INVENTIONS WITHOUT INVENTORS: THE NEED TO RECOGNIZE AI SYSTEMS AS INVENTORS.

• Date: 01 January 2025
• Author: Selvadurai, Niloufer

 TABLE OF CONTENTS
                ABSTRACT 37
                INTRODUCTION 39
                 I. THE HISTORICAL AND THEORETICAL UNDERPINNINGS OF PATENT LAW 43
                 A. The Central Role of Inventiveness in Calibrating 43
                 Competing Interests
                 B. The Absence of Discourse on the Need for a Human Inventor 47
                 C. The Alignment of Patent Theory to Technological Change 48
                 II. JUDICIAL CONSIDERATION OF INVENTIVENESS AND INVENTORS 50
                 A. The Patent Applications for a Beverage Container and a 50
                 Flashlight that Shook the World
                 B. Inventiveness Not in Dispute 51
                 C. Can an AI System be an "Inventor"? 51
                 D. The Cognitive Capacity of AI Systems 56
                 E. The Relevance of Personal Pronouns 59
                 F. The Enduring Appeal of the Romantic Creator 59
                III. THE VIEW FROM THE EDGE (OR IS IT THE FUTURE?) 60
                 A. The Need to Uphold the Overarching Objectives of Patent 60
                 Law
                 B. Alternative Musings on the Cognitive Capacity of AI 62
                 Systems
                 C. Personal Pronouns and Agent Nouns 64
                 D. Patent Law's Demonstrated Capacity for Evolution 64
                 IV. THE NEED FOR STATUTORY REFORM 68
                 A. The Potential Risks of Denying AI Inventors Patent 68
                 Protection
                 B. Is this Patent Law 's Apple Copyright Moment? 69
                 C. A Suggested Pathway Forward 71
                CONCLUSION 73
                

INTRODUCTION

The public policy rationale for granting patents is to support innovation and progress. (1) While this objective is differently worded in patent statutes around the world, and while scholarly theories differently articulate how patents seek to achieve this objective, the enduring aim of patent law is the advancement of innovation. (2) The Constitution of the United States (US) pronounces that "Congress shall have Power... [t]o promote the Progress of Science and useful Arts," (3) leading to the enactment of laws to confer patent protection on "any new and useful process, machine, manufacture, or composition of matter...." (4) The Patents Act 1977 of the United Kingdom (UK) similarly stipulates the need for law to support future patents (5) while the Patents Act 1990 of Australia refers to the need for the law "to provide a patent system in Australia that promotes economic wellbeing through technological innovation and the transfer and dissemination of technology." (6) As such, the central focus when considering a patent application is the "inventiveness" of the device and the extent it departs from prior knowledge. (7) The extent to which the claimed invention reflects the uniquely human attributes of its inventor or can only be attributed to a human inventor, has never been featured in either the theoretical or judicial discourse. It is therefore interesting that references to "individuals" (US) and "persons" (UK and Australia) variously scattered through present patent statutes, often found in incidental enabling or administrative provisions, have resulted in AI inventors being denied the protection of patent law. (8)

As the law presently stands, even if an AI-generated invention is inventive and novel, it will be denied patent protection. This is the unambiguous finding of the courts involved in the global test litigation instigated by Dr. Stephen Thaler. (9) Significantly, in all these proceedings, the inventiveness of the invention claimed to have been created by the AI system was not in dispute, but rather, the main contested issue was whether an AI system could be an inventor. (10) As a result, we presently have the curious situation where there can be inventions without inventors. (11) While the Thaler decisions constitute a correct, formal interpretation of present statutory provisions, they raise the question--should legislators intervene to remove references to individuals and persons from patent statues to enable AI-generated inventions to be patented? (12) While patent statutes refer to patent applications being made by individuals and persons, the substantive criteria for patentability remains firmly invention-centric, not inventor-centric. The reason for such a reform would hence be to maintain the central objective of patent law, the advancement of innovation. (13) The Thaler litigation also raises the tangential, but nevertheless intriguing question--is Thaler the patent law equivalent of copyright law's Apple moment? (14) In the 1980s, Apple computer programs were denied copyright protection in certain jurisdictions on the basis that they did not fall within the definition of "literary works." (15) Subsequent legislative changes to copyright statutes led to copyright protection to object and source codes. (16) Consider the progress in computer software innovation if copyright legislation had not been amended to provide such protection. It is interesting to consider whether society is at a similar law and policy juncture regarding patent law in this new and unforeseen context of AI inventions.

It is regularly said that the AI revolution is upon us, (17) and that generative AI will be responsible for unprecedented technological breakthroughs in future years. (18) This is not merely expected in the commercial sphere, through increased productivity and cost reductions, (19) but also anticipated in the spheres of health and justice, where AI-driven technological advances are expected to shape the quality of life and well-being of individuals. (20) Moreover, as AI has the capacity to broadly diffuse across a variety of technologies, products, and services, AI has been described as the next "general-purpose technology," akin to prior developments of steam power, electricity and semi-conductors. (21) It therefore has the capacity to have a massive effect on societal welfare, security, and economic growth. (22) Fortunately, to date, patent law has proven to be an extremely flexible body of law, transitioning from the protection of components of steam engines, to business methods, to computer software, to quantum computing. (23) such flexibility has been achieved both by expansive judicial readings of existing statutory provisions and legislative amendments. This pragmatism is perhaps explained by the substantial risks of maintaining an outdated patents system to productivity and progress. considering the future role of AI and its potentially momentous economic implications, it is imperative that patent law continues to provide the appropriate form of protection to support economic growth and progress.

In such a complex technological, commercial, and legal environment, the central objective of this paper is to analyze whether the present requirement for a human inventor in patent statutes should be removed to extend patent protection to inventions by AI systems. To do so, this paper will commence by considering the historical and theoretical underpinning of patent law, noting its focus on the inventiveness of the claimed product or process and its disregard for the human characteristics of the named inventor. Building upon this theoretical foundation, the paper will then consider the role of inventiveness in patent law and the relevance of human input to the patentability of an invention, connecting past scholarly discourse to present technological advances. This issue will be considered within the ambit of patent theory, law, and policy, suggesting that a finding of non-human invention neither undermines the value of human creativity nor justifies concerns as to a dystopian android future. As the patentability of AI-generated inventions is of substantial global relevance, it is valuable to transcend national boundaries and identify concerns, musings, and themes that resonate across jurisdictions. In support of such interjurisdictional analysis, the paper will consider the laws of the US, the UK, and Australia. These jurisdictions were selected based on the sophistication of judicial and legislative discourse on the patentability of AI-generated inventions and the value of considering both common and civil law jurisdictions. The paper will conclude by presenting the case for legislative reform to permit the patentability of AI-generated inventions. In doing so, the paper will examine the dissenting Thaler judgments of Justice Beach of the Australian Federal Court (24) and Lord Justice Birss of the UK Supreme Court, (25) as well as various (failed) submissions made by Dr. Thaler. (26) While unregulated AI presents substantial risks--and such risks need to be carefully addressed through law reform--it is suggested that patent law is neither an appropriate nor precise lever through which to control the development of generative AI. The role of patent law is to calibrate incentivizing invention through the balancing of the inventor's benefits of a temporary monopoly with the public's benefit of immediate access to new knowledge. This is the balancing test through which the patentability of AI-generated inventions will be analyzed.

I. THE HISTORICAL AND THEORETICAL UNDERPINNINGS OF PATENT LAW

A. The Central Role of Inventiveness in Calibrating Competing Interests

Historically, the law has been reluctant to confer monopolies on private entities. Yet, in the sphere of patents, this governing principle has been overridden for the clear and stated purpose of supporting innovation. As far back as 1421, a form of patent was granted in Italy to Filioppo Brunelleschi for a "newly invented ship [that] could haul loads more cheaply and that it would provide futher [] benefits to merchants and others." (27) An exclusive three-year grant was awarded to produce a barge with hoisting gear for transporting marble. (28) In 1474, a Venetian law formally conferred privileges lasting ten years to inventors of new arts and machines. (29) The objective was to encourage the building of devices of "utility and benefit." (30) In England, during the reign of Elizabeth I, the Crown granted monopolies to individuals who introduced new inventions. (31) A 1601 Parliamentary debate expressed outrage at the wide range of products which were the subject of such monopolies, including "currants, iron, powder, cards, ox-shin bones, train oil, transportation...