This year's UCL Institute of Brand & Innovation Law innovation seminar in association with CIPA and hosted by Sir Robin Jacob was titled AI: Reinventing Inventorship? The seminar focused on the UKIPO's ongoing consultation launched in October 2021 on AI and IP in the aftermath of the Court of Appeal's decision in Thaler v Comptroller General  EWCA Civ 1374 (our report of the decision can be found here). The judgment held that, under the current legislation, an inventor of a patent must be human and cannot be an AI system.
The UKIPO's consultation asks: Should we protect AI-devised inventions and if so, how should they be protected? The consultation proposes four options for consideration:
- Option 0 maintains the status quo. Patents can only be granted where the deviser of the invention is human.
- Option 1 redefines the definition of "inventor" in the legislation to include "humans responsible for making the arrangements for AI to devise an invention".
- Option 2 permits patents devised by AI. The legislation is amended to either (a) allow AI systems to be named as an inventor; or (b) removing the requirement to name an inventor if it is an AI system.
- Option 3 proposes a new sui generis "patent-like" right to protect inventions devised by AI.
The majority of the panel viewed Option 0 as not viable. AI is here to stay and will play an increasing role in devising patentable inventions. The legislation must be revised to reflect this reality. As for Option 1, it was remarked that this would likely lead to disputes over who qualifies as the "human responsible" - was this the programmer of the AI? Was it the AI operator? Or could it be someone else? Option 3 would also give rise to difficult debates. Would a new sui generis right for AI inventions have the same bar for patentability? Would it offer the same amount of protection? And what about "hybrid" inventions that are partly devised by AI and partly devised by humans? This left Option 2 which, to most of the panel, seemed to be the preferred approach. A key point that ran through the whole discussion was that any solution to AI inventorship will require a degree of global harmonisation to succeed and this will undoubtedly be a significant challenge.
The introduction of AI inventors gives rise to a further debate as to whether the characterisation of the notional skilled person should change. Should there be a "notional skilled AI" when considering the obviousness of AI inventions? The panel generally thought not - the skilled person is and should continue to be formulated irrespective of the actual inventor's characteristics. However as particular sectors begin to increase the use of AI for innovation we may see situations where inventive step is assessed from the position of the notional skilled person in possession of some form of machine learning or AI system, much like the skilled person can in some circumstances take the form of a "skilled team". This in turn may significantly impact the scope of common general knowledge that the skilled person is able to draw from using its AI.
The panel comprised Professor Ryan Abbott, University of Surrey; The Hon. Justice Beach, Federal Court of Australia; Coreena Brinck, Zacco A/S, Member of CIPA Computer Technology Committee and Nicki Curtis, Senior Policy Advisor, UK Intellectual Property Office. The Chair was Gwilym Roberts, Kilburn & Strode LLP, Honorary Secretary of CIPA.
Patent protection for AI-devised inventions. Should we protect them, and if so, how should they be protected?