On 30 July 2021, in Federal Court of Australia case, Thaler v Commissioner of Patents, Justice Beach handed down a ground-breaking decision, giving legal recognition to an AI robot (‘DABUS’) as a valid inventor on a patent application.
Based in the United States, Dr Stephen Thaler is the developer of the ‘DABUS’, (an acronym for Device for the Autonomous Bootstrapping of Unified Sentience’), computer program that he claims is not only capable of autonomous invention, but has already developed a number of new inventions.
The case is one of what will be a number test cases being run globally. These cases have arisen following the filing of a string of patent applications by the Artificial Inventor Project (AIP) in 17 different countries, including the USA, the UK and New Zealand. In each of these applications, DABUS is the named inventor, and Dr Thaler the application owner.
To date test cases have been run in the United States, the United Kingdom, and in Australia. In the US, the initial decision refusing two patent applications was upheld in the United States District Court for the Eastern District of Virginia (‘EDVA’), on the basis that the AI entity is not a human being under US law, and therefore cannot be a valid inventor. While there was a slightly different focus in the UK case, largely due to differing laws regarding the role of inventors in the US versus the UK (and most other jurisdictions globally), the appeal was also unsuccessful, and the applications deemed withdrawn. Unlike the US case, however, the UK decision was not unanimous, and resulted in a strong dissenting opinion.
Under US law the inventor is fundamental. This due to the power of Congress, granted by the Constitution, to make laws that ‘promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries’. By contrast, in the UK (and in most other jurisdictions worldwide), the inventor’s identity is in many ways almost irrelevant in most aspects of patent law.
The negative results in the US and the UK leave Australia alone in granting legal recognition to a non-human entity as a valid inventor. This decision, widely seen as a radical departure from established legal and policy norms, has given rise to significant debate and commentary worldwide.
A Can Of Worms Or A Bright Future Of Innovation?
As a surprise world first, this ground-breaking Australian decision has sparked an ongoing global discussion, with commentators raising many potentially far reaching legal, ethical and policy issues.
Policy – will recognition of AI inventors further the object of the Patents Act?
Section 2A defines the object of the Act, which includes promoting ‘economic wellbeing through technological innovation and the transfer and dissemination of technology’. In Thaler, it was argued by the applicant that this objective would necessarily be progressed if the court moved to recognising machine made inventions. In support of the contention, counsel argued that specific software in sectors such as the pharmaceutical industry, has been responsible for key developments, for example, simulations and other tools that have the ability to discover treatments and perform biological modelling.
However, critics of the decision argue that this focus on select segments of particular industries can in fact obscure potential outcomes in other industries that may run counter to the object of the Act. As several commentators have pointed out, machines have the capacity to produce patents at unimaginable rates, without requiring any human input in terms of ingenuity, creativity, skill, knowledge or judgement. This scenario could ultimately favour large corporations that have the capacity to invest in a legion of AI patent generators, which could in turn could throttle innovation. So, while it is possible that recognition of machine inventors could foster technical innovation in certain sectors, it may also be that the effects in other sectors could be equally detrimental – instead of incentivising the creation of intellectual property, it may simply incentivise large corporations to develop automated patent ‘farms’, that could ultimately monopolise vast segments of technology.
Further, if Australia remains an outlier, as the only jurisdiction to grant patents for such inventions (or one of only a handful of jurisdictions worldwide), it could potentially result in an increase in foreign applicants obtaining exclusive rights in Australia, ahead of local innovators. This is particularly concerning for some, given that Australia is said to already overcompensate foreign patent owners, with the vast majority of Australia’s patent applications being filed by foreign entities.
How do non-human inventors fit within the overall scheme of the Patents Act?
The word ‘inventor’ is not explicitly defined in the Act. As such, in Thaler, the applicant argued that the term is an ‘agent noun’, that by nature does not define the identity of the entity it is referring to. Rather, the term is simply descriptive of any entity that performs the action of inventing. Justice Beach was persuaded by much of this argument, hearing evidence on how English language has evolved over time. For example, the applicant argued that the word ‘inventor’ is similar in nature to words like ‘dishwasher’ which is a noun that can apply to a human or a machine, and ‘computer’, which originally described a person who undertook calculations, but now refers almost solely to machines.
However, as many critics maintain, other elements, both historical and as a matter of legal interpretation and construction, may weigh heavily against this argument.
Under paragraph 15(1)(a) of the Patents Act, a patent may be granted to “a person who … is the inventor”. That a machine is not a person is established. For this reason, at face value, the inventor must be a natural person under the law. This view is consistent with the established legal mechanisms for transferring title to a patent. That is, to obtain and transfer title, the entity in question must be a natural human. To counter this assumption, counsel in Thaler sought to establish that the nature of the entity that is the inventor is irrelevant, because, he argued, what is necessary is that a human be “in possession” of the invention. In this case, Thaler, as the owner of DABUS, was in reality, “in possession” of the invention, and therefore capable of transferring title.
However, in Australia, many agree that there is a traditionally recognised expectation that the ability to acquire a patent stems purely from the efforts and ingenuity of a human inventor. While this is not explicit within the Patents Act, several cases over time have discussed the elements that establish someone as an inventor, and affirmed the legal mechanisms for transferring patent rights. These cases suggest that there is an assumption that inventions are the creation of the human mind and intellect, that title originates with the inventor, and that title can only be transferred according to established legal mechanisms, between human beings.
Impacts on patent law system
A key step in determining whether or not a patent should be granted is that it be “inventive”. The test for determining whether or not the inventive step has been achieved, is whether or not the invention would be obvious to a “person skilled in the art” in possession of relevant skills and the “common general knowledge”. Essentially, this test involves a comparison between what the inventor has created, and what an ordinary person would have done in the same situation.
Critics of the Australian decision have raised concerns about how this test will be impacted if a machine, rather than a human, is capable of being construed as an inventor. For example, the question has been raised whether a different test should apply between machine and human inventors, or if machines should effectively be considered human for the purposes of the inventiveness test (or vice versa).
This concern about the potential stretching of legal tests has also been raised in relation to the interpretation of the Act itself, with some arguing that a broad interpretation of paragraph 15(1)(c) which would essentially sever the connection between the inventor and the legal means for transferring title, was never the intention of parliament. Such changes, it is argued, have the potential to cause harm to the system of patent law as a whole, which in turn could impact the economy, local innovation, consumer needs, and even Australia’s international standing.
How will related legislation be affected?
While Justice Beach rejected the commissioner’s argument that an ‘inventor’ for the purposes of the Patents Act is analogous to the definition in copyright law, which requires a human author, other related IP legislation was not discussed.
In particular, it has been noted that the decision was silent on parallel provisions in the Designs Act, and whether or not there is an equivalence between an ‘inventor’ in the Patents Act, and a ‘designer’ in the Designs Act, which, under Section 13 is referred to as “the person who created the design” (emphasis added).
This is an important observation, highlighting an issue that could give rise to inconsistencies in the application of different aspects of IP law, particularly in relation to products that are currently potentially capable of registration under both the Designs Act and the Patents Act. For example, if the decision stands, it may be that an entity that is deemed the inventor of a product for the purposes of a patent application, will not be considered a designer of the very same product in a design application.
Such questions and potential inconsistencies could have detrimental effects on the overall operation and application of IP law generally.
Where to from here?
While Thaler v Commissioner of Patents has caused an international stir, it is yet to be seen whether the decision to ‘expand’ the law in relation to the non-human inventors will stand on appeal.
On 27 August 2021, the Commissioner of Patents lodged an appeal against the Federal Court decision. Given the controversy surrounding the case, IP Australia has been careful to clarify that the “appeal is centred on questions of law and the interpretation of the patents legislation as it currently stands” and that its decision to appeal is not indicative of any particular policy position of the Australian Government in relation to AI inventors. The appeal is likely to be held from November 2021, potentially in early 2022.
However, considering the strong dissenting decision in the UK case, it could be that some tentative movement in this direction may occur over time, at least in some jurisdictions outside of the US.