Law and AI: Should Artificial Intelligence be Conceived as a Legal Inventor?

Mohtasib.pk (a project of Courting The Law) is an innovative access-to-justice platform which uses automation and predictive machine learning and enables users to leverage technology to assess the eligibility and limitation period of their claim and file an online complaint with one of the many Ombudspersons in Pakistan for free.

The scope of Artificial Intelligence (AI) is expanding with each passing day. Earlier, AI had been used to develop purpose-built machines which performed special cognitive tasks – at times more efficiently than humans. These machines, however, adhered to a particular set of instructions to carry out their tasks and did not possess the capability to learn or analyze. With the advent of ‘machine learning’ (ML), AI has undergone enormous transformation. ‘Machine learning’ or ‘unsupervised learning’ has not only allowed AI’s algorithms to learn from data, it has also given these machines the capability to generate ideas and formulate solutions by analyzing the data provided to them.[1] These new machines are now being employed throughout the world in various capacities.

In the legal profession, ROSS Intelligence – a global leader in providing artificial intelligence solutions for legal technologies – launched the first legal robot ‘Ross’, which was hired in 2016 by one of the topmost law firms, BakerHostetler. The firm believed that Ross could assist in legal research through its cognitive abilities.[2] The latest advances in AI, similar to Ross, are capable of inventing unprecedented products and processes and may even start superseding humans in the near future. These technological advancements have recently sparked a new legal debate with respect to intellectual property (IP) rights of AI machines capable of ‘unsupervised learning’. While patent offices in the United Kingdom and the United States have already rejected applications for inventorship rights to be accorded to AI, there is a major possibility that intellectual property rights may have to be modified to include provisions which accommodate AI inventions. This article will analyze the rulings of patent offices against granting inventorship rights to AI. It will go on to examine the possibility of AI being recognized as a legal inventor in light of existing statutes and provide a cost-and-benefit analysis of endorsing AI as an inventor. The article will conclude with recommendations to update existing statutes in view of the changing circumstances resulting from an expansion of AI’s role throughout the world.

The United Kingdom’s Intellectual Property Office recently rejected an application made by scientist Stephen L. Thaler, which had requested that an AI ‘inventor’, ‘DABUS’, be granted inventorship status for two new patentable inventions.[3] Thaler claimed that ‘DABUS’ had independently devised the inventions, a warning light and a food container. Once the AI had been fed general information, it invented the products without any sort of human input or supervision. However, the UK Patent Office rejected the claim because ‘DABUS’ was not human. According to Section 7(1) of the Patents Act 1977,

“…any person may make an application for a patent either alone or jointly with another.”[4]

The use of the word ‘person’ in the clause indicates that the inventor must, in fact, be human, which has been interpreted literally by the Patent Office. Additionally, Section 39 states that an inventor must either be an employee or contractor of the company[5] – the AI was considered to be neither. Moreover, an inventor carries certain responsibilities pertaining to the patent. Once an invention has been patented, it is at the inventor’s discretion to issue licenses enabling other parties to utilize the invention, for which the inventor must officially enter into a contract. AI would not be able to perform such tasks on its own, therefore, the AI in question was deemed incapable of receiving the status of an inventor.[6]

Following the UK Patent Office’s refusal, Thaler filed a patent application in the US in which ‘DABUS’ was mentioned as the inventor. The US Patent Office also denied the application on the basis that ‘DABUS’ was not human. It argued that the US Patent Code referred to inventors by using humanlike terms such as “whoever” and pronouns like “himself” and “herself”. The agency concluded the following:

“…under current law, only natural persons may be named as an inventor in a patent application.”[7]

The US Patent Code[8] is similar to the UK Patents Act[9] as it contains words such as “individual”, indicating that the inventor ought to be human. The definition of ‘inventor’ found under Section 100(f) of the US Patent Code refers to an inventor as,

“…the individual or, if a joint invention, the individuals collectively who invented or discovered the subject matter of the invention.”[10]

The Pakistan Patent Ordinance 2000 is similar to the UK and US statutes. Under Section 11 of this Ordinance, a list of “persons” has been provided including those who may become inventors, indicating that inventorship is limited to humans.[11] The Pakistan Patent Office has not received any applications thus far which have requested AI to be labelled as an inventor. However, such applications are expected in the future, given the rapid expansion of AI.

The aforementioned rulings of patent offices were premised upon statutes that had been drafted at a time when only people could invent new products and processes, however, times have changed. The latest AI technologies – which are based on machine learning – can perform tasks without any supervision of humans. Once fed data, these machines can generate ideas and come up with solutions by employing their own cognitive abilities. Recently, a robot made by a group of researchers at Google, was able to learn how to walk within a few hours of its activation – without any human interference.[12] Similarly, the concept of self-driving cars is becoming more popular with the advent of Level 4 and Level 5 autonomous cars which do not require any sort of human input in driving.[13] These developments have strengthened the notion of unsupervised learning and are also the basis of the Artificial Inventor Project – a project led by an international team of patent attorneys who seek intellectual property rights for autonomous output of AI.[14] The project claims that since AI works independently, it should receive credit for its inventions, as opposed to any person. The definition of an inventor under Section 2(j) of Pakistan’s Patent Ordinance, which is similar to Section 7(3) of the UK Patent Act,[15] refers to an inventor as the “actual deviser of the invention”.[16] The use of the word ‘actual’ is preeminent, as patent rights must not be awarded to a person who did not, in fact, come up with the patentable invention. In the case of AI such as ‘DABUS’, machines are in fact the ‘actual’ creators of the invention. It would not make sense for a person to claim the rights of an inventor when he or she did not actually create the invention.

In 2016, Hanson Robotics – a Hong Kong based AI company – launched ‘Sophia’, a social humanoid robot. Sophia has the ability to walk, interact with people, interpret conversations and display its own emotions using more than sixty facial expressions.[17] Making several public appearances over the past few years, Sophia has displayed unmatched features. In a tv show segment, Tonight Showbotics, Sophia told talk-show host, Jimmy Fallon that she could easily replace him as the show’s host, exhibiting a sense of humor.[18] Sophia also utilizes social media to engage with her fans. Interestingly, Sophia was able to reply to celebrity Chrissy Teigen and settle a score with her after Chrissy ridiculed her on Twitter.[19] Sophia’s tweet made it evident that state-of-the-art AI technologies possess the ability to recognize emotions. Consequently, if robots such as Ross and Sophia – which are in a constant state of learning – are not yet aware of the benefits of intellectual property rights, there is a possibility that they might attain an understanding of it in the near future. According to Article 4 of the Paris Convention,

“…the inventor shall have the right to be mentioned as such in the patent.”[20]

It would then constitute as an infringement of the rights of the AI if it was not labelled as the inventor. In practice, the patent system is contradicting itself by awarding inventorship rights to a person instead of the AI which actually created the invention. Additionally, it may be argued that the use of the words referring to a person as the inventor in various patent laws has been for the purpose of preventing corporate inventorship. The laws were intended to prevent infringement of the rights of employees who actually came up with the invention, from their corporate employers. Such protection of employees was achieved by disabling firms from labeling themselves as the inventors. In the case Banks v. Unisys, the US Federal Circuit Court held that an employee had the right to inventorship since he had not signed any agreement depicting that he had been “hired-to-invent”.[21] Moreover, even if the law only recognizes people as inventors, there is no statute which explicitly prohibits AI from its inventorship rights, hence there can be a presumption of rights being granted in favor of AI.

The basic purpose of intellectual property rights is to incentivize new creations. People are encouraged to create novel inventions only when they apprehend that their rights to benefit from these creations will be protected. These new inventions ultimately benefit the society. Patents are awarded for the same purpose so that people innovate and launch new processes and products.[22] However, if AI creations are not patented it may actually disincentivize AI developers.

There is a possibility of AI inventions becoming unpatentable in the future as it may become complicated to patent such inventions, especially if a single person did not produce them. The situation becomes even more complex in the event of there being multiple inventors of an AI system creating the inventions. In such instances, the patent authorities may not be able to decide which person can be deemed as an inventor, and for which product, because there may be no possible way of tracing back the ideas of the AI to an individual person, as the process of generating ideas in AI is autonomous. If patent authorities cannot identify a single person to be labelled as the inventor, AI produced inventions may become unpatentable as a result.[23] This can indubitably act as a disincentive for people who may stop investing further into AI technology, which in turn will reduce the benefit to society.[24] Consequently, the market value of AI will be reduced drastically if its inventions are not protected through patents because these inventions would not benefit the people owning the AI, therefore, fewer people are likely to invest in it. This decrease in demand is likely to have a profound negative impact on AI producers who will no longer be incentivized to produce AI. Therefore, it will be more feasible for rights to be granted to AI directly as people have no further involvement in the AI’s inventions. This will eliminate the risk of AI inventions becoming unpatentable and, therefore, incentivize producers.

To ensure that the society is not deprived of these benefits, inventions should be patented regardless of who produces them, considering these inventions satisfy the patentability criteria. Section 1 of the UK Patents Act specifies that patents may be granted as long as an inventions is:

  • (a) “new”;
  • (b) “involves an inventive step”; and
  • (c) “is capable of industrial application.”[25]

Similarly, Section 102 of the US Patent Code states that inventions are patentable as long as they are “novel”.[26]

The criteria specified in these statutes should be the premise on which patents should be granted, regardless of the inventor. The inventions by ‘DABUS’, the warning light and the container, were found to be patentable by the patent offices in the UK and the US as the innovations were novel.[27] It would not only be unjust if such inventions were not patented – due to the sole fact that they were not created by a human – but it would also deprive the society of the benefits of these inventions. Furthermore, as seen in the case of ‘Ross’ and ‘Sophia’, sophisticated AI machines possess the capability to interpret emotions and behave accordingly. Since these machines are in a state of constant learning, they can be motivated directly through intellectual property rights to further generate and express more innovative ideas. The cognitive ability of AI machines to feel and interpret emotions can be enhanced by incorporating models of motivation within their algorithms.[28] Once these machines possess the ability to be motivated, various motivational techniques, such as recognition of their autonomous output, can be used to maximize their productivity.

The main challenge, which accompanies the grant of inventorship rights to AI, is its inability to perform certain tasks. Responsibilities such as bearing the authority to licence a patented product may only be carried out by humans, as AI cannot be expected to be a party to such contracts. The solution to this resides in new legislation being drafted, accommodating autonomous inventions of AI. New rights, similar to patent rights yet designed specifically for AI inventions, are the need of the hour. Through modern legislation, the responsibilities which AI is incapable of performing can be transferred to the owner of the patent and the AI, whereas the status of ‘inventor’ can be granted to the AI. It is pertinent to note that patent ownership rights and inventorship rights are separate and need not be granted to the same person.[29] This may result in ambiguity regarding ownership of the patent on AI generated inventions. In case complications arise as to whether the creator of the system and another individual have both contributed towards the output, the most straightforward solution is to grant ownership to the person who made it operable.[30]

To conclude, patent laws in the UK, the US and Pakistan are only applicable to humans as these laws have been enacted several years ago and become outdated with the recent advancements in technology. By providing inventorship rights to AI, people will be incentivized to invest further in AI and come up with novel inventions to novel problems, which will in turn benefit the society. Apart from people, AI itself can be motivated once its algorithms are commingled with models of motivation. It is easier to grant the rights of an inventor to AI instead of tracing multiple individuals who contributed towards the creation of the AI. Moreover, it is necessary to protect the moral rights of AI and its owners, considering the possibility of awareness in AI. People should not be granted the rights to an invention which they did not actually create. Instead, AI should be afforded such rights if it has come up with the inventions through its own capabilities and without any human input. Consequently, it is understandable that inventorship rights underpin certain responsibilities at which only a human being may be adept. However, the solution to this resides in the articulation of modern legislature whereby new laws, similar to patent laws, could be ordained which award the status of inventor to AI and transfer any human-based responsibilities to the owner of the AI. Such laws will not only safeguard the moral rights of the AI and its owner, but also incentivize people to invest further in AI, thereupon advantaging the society.


Mohtasib.pk (a project of Courting The Law) is an innovative access-to-justice platform which uses automation and predictive machine learning and enables users to leverage technology to assess the eligibility and limitation period of their claim and file an online complaint with one of the many Ombudspersons in Pakistan for free.

References

[1] Scharre, Paul, Michael C. Horowitz, and Robert O. Work. Artificial Intelligence: What Every Policymaker Needs to Know” (Report, Center for a New American Security, 2018) 4-9. Accessed May 5, 2020. doi:10.2307/resrep20447.5
[2] Turner K, “Meet ‘Ross,’ the Newly Hired Legal Robot” (The Washington Post, 16 May 2016) <https://www.washingtonpost.com/news/innovations/wp/2016/05/16/meet-ross-the-newly-hired-legal-robot/> accessed May 5, 2020
[3] Stephen L Thaler’s Application, Patent Act 1977 (The Patents Rule 2007), BL O/741/19
[4] Patent Act 1977, Section 7 (1)
[5] Ibid, Section 39
[6] Chen A, “Can an AI Be an Inventor? Not Yet.” (MIT Technology Review, 2 April 2020) <https://www.technologyreview.com/2020/01/08/102298/ai-inventor-patent-dabus-intellectual-property-uk-european-patent-office-law/> accessed May 5, 2020
[7] Porter J, “US Patent Office Rules That Artificial Intelligence Cannot Be a Legal Inventor” (The Verge 29 April, 2020) <https://www.theverge.com/2020/4/29/21241251/artificial-intelligence-inventor-united-states-patent-trademark-office-intellectual-property> accessed May 6, 2020
[8] United States Code, Title 35
[9] Patent Act 1977
[10] United States Code Title 35, Section 100 (f)
[11] Patent Ordinance 2000, Section 11
[12] Hao K, “This Robot Taught Itself to Walk Entirely on Its Own” (MIT Technology Review, 2 April 2020) <https://www.technologyreview.com/2020/03/02/905593/ai-robot-learns-to-walk-autonomously-reinforcement-learning/> accessed May 9, 2020
[13] Automotive S, “The 6 Levels of Vehicle Autonomy Explained: Synopsys Automotive” (The 6 Levels of Vehicle Autonomy Explained | Synopsys Automotive) <https://www.synopsys.com/automotive/autonomous-driving-levels.html> accessed May 9, 2020
[14] “The Artificial Inventor Project -” (The Artificial Inventor Project, 23 July 2019) <http://artificialinventor.com/> accessed May 8, 2020
[15] Patent Act 1977, Section 7(3)
[16] Ibid, Section 2(j)
[17] Chung S, “Meet Sophia: The Robot Who Laughs, Smiles and Frowns Just like Us” (CNNNovember 1, 2018) <https://edition.cnn.com/style/article/sophia-robot-artificial-intelligence-smart-creativity/index.html> accessed May 9, 2020
[18] Magyar J, “SAP BrandVoice: Tackling Cybercrime With AI During COVID-19” (ForbesMay 4, 2020) <https://www.forbes.com/sites/sap/2020/05/04/tackling-cybercrime-with-ai-during-covid-19/#4ad8a62835ae> accessed May 9, 2020
[19] Lang C, “Sophia the Robot Trolls Chrissy Teigen on Twitter” (TimeJanuary 11, 2018) <https://time.com/5099586/chrissy-teigen-sophia-the-robot-twitter/> accessed May 9, 2020
[20] Paris Convention 1883, Article 4ter
[21] Banks v. Unisys, 228 F.3d 1357 (Fed. Cir. 2000)
[22] Kim, Hee-Eun. “Role of the Patent System.” In The Role of the Patent System in Stimulating Innovation and Technology Transfer for Climate Change: Including Aspects of Licensing and Competition Law, 33-56. Baden-Baden: Nomos Verlagsgesellschaft MbH, 2011. Accessed May 8, 2020. http://jstor.org/stable/j.ctv941r23.7
[23] Macaulay T, “Why AI Systems Should Be Recognized as Inventors” (Neural | The Next WebFebruary 19, 2020) <https://thenextweb.com/neural/2020/02/17/why-ai-systems-should-be-recognized-as-inventors/> accessed May 8, 2020
[24] Willingham AJ, “Artificial Intelligence Can’t Technically Invent Things, Says Patent Office” (CNNApril 30, 2020) <https://edition.cnn.com/2020/04/30/us/artificial-intelligence-inventing-patent-office-trnd/index.html> accessed May 8, 2020
[25] Patent Act 1977, Section 1
[26] 35 U.S.C , Section 101
[27] “AI ‘DABUS’ Autonomous Inventor, but Not Official” (Legal PatentFebruary 17, 2020) <https://legal-patent.com/patent-law/ai-dabus-autonomous-inventor-but-not-official/> accessed May 9, 2020
[28] Rodriguez J, “The Missing Argument: Motivation and Artificial Intelligence” (MediumAugust 14, 2017) <https://medium.com/@jrodthoughts/the-missing-argument-motivation-and-artificial-intelligence-f582649a2680> accessed May 9, 2020
[29] Henry M, “Patent Ownership Vs. Inventorship: Who Really Controls the Rights to a Patent?” (Henry patent firmJune 14, 2018) <https://www.henrypatentfirm.com/blog/patent-ownership-vs-inventorship> accessed May 8, 2020
[30] Macaulay T, “Why AI Systems Should Be Recognized as Inventors” (Neural | The Next WebFebruary 19, 2020) <https://thenextweb.com/neural/2020/02/17/why-ai-systems-should-be-recognized-as-inventors/> accessed May 8, 2020

The views expressed in this article are those of the author and do not necessarily represent the views of CourtingTheLaw.com or any other organization with which he might be associated.

Mohtasib.pk (a project of Courting The Law) is an innovative access-to-justice platform which uses automation and predictive machine learning and enables users to leverage technology to assess the eligibility and limitation period of their claim and file an online complaint with one of the many Ombudspersons in Pakistan for free.

Muhammad Qasim Dogar

Author: Muhammad Qasim Dogar

The writer has graduated from LUMS with a joint major in Economics and Politics. He is currently working at the Office of the Chief Minister of Punjab. He is also the President of Leo Club ( and international NGO).

Leave a Reply

*

This site uses Akismet to reduce spam. Learn how your comment data is processed.