On the 12 February 2024, the USPTO issued guidance on the patenting of inventions developed with the contribution of Artificial Intelligence.
This guidance originates from the Thaler v. Vidal case, one of a series of test cases considered in courts and patent offices around the world in response to patent application filed by Stephan Thaler declaring only the DABUS AI system as an inventor. The US Supreme court confirmed that an Inventor must be a natural person, and the new guidance explores the implications of this ruling.
Specifically, the USPTO take the view that insofar as an invention originates with an AI to some degree, the contribution of the AI weighed together with other (human) contributors in the same way as the relative contributions of various human contributors to determine which should be designated as inventors. The factors underlying in this analysis are referred to as the Pannu factor, according to which, each inventor must:
(1) contribute in some significant manner to the conception or reduction to practice of the invention (ed. as defined in any of the claims),
(2) make a contribution to the claimed invention that is not insignificant in quality, when that contribution is measured against the dimension of the full invention, and
(3) do more than merely explain to the real inventors well-known concepts and/or the current state of the art
From this starting point, the USPTO consider that the involvement of AI does not rule out patent protection as long as at least one natural person can be properly identified as an inventor in view of the Pannu factors, but that many common interaction scenarios may nevertheless not justify that person to be so identified.
In particular, considering a sequential interaction starting from human input, and resulting in oriented output from the AI, merely recognizing a problem and feeding it into an AI to obtain a solution will not generally justify a person as being recognized as an inventor, although careful construction of the prompt language in order to guide the AI to a meaningful solution may clear this hurdle.
On the output side, a person who recognizes a particular usefulness in the output of an AI system as an invention, particularly when the properties and utility of the output are apparent to those of ordinary skill, is not necessarily an inventor, although performing significant work to render the output operational or to determine effectiveness may justify an individual’s designation as an inventor.
In some situations, the natural person(s) who designs, builds, or trains an AI system in view of a specific problem to elicit a particular solution could be an inventor, where the designing, building, or training of the AI system is a significant contribution to the invention created with the AI system. On the other hand, a person simply owning or overseeing an AI system that is used in the creation of an invention, does not make that person an inventor.
These considerations give rise to some very tangible implications for inventors and patentees. It is now necessary to document and assess any involvement of AI in the invention process, firstly to be sure that a Patent can properly be filed (in the US, if and when required), i.e. with a properly designated human inventor, and secondly so that in case any doubt is cast on the validity of an inventor’s designation in the future, sufficient detail as to their contribution is available to demonstrate that they made the “substantial contribution” required by the USPTO.
Some commentators now recommend an explicit “did you use AI in developing this invention, and if so, how” question in Invention Disclosures. Of course, such an approach raises the risk of inventors making declarations without properly appreciating the significance of their replies or the particular legal meaning of the terms used, in a way which may later prejudice the validity of any US Patent. In any case, it is clear that Patentees or their representatives will be well advised to assess these questions and record the facts retained and resulting conclusions during the invention evaluation process.
It should be pointed out that the current guidelines raise many questions, from basic questions of vocabulary (what is an “AI system” anyway?) to fundamental policy (the guidelines imply a growing category of inventions where no human inventor can be designated – DABUS is perhaps the first example – how are investments in the generation of such inventions, and their divulgation to the public to be incentivized if not by patents?). The USPTO’s guidance is extrapolated mainly from case law which did not anticipate application in this particular context, and it may well be that the US courts will develop new lines of reasoning in view of these policy considerations.
We can look forward to further developments addressing these points. Indeed, it will be interesting to see to what extent other important Patent Offices follow the approach of the USPTO. New guidelines from the EPO are to enter into force on 1 March 2024 addressing certain AI specific considerations, but are not understood to go beyond setting out the basic requirement that an inventor be a natural person in view of the parallel DABUS decision. Watch this space for future developments.