Junghi Woo is an IPilogue Writer and a 3L JD Candidate at Osgoode Hall Law School.
Artificial intelligence (“AI”) is always popping in and outside of our news feeds. A prime example is Tesla’s newly announced AI-powered robot, the “Optimus”. Recently, AI technology once again exceeded the legal community’s expectations by filing a patent for its invention of interlocking food containers. Why is this an issue? Under patent law, it is the general expectation that inventors are humans, not robots.
Dr. Stephen Thaler created DABUS (‘Device for the Autonomous Bootstrapping of Unified Sentience’), an artificial neural system, and claimed that DABUS was the sole inventor of the patentable invention. Currently, from its filings in the U.S., Europe, Australia, and South Africa, only Australia and South Africa granted this patent. The main question for patent offices is: can AI technology be considered an “inventor” within patent law?
A Flexible Australian Approach
The Federal Court in Australia found that no provision expressly excludes AI from the definition of “inventor”, nor does the actual term require a human author. Resultantly, the court confirmed that it will take a flexible approach, aligning with the Australian Patents Act’s objective which reads:
“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.”
Interestingly, the court also distinguished between ownership and inventorship, emphasizing that only humans can be capable of ownership, but inventorship only requires the capacity to invent, which AI arguably possesses.
Patent Law in Canada
Contrary to other jurisdictions where “inventor” is legislatively defined, in Canada, the term “inventor” is mentioned but not defined within the Patent Act. For example, under section 2, “applicant” is defined as “... an inventor and the legal representatives of an applicant or inventor.” But, it does not define the term “inventor” or specify whether an inventor must be human.
When defining this term, experts refer to Apotex Inc. v Wellcome Foundation Ltd., 2002 SCC 77 (“Apotex”). Here, the Supreme Court interpreted “inventor” to mean “the person or persons who conceived of” the invention. This approach favours a stricter interpretation and defines “inventor” as a human individual.
However, the purpose of Canada’s Patent Act aligns closely with Australia’s Patents Act, which is to promote innovation and economic growth in Canada. In this sense, Canada could take a more flexible approach and follow the lead of our friends from across the pond. It can be further argued that Apotex did not focus on the issue of defining the term “inventor”, and the Supreme Court never considered or debated the possibility of AI inventing patentable inventions in 2002. Thus, one may say that this case does not truly address this issue, nor does it lead us to a clear conclusion of how Canada may treat such a patent filing.
Granting patents to AI “inventors” would address the gap between AI and AI-created innovations, and would align with the objective of promoting innovation. DABUS is a creativity machine, which means it can participate in machine learning, processing, and critically analysing data. With such capabilities, food containers would only be the start of a novel line of inventions. Restrictive legislation may deter future innovations from being patented in Canada, and hinder our technological advancements as a society.
Unfortunately, expanding this definition could also afford companies more inventive liability loopholes. A similar example is Tesla’s Autopilot system that caused automotive fatalities. In these cases, Tesla has argued that despite AI involvement in the driving of such cars, the human driver solely determines the vehicle’s actions, so Tesla should be exempt from liability. AI technologies’ ease in collecting, storing, and analyzing data could also raise privacy concerns when looking at the inventive process more closely.
Additionally, there is no doubt that expanding this definition would confuse some areas of intellectual property law, and perhaps even other areas of law. It would take decades to adapt, interpret, and clarify what role AI will have on Canadian legislation and case law unless AI can someday predict this as well.
Whatever the decision is, AI seems like it is here to stay and will only continue to grow and advance. The only question left is whether Canada will warmly embrace this giant step of innovation, or stay back and observe.