The article “One small step for “artificial intelligence”, and a giant leap for the Australian patent system? The Federal Court decision in Thaler v Commissioner of Patents” analyses the reasoning of Beach J in the Thaler decision at first instance, noting that this decision is the first judicial consideration in Australia relating to the impact of artificial intelligence (AI) on the Australian patent system. The author Adam Liberman carefully sets out his analysis and considers that each of the conclusions of his Honour can be genuinely challenged. The author details his concerns that the principles of the decision, if embraced, will lead to greater uncertainty and a lack of coherence in the operation of the Patents Act from a number of legal, commercial and practical perspectives.
The article briefly explains the background to the Federal Court decision, which arose from an application for judicial review of a Patent Office decision rejecting the patent application for failure to comply with the formalities requirement that the patent applicant must provide the name of the inventor. The Patent Office did not accept that Australia’s patent statute and associated regulations permitted an AI system (as opposed to an individual) to be identified as an inventor.
In the article, the author isolates the key elements of Beach J’s reasoning, and proceeds to carefully analyse their merit. The analysis covers each of the key submissions of the Commissioner of Patents (which includes a number of linguistic statutory constructions issues and the objects clause in the Patents Act), his Honour’s consideration of “inventive step” in connection with AI inventorship issues, and the implications of section 15 of Patents Act, which deals with eligibility to be granted a patent.
The article considers his Honour’s dismissal of the Commissioner’s argument that the Federal Court decision in JMVB Enterprises Pty Ltd v Camoflag Pty Ltd ((2006) 154 FCR 348) should be followed by holding that the word “inventor” in the Patents Act bears its ordinary English meaning and refers to the person who makes or devises the invention. The article considers the manner in which his Honour dismisses this argument and argues that the key point of the JMVB decision was not that it was not concerned with AI systems, but rather that it characterised an inventor only in human terms. The author then argues that, contrary to the holding of his Honour, it is equally valid to contend that the concept of inventor as considered by the JMVB case was exhaustive, or at the very least consistent with the Act.
The article discusses his Honour’s treatment of the argument that the ordinary meaning of “inventor” is inherently human, with particular focus on his Honour’s criticisms of the use of dictionary definitions in interpretating the meaning of “inventor” in the statute. The article contrasts his Honour’s approach with other established judicial approaches to using dictionary definitions in statutory interpretation. The article also considers the closely related point that his Honour considers that “inventor” may be characterised as an “agent noun”; that is, a noun that describes the agent who invents, without limitation as to whether that agent is a person or a machine. The author makes the interesting observation that in all his Honour’s examples of agent nouns (e.g. computer, controller, regulator etc.), the dictionary definitions recognise that these nouns may extend to non‑human agents in the appropriate context. In particular, the author notes that dictionary definitions are reflective of common usage, and none of the cited dictionary definitions for “inventor” recognise that an inventor may be a non-human agent that invents.
The article also considers his Honour’s significant reliance on the recently added objects clause in the Patents Act to reach his conclusions and considers additional principles of statutory interpretation relating to objects clauses not referred to by his Honour.
A significant portion of the article is devoted to critiquing his Honour’s treatment of the relationship between section 15 of Patents Act (concerning eligibility to be granted a patent) and the concept of inventorship.
There is a particularly interesting discussion of the concept of possession of an invention giving rise to entitlement under section 15 as distinct from deriving title by way of assignment. The author identifies an important distinction between physical possession and possession of a concept in the mind of a human inventor. The article also identifies a number of practical difficulties with applying the physical possession route to entitlement in circumstances where the owner, controller and programmer of the AI is not all the same person as was the case with Dr Thaler in the instant decision.
Overall, this is a highly interesting and thought-provoking analysis to be commended to practitioners and others interested in examining the development of this area of Australian patent law.
Read the full article in the Journal of Intellectual Property Law & Practice, Volume 17, Issue 2, February 2022.