Inventions generally refer to products or processes involving an inventive step and having industrial application. Such novel creations are recognized and protected under patent laws. A patent entitles the holder to certain exclusive rights over his invention for a prescribed period of time, generating economic rewards.
Such protection is founded on the rationale that allowing an inventor to reap the benefits of his creation will incentivize innovation and will also prevent people from obtaining undue benefits from the invention (subject to fair and unreasonable terms).
Patent protection laws have stimulated significant breakthroughs in technology and technological capabilities. Such advancement has made it possible for inventions to be created by computers and computer programs themselves, otherwise known as artificial intelligence (AI). AI refers to the capability of machines and systems to enable tasks like problem-solving or decision-making that would otherwise require human intelligence. AI is based on machine learning, which allows the system to improve and automate without needing to be programmed with instructions at every step. However, science cannot reasonably predict when AI will be completely automated and at par with human intelligence. Nonetheless, AI technology has made large strides and has become capable of discovering drugs, developing computer languages, and inventing products and processes.
Legal developments on non-human inventorship
The legal framework on non-human inventorship is inadequate and in need of attention from lawmakers. In August 2019, the United States Patent and Trademark Office made a request for comments to address AI inventions. In September 2019, the WIPO Conversation on Intellectual Property and Artificial Intelligence was set up to discuss the impact of AI on Intellectual Property Rights. Three sessions have been conducted thus far on this.
The question of inventorship of AI-generated inventions came to the fore when an application was made by Dr Stephen Thaler for inventions made by an AI system called ‘Device for Autonomous Bootstrapping of Unified Sentience’, commonly referred to as DABUS. Patent applications in numerous countries naming DABUS as an inventor were filed as part of the Artificial Inventor Project headed by Dr. Ryan Abbott, University of Surrey, to advocate for the recognition of non-human inventors. DABUS invented an improvised beverage container and a neural flame for search and rescue operations entirely on its own and Dr. Thaler sought a patent, naming DABUS as the inventor in the United States of America, United Kingdom, Canada, China, South Africa, Germany, India, Japan, Israel and Europe.
While the United Kingdom Intellectual Property Office, European Patent Office and the US Patent and Trademark Office did not object to the patentability of the inventions themselves, they refused to grant a patent for a non-human inventor. Conversely, DABUS was designated as an inventor in South Africa’s Patent Journal on July 28, 2021. Furthermore, on July 30, 2021, the Federal Court of Australia delivered a landmark decision in the case of Thaler v. Commissioner of Patents, stating that AI could be named as the inventor in a patent application.
Setting the ball rolling: Australian Court affirms AI as an inventor
When the patent application naming DABUS as an inventor was filed in Australia, the Commissioner of Patents rejected it on the ground that the inventor named had to be a human. The matter came before the Federal Court after Dr. Thaler sought a judicial review of the rejection.
The Court considered the use of AI technology in inventions, especially in the pharmaceutical industry and adopted a broader understanding of the term 'inventor' so as to prevent inhibition of innovation in computer science and other allied sciences and held as under:
Since the Patents Act 1990 in Australia does not define an inventor, they relied on the ordinary term of the meaning.
The underlying idea of the word must be examined, rather than its usage over the past years so that a broader meaning may be affixed to the term, in line with innovation and advancement.
The Act neither specifically provides that an inventor must be a human, nor does it expressly prohibit an AI system from being designated as an inventor.
The Court also examined the object of the Act, which is to promote technological innovation and dissemination. Therefore, to not recognize non-human inventors would be inconsistent with the legislative intent.
The purpose of the Act was to also enable inventors to enjoy the benefits of their inventions and to not recognize AI-inventorship and the inventive process would result in improper recognition and reward.
A patent can be granted when an application under Section 15(1)(b) and (c) of the Act is made, naming an AI as an inventor by an applicant who is entitled to have the patent assigned to them or if they derive a title to the invention from the inventor, which Dr. Thaler in this case did.
The Court also affirmed Dr. Thaler's contention that as the owner of the AI, he was entitled to reap its benefits the same way that the owner of a tree is entitled to its fruit.
Indian scenario: The Patents Act, 1970
Patent law in India (under the Patents Act, 1970) does not define an 'inventor'. However, Section 2(1)(p) of the Act defines a 'patentee' as a person who is entered in the Register of Patents as a grantee or a proprietor. The Act has not clearly defined the term 'person’ and Section 2(1)(s) merely states that a 'person' includes the government. Section 6 of the Act enumerates persons entitled to make an application for a patent -the true and first inventor, an assignee of the true and first inventor or a legal representative of the person entitled to make an application. This provision bears a likeness to Section 15 of the Australian Act and it may be possible, following the Thaler decision, to include AI within the ambit of an inventor.
While these provisions may be analyzed to determine whether an 'inventor' can include AI, reliance should be placed on Form 1 for the ‘Application for the Grant of Patent’. As per this form, an applicant may be a natural person or a non-natural person that includes a small entity, a startup or 'others'. Further, the form also provides an option for the inventors named to be the same as the applicants. This open-ended categorization leaves ample room for a broad interpretation that has within its ambit, a non-human like AI. However, it remains to be seen how the DABUS application will be treated under the Indian patent framework.
Examining AI as an inventor
Before dwelling into whether AI’s inventorship can be legally recognized, one must examine the extent to which AI is actually capable of 'invention'. While AI technology has evolved to the extent of inventing pharmaceutical drugs, computer languages and even products, it can do so through machine learning and algorithms. These enable the AI to automate programming, problem-solving and decision-making, and therefore, it can be argued that AI is a tool used for an invention, rather than the inventor itself. However, the US Patent and Trademark Office has included within the ambit of AI inventions, those inventions that utilize AI as well as ones developed by AI.
It is important to understand that AI still requires human intervention to identify the task to be performed, and sometimes even the manner in which it should be performed. An AI system that can do so on its own, possessing an intelligence at par with humans, is known as Artificial General Intelligence and is predicted to be engineered only by 2099. This necessitates a distinction between AI that is automatic and AI that is autonomous. Since AI today is merely automatic, human contribution to the invention can be clearly determined and this may in some cases even be sufficient to grant the status of inventorship to the human as well.
Recognition of AI as an inventor, being a progressive move, can certainly pose challenges, primarily to human inventors. Patent law requires that a patent should be non-obvious to a person skilled in the art, and the recognition of an AI inventor could possibly raise the bar for a person skilled in the art, making it difficult for human inventors to obtain patents.
Another aspect that warrants attention is the lack of autonomy in AI and the human contributions made in AI-generated inventions. The extent to which an AI has made an invention in an autonomous manner with minimal human input will vary on a case-to-case basis and lawmakers need to clearly define the ‘extent of contribution’ that will be substantial enough to claim inventorship and be able to identify that contribution in every case. An inadequate system for the same would result in incorrect patent applications, which will threaten the validity of the patent. A possible solution to this could be to obtain the patent under an agreement for co-inventorship.
It has also been argued that anthropomorphizing AI poses legal and social challenges and can be done away with since AI is a mere executor of instructions. To designate AI as an inventor would mean to attribute it with a legal personality possessing proprietary rights, along with liabilities that need to be defined.
The European Parliament has stated that any wrongful acts or omissions by an AI will make its human operator/manufacturer/user liable and not the AI itself. This could discourage the use of AI for inventions. Further, the US District Court for the Eastern District of Virginia held that only a human being can be legally listed on a patent as an inventor.
AI technology may still be in its nascent stages, but it is being used, especially in the pharmaceutical industry, for the invention of drugs in an efficient manner. It is necessary that legal provisions to address AI inventorship are laid down, clearly defining the extent of contribution that would entitle an AI or a human to the status of an inventor. As Justice Beach noted in the Thaler decision, a broader interpretation of the term 'inventor' must be adopted for the advancement of technology and innovation.
*The authors are Founder and Associate respectively with Shivadass & Shivadass (Law Chambers). The authors would like to acknowledge the contributions of Aradhana Pandit, a 4th Year law student from School of Law Christ University, Bangalore.
The contents and comments of this document do not necessarily reflect the views/position of Shivadass and Shivadass (Law Chambers) but remain solely of the author(s). For any further queries or follow up, please contact .