The global intellectual property (IP) community has been abuzz with opinions on the opportunities and challenges presented by the development of artificial intelligence (AI) for quite some time now. These voices have only become more pronounced in the past months as AI has rapidly captured the imagination of people across industries and borders, especially since the public launch of OpenAI’s popular chatbot, ChatGPT, on 30th November 2022. This is not to say that other AI systems were far behind on technology before ChatGPT— in fact, ChatGPT used available technology which was also being used by other systems at the time— but it still marked a turning point in AI development. Firstly, it increased the accessibility of AI as it brought a sophisticated, generative, free-to-use, linguistic language model to the fingertips of the public at large. Secondly, ChatGPT used very large data sets for training with impressive neural networks and architecture. This significantly improved the results, leading to better contextual comprehension, the ability to deal with complex queries, and overall, a more conversational interaction. ChatGPT represented the first time that a large language model of its scale and ease of access showed the ability to produce human-like results.[i] The wide applicability soon became apparent— it could generate high-quality and coherent articles, poetry, fictional stories, news reports, and even computer codes with the assistance of prompts from users. Its prowess in vastly reducing the time and effort required to perform certain tasks has wide ramifications for various industries, and its versatility has allowed a wide variety of people from across the world to use the system, either out of curiosity or to make their work easier. The magnitude of users has also allowed ChatGPT to evolve and improve quickly due to network effects, leading to newer and better versions with further enhanced capabilities. The success of ChatGPT has led businesses to invest more in AI[ii] and the development of AI tools as well as innovation in the field has seen significant growth in 2023[iii]
ChatGPT is merely an example of generative AI, which are essentially deep-learning models capable of creating high-quality text, videos, images, and other content based on data they are trained on.[iv] It is important to note here that the output generated by such models is based on training data and, therefore, shares some resemblance with them. However, they are usually new/creative and not mere reproductions of the input.[v] Thus, for all means and purposes, generative AI creates something new; and while one could argue that the output is a mosaic of the inputs processed and transformed by the neural network, how is this significantly different from a human creator inspired by prior works? Now, many will (and are) taking exception to this perspective[vi] and legal challenges loom large upon generative AI as courts grapple with the legality of how generative AI models are trained and whether the similarity between the inputs and the output would amounts to infringement of various IP rights, especially copyright.[vii]
The rise of AI, generally, and improvements in generative AI, particularly, are transforming how people perform tasks. Early adopters are already benefitting from the use of AI by reducing costs and improving efficiency and quality. Businesses which adopt ChatGPT have shown better results, including in consumer satisfaction.[viii] Studies show that the use of ChatGPT in education has a positive impact on higher education[ix] and the teaching-learning process[x]. AI has proved useful in various fields, such as healthcare[xi] and scientific research[xii]. AI has also proved useful in creating original research in complicated fields of science, such as pharmaceutical development, provided it is guided by human intervention and data validation.[xiii] The effectiveness of AI in these fields is likely to grow manifold in the coming years. The ease with which users can now use simple prompts to create something or improve upon something reduces the time and effort of creation and improves the overall quality of the output. Computer codes which took months to write can now be written in days or even hours; even through prompts supplied by people who have limited knowledge of programming. Scientific data analysis or simulation of physical or biological phenomena can now be done faster and with lesser computing power. Scientific predictions can be made faster and more accurately through the use of AI systems. Thus, the scale and speed of technological development and societal transformation are likely to be exponentially higher than ever before in this new AI era.
In times when AI is increasingly being used to create, it is not surprising that courts are being compelled to decide the extent to which AI-generated or AI-assisted creations can be protected under extant IP laws. The most popular controversies in the field of patents, for example, revolve around the patent applications filed by Dr. Stephen Thaler in various jurisdictions, claiming patent protection for inventions allegedly created by the AI system named DABUS (Device for the Autonomous Bootstrapping of Unified Sentience). This has led to an interesting array of decisions by patent offices and judicial authorities across the world. In most jurisdictions, courts have broadly ruled that existing domestic patent laws do not allow for anyone but natural persons to be inventors. This includes Australia[xiv], the E.U.[xv], the U.K.[xvi], and the U.S.A.[xvii] In India too, Dr. Thaler’s patent applications have faced objections from the Patent Office citing similar reasons. The most prominent anomaly in this slew of judicial decisions denying patent protection to AI-created inventions is South Africa, which went ahead and granted a patent to DABUS in 2021, making it the first country to grant a patent to an AI system.[xviii] That aside, it seems most jurisdictions have determined that existing laws do not permit a patent to be granted to a computer system/AI. This is not surprising. Most patent laws were drafted at a time when the legislatures could not have foreseen a non-human intelligence capable of creation. However, while much is being said about how existing laws should be interpreted to deal with AI creations, these controversies will likely pale into insignificance in the long run, for there is no reason to expect that legislation will remain constant indefinitely. For instance, in India, the Parliamentary Standing Committee on Commerce, in its 161st Report[xix]recommended the amendment of The Patents Act, of 1970 and The Copyright Act, of 1957 to incorporate AI and AI-related inventions. The Committee also recommended the creation of a separate category of rights for AI and AI-related inventions and solutions.[xx] Similar arguments for amendment to existing laws are being made across the world. The real question, therefore, is not whether AI creations can be protected under existing laws. The question is whether they ought to be.
This inquiry will most certainly lead us back to the rationale for IP protection, and I submit, that it will necessitate treating different forms of creations differently. Firstly, we must distinguish those inventions and works created by humans with the assistance of AI (AI-assisted creation), and those created predominantly by an AI system with little or no human contribution (AI-generated creation). In the case of the former, where an AI system simply assists the human creator, thus, still requiring the human creator to be creative/innovative, copyright and patent protection should follow even under existing laws. This is the stand taken by the USPTO, which in a guidance document released in February 2024, has clarified that while AI-assisted inventions are not unpatentable, “the inventorship analysis should focus on human contributions, as patents function to incentivize and reward human ingenuity.”[xxi] The real issue lies in the latter category, where human contribution is negligible. For instance, I asked ChatGPT to “write a poem of IPR”. Within seconds, the chatbot produced a poem, starting with the following paragraphs:
In the realm of ideas, a dance of light,
Where thoughts take flight in the silent night.
Creativity's spark ignites the fire,
Invention's birth, a world to inspire.
Yet in this realm, a shadow looms,
Where theft of dreams becomes the tombs.
For here resides a sacred guard,
The sentinel of the intellectual yard.
Undoubtedly, the aforesaid paragraphs have some literary quality, and if they were to be written by a human being, copyright protection would follow. However, this has been written by ChatGPT and my contribution is limited to simply stating the topic on which the poem was to be written. In such a case, it would be unfair if I were allowed to obtain statutory protection over the poem, excluding others (and possibly even ChatGPT) from reproducing, adapting, or otherwise dealing with the work.
In such a case, where the role of AI in creation is significantly more than the role of the human being guiding the AI system, the question of the protection of creations is a daunting one. While it may be tempting to opine that all AI creations of this kind should be part of the open commons devoid of statutory protection from third-party appropriation, jumping to this conclusion without careful thought may be dangerous.
IP systems across the globe are justified largely under four theories: the utilitarian theory, the labour theory, the personality theory, and the property theory.[xxii] Amongst these, the utilitarian theory has, for all practical purposes, attained significant prominence. According to this theory, IP rights exist to maximize social welfare, and the limited monopoly granted by IP laws is an attempt to balance the need to incentivize intellectual endeavour and creation of useful works and inventions on one hand, with the public’s interest in accessing and using such creations on the other.[xxiii] If we stipulate the correctness or desirability of the utilitarian theory of IP, protection for AI creations must be granted if it is likely to encourage the use of AI to create useful things for society. If AI can help medical research, improve our existing technologies, mitigate climate change, and find answers to questions regarding our universe, very few would suggest the non-deployment of AI systems. After all, if the use of AI can find better treatments for cancer and speed up research in the field, would it be wise to discourage the use of AI by disqualifying such an invention from patent protection? If such inventions were to be disqualified, inventors would likely be encouraged to not use AI systems at all, or at least to their fullest capacity, to ensure their invention obtains patent protection enabling better commercial exploitation. This situation would not maximise social welfare and could slow down progress. But this is related to inventions. Would the scales remain unchanged in the case of works usually protected by copyright? I submit not. The reason for this is the stronger influence of other theories of IP in the field of copyrights. Copyrightable works are not only protected because of their utility to society but also because they are considered a form of personal expression (the personality theory) of the author. Thus, most copyright regimes provide moral rights to authors to claim attribution and prevent modifications to a work. These moral rights move beyond commercial considerations[xxiv] and even beyond societal utility as authors also have the right to withdraw their work (author’s right to retraction)[xxv] and thereby deprive society of any utility the work may possess. This is quite unlike patent law, which considers non-use as a ground for compulsory licensing and subsequent revocation.[xxvi] Thus, the law as it stands today, already treats inventions and copyrightable works differently, with the utilitarian theory having a much larger say in patent law than in copyright law.
Following the utilitarian theory, it would make sense to provide some form of patent protection to AI-generated inventions, as the society at large will benefit from the technical progress that this would entail. Of course, this has many practical challenges. For instance, if an AI system is a patentee, will it also sue for infringement? How can such a system negotiate licenses? Who owns the royalties and other remuneration earned out of the exploitation of a patent, and how are such funds used? The questions that arise are numerous. In all likelihood, a legislature which grants patent protection to AI-generated work will grant it to the person(s) who instructed the AI system and/or the person(s) who took responsibility and initiative for the invention (similar to a producer of a cinematograph under copyright law[xxvii]). This would clarify the issue of ownership and prevent the creators of AI systems from taking credit and obtaining protection for all that the AI system later creates.
At the same time, legislatures may not extend copyright protection to AI-generated works, limiting copyrights to human creativity. Now, it is true that this would disincentivize the creation of useful works through AI and increase the time and effort necessary to create stories, poems, paintings, etc., but the negative impact of such discouragement on society is limited. At the same time, such a legislative policy would continue to value human creativity and contribution, and financially support human creators. This is also crucial as generative AI already threatens the livelihood of much of the human workforce, including creators, and society is likely to face large-scale unemployment challenges as AI takes over several tasks which human beings perform today, or at least significantly reduces the amount of human labour necessary. Even under the labour theory, such an approach would be acceptable as the fruits of human labour will belong to the individuals who contributed their labour and creativity.
This difference in approach under patent law and copyright law will be crucial in ensuring the fruits of AI are enjoyed by society at large without shunning human skills and creativity. The exact manner in which this distinction plays out will, however, depend on a number of factors and will require extensive discourse spanning not only legal theory but economic and technical realities. Exponential growth in the field in the coming years and the development of Artificial General Intelligence may well change the paradigm. Therefore, this is by no means a complete guidebook to how IP laws must develop to deal with developments in AI technology, but merely a start as we begin to seriously consider the future of creation and creators.
[i][i] Marr, Bernard, A Short History of ChatGPT: How We Got To Where We Are Today, Forbes, (19 May 2023), available at: https://www.forbes.com/sites/bernardmarr/2023/05/19/a-short-history-of-chatgpt-how-we-got-to-where-we-are-today/?sh=41da50d1674f.
[ii] Gartner Poll Finds 45 Percent of Executives Say ChatGPT Has Prompted an Increase in AI Investment, Gartner, (3 May 2023), available at: https://www.gartner.com/en/newsroom/press-releases/2023-05-03-gartner-poll-finds-45-percent-of-executives-say-chatgpt-has-prompted-an-increase-in-ai-investment#:~:text=poll%20of%20more%20than%202%2C500,in%20pilot%20or%20production%20mode.
[iii] Rashi, Shrivastava, How ChatGPT and Billions in Investment Helped AI Go Mainstream in 2023, Forbes, (27 December 2023), available at:https://www.forbes.com/sites/rashishrivastava/2023/12/27/how-chatgpt-and-billions-in-investment-helped-ai-go-mainstream-in-2023/?sh=7a71e4ee7176.
[iv] Kim Martineau, What is Generative AI?, IBM Research Blog, (April 20, 2023), available at: https://research.ibm.com/blog/what-is-generative-AI.
[v] Gartner, “Generative AI”, Gartner IT Glossary, available at: https://www.gartner.com/en/information-technology/glossary/generative-ai.
[vi] Gil Appel et al, Generative AI has an Intellectual Property Problem, Harvard Business Review, 7 April 2023, https://hbr.org/2023/04/generative-ai-has-an-intellectual-property-problem.
[vii] See, for eg., Getty Images (US) Inc. v. Stability AI Ltd., [2023] EWHC 3090 (Ch); Michael M. Grynbaum & Ryan Mac, The Times Sues OpenAI and Microsoft Over A.I. Use of Copyrighted Work, (27 December 2023), available at: https://www.nytimes.com/2023/12/27/business/media/new-york-times-open-ai-microsoft-lawsuit.html; Min Chen, Artists and Illustrators Are Suing Three A.I. Art Generators for Scraping and ‘Collaging’ Their Work Without Consent, (24 January 2023), available at: https://news.artnet.com/art-world/class-action-lawsuit-ai-generators-deviantart-midjourney-stable-diffusion-2246770.
[viii] Georges Maalouf, The Effect of ChatGPT on Business Success, Int’l J. of Professional Bus. Rev. São Paulo, vol. 8, No. 12 (2023), pp. 1-19, available at: https://doi.org/10.26668/businessreview/2023.v8i12.4134.
[ix] Juan Dempere et al, The Impact of ChatGPT on Higher Education, Front. Educ., vol. 8 (2023), 1206936, available at: https://doi.org/10.3389/feduc.2023.1206936.
[x] Marta Montenegro-Rueda et al, Impact of the Implementation of ChatGPT in Education: A Systematic Review, Computers, vol. 12, no. 8(2023), p. 153, available at: https://doi.org/10.3390/computers12080153.
[xi] Fazal Khan, Regulating the Revolution: A Legal Roadmap to Optimizing AI in Healthcare, 25 MINN. J.L. SCI. & TECH. 49 (2023); Blake Murdoch, AI in Healthcare Is Coming, and We Need to Be Ready, Lawnow, vol. 43 (2019), p. 7.
[xii] A. Ghosh, How can artificial intelligence help scientists? A (non-exhaustive) overview, Artificial Intelligence in Science: Challenges, Opportunities, and the Future of Research, (OECD Publishing, 2023), available at: https://doi.org/10.1787/a8d820bd-en.
[xiii] Moe Elbadawi et al, The Role of Artificial Intelligence in Generating Original Scientific Research, Int’l J. of Pharmaceuticals, vol. 652 (2024), 123741, available at: https://doi.org/10.1016/j.ijpharm.2023.123741.
[xiv] Thaler v Commissioner of Patents, [2021] FCA 879 (Federal Court, Australia).
[xv] Opinion on Designation of inventor/DABUS, J 0008/20 (European Patent Office) (21 December 2021), available at: https://www.epo.org/en/boards-of-appeal/decisions/j200008eu1.
[xvi] Thaler v. Comptroller-General of Patents, Designs and Trade Marks, [2023] UKSC 49 (Supreme Court of the UK), available at: https://www.bailii.org/uk/cases/UKSC/2023/49.html.
[xvii] Thaler v. Vidal, 43 F.4th 1207 (Fed. Cir. 2022; cert. denied), available at: https://cafc.uscourts.gov/opinions-orders/21-2347.OPINION.8-5-2022_1988142.pdf.
[xviii] IPWatchdog, DABUS Gets Its First Patent in South Africa Under Formalities Examination, (29 July 2021), available at: https://ipwatchdog.com/2021/07/29/dabus-gets-first-patent-south-africa-formalities-examination/id=136116/.
[xix] India, Rajya Sabha, Parliamentary Standing Committee on Commerce, Report on Review of the Intellectual Property Rights Regime in India, (Report No. 161 of 2021-22), available at: https://files.lbr.cloud/public/2021-07/161_2021_7_15.pdf?VersionId=S01fCQEC5DzDqKNymsGgxal6YXmJbUwM.
[xx] Id. at 100-101.
[xxi] USPTO, Inventorship Guidance for AI-Assisted Inventions, 89 Fed. Reg., no. 30, 10043, at p. 10044, available at: https://www.govinfo.gov/content/pkg/FR-2024-02-13/pdf/2024-02623.pdf.
[xxii] William T. Fisher, Theories of Intellectual Property, Berkman Klein Center, 1987, available at: https://cyber.harvard.edu/people/tfisher/iptheory.pdf.
[xxiii] Id.
[xxiv] See, for eg., §57, The Copyright Act, 1957 (India); Amar Nath Sehgal v. Union of India, 2005 (30) PTC 253 (DEL) (“24. In the material world, laws are geared to protect the right to equitable remuneration. But life is beyond the material. It is temporal as well. Many of us believe in the soul. Moral rights of the author are the soul of his works. The author has a right to preserve, protect and nurture his creations through his moral rights… 26. Except for the 'divulgation or dissemination right' which perhaps is guided by commercial considerations, the other three rights originate from the fact that the creative individual is uniquely invested with the power and mystique of original genius, creating a privileged relationship between a creative author and his work.”)
[xxv] Amarnath Sehgal, Id. at para 25.
[xxvi] See, §84 & §85, The Patents Act, 1970 (India).
[xxvii] §2(uu), The Copyright Act, 1957 (India).