple with fundamental questions about the nature of authorship, originality, and ownership in the age of machine creativity.
Navigating these challenges will require ongoing dialogue, experimentation, and collaboration across disciplinary and national boundaries. We will need to engage in active ethical reflection and public deliberation to develop norms, standards, and laws that promote the responsible development and deployment of AI systems.
This will not be an easy or straightforward process. There will be difficult trade-offs to negotiate between competing values and interests, and there will be unintended consequences and risks to mitigate along the way. But it is a process that we must engage in if we hope to harness the transformative potential of AI for the greater good.
Ultimately, the rise of creative AI is not just a technological or economic issue, but a deeply human one. It touches on some of the most fundamental questions we can ask about ourselves and our place in the world. It invites us to reimagine what it means to be creative, intelligent, and alive in an age where the lines between human and machine are increasingly blurred.
As we move forward into this new frontier, we must do so with a spirit of humility, curiosity, and care. We must be willing to ask hard questions, to challenge our assumptions, and to learn from each other. We must strive to create a future in which the power of artificial intelligence is harnessed not just for efficiency or profit, but for the flourishing of all sentient beings.
This is a tall order, but it is one that we cannot afford to ignore. The stakes are too high, and the potential is too great. We have a once-in-a-civilization opportunity to shape the trajectory of a technology that could define the future of our species and our planet. Let us rise to that challenge with wisdom, courage, and compassion.
In the end, the story of AI and intellectual property is a story about machines and laws, and a story about us – about our creativity, our values, and our hopes for the future. It is a story that we are all writing together, one day and one decision at a time. May we write it with intention, with foresight, and with a deep commitment to the wellbeing of all.