01 August 2021

In Becoming the First Country to Recognise Non-Human Inventors, is Australia a Hero of Progress, or a Chump?

Menacing cyborgAs I recently (tentatively) predicted, on Friday 30 July 2021 Justice Beach in the Federal Court of Australia handed down a judgment giving Australia the dubious honour of becoming the first country in the world to legally recognise a non-human as a valid inventor on a patent application: Thaler v Commissioner of Patents [2021] FCA 879.  I would suggest that the remarkable speed with which this unnecessarily lengthy (228 paragraphs) decision was rendered, after being heard on 2 July 2021, may reflect the judge’s enthusiasm for issuing such a ground-breaking ruling.  Unfortunately, I do not share that enthusiasm, and I am confident that there are many others who are equally uncomfortable with the outcome.  My hope is that this includes officials within IP Australia and the Department of Industry, Science, Energy and Resources, and that the decision will be duly appealed to a Full Bench of the court.  It is, in my view, deeply regrettable that the Commissioner of Patents did not put on a stronger defence in the first instance because, even though an appeal was probably inevitable either way, the worldwide publicity that this decision is now generating is not necessarily beneficial for Australia.

The judge summarised his reasoning (at [10]) that:

…in my view an artificial intelligence system can be an inventor for the purposes of the Act. First, an inventor is an agent noun; an agent can be a person or thing that invents. Second, so to hold reflects the reality in terms of many otherwise patentable inventions where it cannot sensibly be said that a human is the inventor. Third, nothing in the Act dictates the contrary conclusion.

The patent system faces many challenges, but right now a need to grant more patents in a wider range of circumstances in not one of them.  We are in the grip of a global pandemic, and very serious questions are being asked about whether patents deliver a net benefit to the people of the world by incentivising the development of new vaccines and treatments, or whether they have the detrimental effect of denying affordable access to vital care and protection in poor and developing nations.  While I am firmly in the former camp, it only becomes harder to defend the patent system when opponents see the law expanding access to allow inventions generated by machines – potentially including those owned and controlled by giant corporations.

The standing and reputation of Australia and our patent laws are also at risk.  The country is already the target of criticism – rightly or wrongly – for declining to support calls for a waiver of IP provisions of the WTO Agreement on Trade-Related Aspects of Intellectual Property (TRIPS).  The Federal Court’s decision in Thaler is receiving global attention, and not all of it is positive.  One tweet (in Spanish) compares Australia’s patent-friendly approach unfavourably with the infamous incident of an early innovation patent being granted for a ‘circular transportation facilitation device’, a.k.a. the wheel.

Just because patents are (or, at least, can be) good, it does not follow that more patents, generated in more ways, by more entities, must be better.  Australia should not think that we will necessarily come across as a socially and technologically progressive nation by ‘leading the way’ on allowing patents to be granted for inventions generated by non-humans.  On the contrary, we risk being left out on our own and looking like chumps.  The United States will not follow our lead – there are Constitutional, statutory and procedural barriers to permitting US patent applications naming non-human inventors.  The European Patent Office will not follow in the foreseeable future – it has already established its position via an academic study and discussions with member states of the European Patent Convention.  The UK has so far rejected any expansion of inventorship to non-humans.  And, contrary to recent reports (and the claims of the Artificial Inventor Project’s Ryan Abbott), the recent grant of a patent in South Africa naming DABUS as inventor indicates nothing about that country’s law or position on the issue.  As South African patent attorney Pieter Visagie has explained, the application effectively avoided any scrutiny of the legitimacy of the inventor by virtue of being filed via the international (PCT) system.

So what does Australia gain by being the first – and possibly only – country in the world to legally recognise non-human inventors?  Nothing, as far as I can see, other than a whole lot of unneeded publicity and global scrutiny of our patent laws.  If we are lucky, we will not receive many serious patent applications for inventions generated by machine inventors, and little practical harm will be done.  At worst, however, we could become the only country in the world to grant patents on such inventions, mostly filed by foreign applicants, creating exclusive rights that are enforceable only in Australia to the relative detriment of Australian innovators and consumers.

Why the Decision is not Really (or Just) About AI

A large section of the decision (25 paragraphs, from [19]-[43]) is devoted to a discussion of artificial neural networks in general, and to the alleged machine inventor DABUS (‘Device for the Autonomous Bootstrapping of Unified Sentience’) in particular.  I find it difficult to identify the relevance of much of this explication.  The explanation of neural networks is superficial and outdated, while the discussion of DABUS is based on the writings of its developer, Dr Stephen Thaler, which I personally find largely incomprehensible, and certainly no basis for reproducing any of his results (assuming that is possible).

In any event, as the judge rightly observes (at [142]):

The Act focuses on inventive step, which uses a hypothetical and objective construct. One is not at all concerned with the inventor’s mental processes as such.

In other words, if we are going to recognise non-human inventors, their nature is irrelevant.  They could be artificial intelligences, but they could equally be purely deterministic search and optimisation algorithms designed to find solutions to specific problems.  Or they could be monkeys.  Whether DABUS is everything that Thaler claims it to be is beside the point.  The issue here is that the inventor is said to be a machine and, taking that at face value, the task of the court is to decide whether that is permissible.

The media have, of course, picked up on the theme of an artificial intelligence becoming an inventor, because this challenges our very notion of what distinguishes us as humans and raises the spectre of us being replaced by machines (thus amusing headlines like I’m sorry Dave I’m afraid I invented that).  But really the decision only addresses whether the Australian Patents Act 1990 permits the grant of a patent based on an application that does not name any human inventor and, if so, who can claim ownership of the resulting patent.

What is an ‘Inventor’?

The judge accepted Thaler’s submission that the word ‘inventor’ is not limited to humans (at [120]):

…as the word “inventor” is not defined in the Act or the Regulations, it has its ordinary meaning. In this respect then, the word “inventor” is an agent noun. In agent nouns, the suffix “or” or “er” indicates that the noun describes the agent that does the act referred to by the verb to which the suffix is attached. “Computer”, “controller”, “regulator”, “distributor”, “collector”, “lawnmower” and “dishwasher” are all agent nouns. As each example demonstrates, the agent can be a person or a thing. Accordingly, if an artificial intelligence system is the agent which invents, it can be described as an “inventor”.

Personally, I am not persuaded, and would suggest that this somewhat stretches the application of ‘ordinary meaning’.  While dictionary definitions typically reflect historical usage, and may take some time to catch up as language evolves, there are very few people in the world today who would consider the word ‘inventor’ to refer to anything other than a human being.  It is not that the dictionaries are lagging behind.  Rather, it is that in no sense can the ‘ordinary meaning’ of the word ‘inventor’ currently be said to encompass a machine.

What is the Legal Basis for an Expansive Interpretation?

The judge argues that taking a ‘flexible’ approach to the meaning of ‘inventor’ is analogous to the approach to the ‘manner of manufacture’ test for subject matter eligibility under Australian patent law (at [121], citation omitted):

…it has been said that a ‘widening conception of “manner of manufacture” is a necessary feature of the development of patent law in the twentieth and twenty-first centuries as scientific discoveries inspire new technologies’…. I see no reason why the concept of “inventor” should not also be seen in an analogously flexible and evolutionary way. After all the expressions “manner of [new] manufacture” and “inventor” derive from the 21 Ja 1 c 3 (Statute of Monopolies) 1623 (Imp) s 6. There is a synergy if not a symmetry in both being flexibly treated. Indeed, it makes little sense to be flexible about one and not the other. Tension is created if you give flexibility to “manner of manufacture” and then restrict “inventor”. You would be recognising an otherwise patentable invention and then saying that as there is no inventor it cannot be patented.

This is an appealing argument, but in my view it is false.  When the current Act was drafted, the ‘manner of manufacture’ formulation was purposefully retained to enable the courts to continue to adopt a flexible approach to assessing the patent-eligibility of new technologies.  The phrase is to be interpreted as per the High Court’s 1959 NRDC decision, i.e. by asking ‘Is this a proper subject of letters patent according to the principles which have been developed for the application of s. 6 of the Statute of Monopolies?’ ([1959] HCA 67 at [14]).  By contrast, I think that it is safe to say that nobody involved in the drafting of the Act, or in its subsequent application (until now), has ever considered that the word ‘inventor’ would refer to anything other than a human being.

Furthermore, it may be perfectly reasonable to exclude those inventions that have not been devised by a human inventor.  It all depends upon what types of activity you wish to incentivise through the patent system.  If you want to promote the advancement of human ingenuity, while offering lesser (or no) rewards for automated innovation, then it is perfectly sound policy to deny patents on machine-made inventions.  Australia is in no obligation to grant such patents.  Nothing in any treaty or convention to which the nation is a party requires it, and (thus far, at least) no comparable country, including all of Australia’s major trading partners, recognises machines as legitimate inventors for the purposes of obtaining patents.

Is Dr Thaler Entitled to Patent DABUS’ Inventions?

Beyond the definition of ‘inventor’, the other potential obstacle to patenting of inventions made by non-humans is that there must be some legal person (i.e. a natural person or a corporation) that is both capable of owning a patent (which is a form of property) and legitimately entitled to ownership.  Where a human is the inventor, the normal process by which another party gains entitlement is by a sequence of one or more transfers of legal rights (e.g. assignments, or other contractual arrangements) originating with the inventor.  This fundamentally requires that the inventor is able to own rights in the invention in the first instance, and to legally transfer those rights to other parties.  In other words, the inventor (on this conventional view) must be a natural person.

Thaler has not suggested that DABUS should be recognised as a legal person, and nobody is suggesting that machines should have rights, be able to own property, or be capable of executing legal agreements.  The judge has observed, at [8], that ‘DABUS is not a natural or legal person’, and at [178] that ‘DABUS, as an artificial intelligence system, is not a legal person and cannot legally assign the invention…’.

Most relevantly (although the decision discusses other provisions) paragraph 15(1)(c) provides that a patent may be granted to ‘a person who … derives title to the invention from the inventor…’.  As noted above, this would normally imply a chain of one or more legal transfers of the relevant rights.  The judge found (at [189]) that:

In my view, Dr Thaler, as the owner and controller of DABUS, would own any inventions made by DABUS, when they came into his possession. In this case, Dr Thaler apparently obtained possession of the invention through and from DABUS. And as a consequence of his possession of the invention, combined with his ownership and control of DABUS, he prima facie obtained title to the invention. By deriving possession of the invention from DABUS, Dr Thaler prima facie derived title. In this respect, title can be derived from the inventor notwithstanding that it vests ab initio other than in the inventor. That is, there is no need for the inventor ever to have owned the invention, and there is no need for title to be derived by an assignment.

In other words, the phrase ‘derives title to the invention from the inventor’ in paragraph 15(1)(c) includes within its scope – according to the judge – ‘obtains possession in circumstances where the possessor has ownership and control of the inventor’.  There are some venerable legal authorities that are said to support this interpretation of the phrase ‘derives title’, however I think that it is far from apparent this is correct.

As I have noted, the Act was drafted, and has always been applied, on the assumption that an ‘inventor’ is a human being, and that ‘derives title’ therefore envisages a chain of legal title, such as assignments, employment contracts, and so forth.  However, if non-human entities are to be permitted to be inventors, then it becomes necessary to interpret the phrase ‘derives title’ more flexibly, so that paragraph 15(1)(c) can continue to function.  It is not that this broader interpretation was always present, or was ever intended.  It was never previously required or contemplated.  The need only arose when the meaning of ‘inventor’ was expanded.  In this sense, the wider interpretation of ‘derives title’ rather seems to presuppose the conclusion that the scheme of the Act is consistent with recognising non-human inventors.  On this view the judge’s interpretations of ‘inventor’ and ‘derives title’ are a house of cards.

Do Non-Human Inventors Further the Object of the Patents Act?

Finally, it is necessary to deal with the policy implications of the decision.  The judge determined that he was entitled to develop the law to adopt a more flexible and expansive construction of ‘inventor’ than has historically been the case on the basis that it is consistent with the object of the Act to do so.  Conveniently, this object is now explicitly set out in section 2A:

The object of this Act is to provide a patent system in Australia that promotes economic wellbeing through technological innovation and the transfer and dissemination of technology. In doing so, the patent system balances over time the interests of producers, owners and users of technology and the public.

The judge’s discussion of this issue can be found principally at [122]-[134], and includes the following points:

  1. ‘it is consistent with the object of the Act to construe the term “inventor” in a manner that promotes technological innovation and the publication and dissemination of such innovation by rewarding it, irrespective of whether the innovation is made by a human or not’ (at [124]);
  2. ‘computer inventorship would incentivise the development by computer scientists of creative machines, and also the development by others of the facilitation and use of the output of such machines, leading to new scientific advantages’ (at [125]); and
  3. ‘[n]ot recognising the reality [that machines are capable of generating inventions] could produce inefficiency if not logical difficulties, which would be the antithesis of the s 2A object’ (at [129]).

In my view, none of these points addresses the section 2A object as a whole.  While all could be said to address an object of promoting technological innovation and the transfer and dissemination of technology, it is not clear that any consideration has been given to the additional words ‘economic wellbeing’.  There is no evidence of any broader benefit to the Australian economy, beyond the direct private benefits to those involved in the development and use of ‘creative machines’.

Furthermore, I can see no consideration of how allowing patents on inventions devised by non-humans serves to ‘balanc[e] over time the interests of producers, owners and users of technology and the public.’  Certainly there are benefits to producers and owners of ‘creative’ technology, who will be entitled to obtain patents in Australia (that may not be available to them anywhere else in the world).  Whether these patents will provide balancing benefits to users and the wider public is less certain.  Again, there is no evidence.

To the extent that the judge relied on extrinsic materials to support his observations, these were academic publications that represent ideas rather than any concrete evidence of economic or technological advantages or disadvantages of granting patents on machine-made inventions.  There are, to be sure, documented reports of algorithms and AI playing an increasingly important role in the pharmaceutical industry – an area in which the judge appears to have a particular interest, presumably due to his educational background in chemistry and the philosophy of science – but there is no evidence that the unavailability of patents for machine-made inventions is creating any significant difficulties, or that this translates to any other fields of technology.  If there were a real need, major corporations would surely be crying out for change, and they are not.

It concerns me that a significant portion of the material cited in the decision to support a policy argument for allowing patents on machine-made inventions was authored by Professor Ryan Abbott of Surrey University.  It is not clear to me whether the judge is aware that Professor Abbott leads the Artificial Inventor Project, which is the true entity driving the DABUS patent applications around the world, and which may be funding the court action in Australia.  In any event, I would question whether it is appropriate to treat this material as an independent source of supporting analysis, and I wonder whether to court would do so if Professor Abbott was actually a party named in the proceedings.

The legislative history of section 2A was not considered by the judge, but is also relevant.  The Australian government’s Productivity Commission (PC), in its 2016 Review of Australia’s IP Arrangements, formed the view that Australia is not a net beneficiary of its own patent system, in large part because over 90% of all patents in this country are issued to foreign entities, and that too many represent relatively low-level innovations that would have occurred without the added incentive of a patent.  The PC’s final Report stated that ‘targeting socially valuable and additional innovations is a necessary condition for the patent system to meet the principles of a well-functioning IP system’.  One reason for recommending the inclusion of an objects clause was to put additional focus on the interests of users and the public, which might lie in fewer, rather than more, patents being issued – particularly to foreign entities.  The government accepted this recommendation, and legislated accordingly.  It is hard to see how providing more opportunities for foreign applicants to obtain Australian patents satisfies this goal.

Conclusion – the Commissioner may Appeal, but is the Damage Done?

I do not consider the decision in Thaler to serve Australia’s interests.  It makes us an international outlier, and it is neither as clever nor as progressive as the judge appears to believe.  I think that it represents a form of judicial activism that results in the development of policy – in this case, the important matter of who, or what, can form the basis for the grant of a patent monopoly enforceable against the public at large – from the bench.

As discussed above, it is also not at all certain that the judge is correct in finding that the word ‘inventor’ includes non-humans, that entitlement to a patent can be derived in any way other than a chain of legal title from a human inventor, or that allowing patents on inventions by non-humans is consistent with the object of the Patents Act.

I am hopeful that the Commissioner will appeal the decision, and that she will mount a stronger case on appeal.  Unfortunately, however, a certain amount of damage has already been inflicted.  Even if the Thaler decision is overturned, that event would be unlikely to garner the same level of global interest as Australia becoming the first country in the world to legally recognise an ‘AI’ inventor, which is likely to become as ‘celebrated’ as Saudi Arabia granting citizenship to a robot.


Post a Comment

Copyright © 2014
Creative Commons License
The Patentology Blog by Dr Mark A Summerfield is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Australia License.