PATENTABILITY OF ARTIFICIAL INTELLIGENCE INVENTIONS
With AI being increasingly used for developing applications without human intervention, a lot of questions arise for intellectual property rights. One important question is whether or not to grant a patent to AI machines. Most AI today are machine learning devices. Machine learning involves feeding an AI data, having the AI score or rank the data, then presenting the AI with new data to analyse based on its learned responses to the prior datasets. Machine learning and deep learning are the branches of AI that are used most frequently in the pharma and biotech industry to generate novel molecules that may become new, patented drugs.
As algorithms become more refined, human intervention in the development process is lessened, thus, the actual inventor of an AI invention becomes ambiguous. Identifying the true and first inventor(s) of an invention is necessary in the application process. This article talks about the patentability of artificial intelligence inventions, the concerns developed from the recent case of AI DABUR and discusses the extent to which the current law is capable of including AI inventions.
Evolution of AI patent applications
One of the critical reasons for advancing artificial intelligence is to create the ability for machines to operate without human intervention. Therefore, when the machine manages to develop an invention without any kind of involvement from humans, it should be acknowledged for its contributions. Besides, according to existing patent law, as elaborated below, a machine cannot be termed ‘owner’. Thus, if humans aren’t taking the responsibility of the invention, it might not be patentable at all. This, in turn, could hamper the authenticity of the invention without any legal protections. Additionally, the commercialisation of the invention might create resistance among businesses to invest more in such AI inventions.
According to US patent law, the term — inventor, can only be used for individuals and natural persons who have invented or discovered the application and can only be the one who can claim the intellectual property.
The United States Court in the case of New Idea Farm. Equip Corp. v. Sperry Corp.[i] stated that only people conceive ideas and not machines. The Copyright Office reinforced this decision by saying that it will decline to register works produced by a machine or mechanical process if there has been no creative input or intervention from a human.
The Gulf Cooperation Council has stated that algorithms and code cannot be patented. Arguably the same rule applies to artificial intelligence machines. According to the Copyright Designs and Patents Act of 1988 in the United Kingdom, the author of any literary, dramatic, musical or artistic work created by a computer is the person that made the arrangements necessary for the creation of the work undertaken. Artificial intelligence cannot be legally protected unless some human intervention takes place. The founder or owner or the head of the company who owns the artificial intelligence will be named the inventor of the inventor.
On the other hand, there are no guidelines in India that can be applied to AI-related patent applications. At present, the guidelines for Computer Related Inventions (CRIs) encompass AI technology. The CRIs guidelines exclude a computer programme or algorithms from being patented. To claim patent protection, it is recommended to describe hardware along with AI algorithms, the method of device which uses AI, and not focussing on codes/algorithms.
The Patents Act in the Indian law, is a little ambiguous — Section 6 of the Patents Act, 1970 states that a patent can only be granted to the first ‘inventor’ of the invention, and suggests that the patentee needs to be a natural ‘person’ in order to gain ownership. However, the act also includes non-person entities like governmental organisations as patentee of an invention, thereby creating a dubiety in interpretation as to whether or not AI machines can be granted patents in India.
Further, the Act, in Section 2(1)(p), provides that a patentee is a 'person' entered in the register of the patent office as the grantee or proprietor of the patent. Having said that, the Act, provides that the person includes the Government (non-person entity), and in Section 2(1)(y), provides that 'true and first inventor' has an exclusionary definition and only excludes the first importer of an invention into India, or a person to whom an invention is first communicated from outside India. Therefore, there exists an inherent ambiguity as to whether an AI machine could be regarded as an inventor in a patent application filed in India. Since India is yet to see a case regarding AI inventorship, the current ambiguity in the law leaves scope for wider interpretation of the term 'inventor'.
The Case of DABUS
In April of 2020, the United States Patent and Trademarks Office denied giving a patent to an AI machine — DABUS. Several inventions were attributed to the connectionist system DABUS, including a method for constructing and simulating artificial neural networks, a food container, and devices and methods for attracting enhanced attention.
The decision concluded that artificial intelligence (AI) cannot be a named patent inventor. The USPTO held that interpreting "inventor" broadly to encompass machines would contradict the plain reading of the statutes that refer to persons and individuals. Accordingly, the USPTO has held that the application does not comply with the requirement that an application include, or be amended to include, the name of the inventor for any invention claimed in the application.
The Assignee also filed patent applications in Europe and the United Kingdom, identifying DABUS as the inventor. Similar to the US, the European Patent Office (the "EPO") and the UK Patent Office (the "UKIPO") did not accept 'DABUS' as an inventor. Both the EPO and UKIPO held that AI DABUS could not be regarded as an inventor under the applicable legal framework, which in the context of inventorship only refers to natural persons. The EPO also referred to the fact that under the European Patent Convention, the position of inventors is safeguarded by granting them various rights such as the right to be referred to as such in the European patent applications. However, AI systems or machines do not have any such rights because they have no legal personality comparable to natural persons or legal persons.
The UKIPO recognized the need for further deliberation in its decision when it concluded that “the present system does not cater for such [AI] inventions and it was never anticipated that it would, but times have changed and technology has moved on. It’s right that this is debated more widely and that any changes to the law be considered in the context of such a debate…)”.
Granting a patent to a machine comes with several challenges. The patent law for every country runs with strict guidelines that state that an inventor can only be termed for an individual and to file a patent that inventor needs to be employed by the parent company, which doesn’t work with machines. According to authorities, AI can only act as a medium for humans to create inventions, rather than the inventor itself. Alongside, being an owner of an invention or an application comes with the responsibility of signing contracts, taking responsibility for the invention, filing lawsuits and even to permit licenses, which can only be done by humans, and thus only they can be called as an owner.
Also, as artificial intelligence cannot hold any grounds in legal processes. Consistent with this view the 1984 case of United States v Athlone Indus., Inc.,[ii] the court stated that robots could not be sued. There is a way to confer legal personality to an artificial intelligence machine. In the United States of America, an owner of a limited liability company in the United States can hand over his or her rights of ownership to a machine. Legal personality and the ability to own assets go hand in hand. The artificial intelligence machine can then be said to hold a legal personality. Currently, an artificial intelligence machine cannot be sued because the law refers to it either as a product or a service.
But it is pertinent to note that even if the AI is designed to produce a patentable output, there will generally still be a human inventor or likely to have a human interpreting the results. This must be further further deliberated upon, in order to ensure advancement of the law in this niche.
The underlying debate on the patentability of AI inventions has just commenced, and is in the nascent stage. Indeed, at a time when the output of AI technology, including AI-based patent applications, is growing exponentially, the current legal framework is heavily inadequate and filled with lacunae. Exploring the legal, technical and policy aspects of new technologies and AI on the patent system and on the operations of the Intellectual Property Rights should come to the forefront. The question of how to handle applications for inventions created by machines must be answered fast, and rightly so.
However, until this issue is addressed, this advanced technology can only be termed as the machine or service that aids humans to invent ideas. Thus, businesses should ensure that humans are well involved in the process of creation of the invention, alongside AI, for them to apply for patents.
Title Image source: Dezeen
This article has been written by Tanya K Y. Tanya is a second year law student at GNLU, Gandhinagar.