Similarly to the European cases, the decision issued on 22 April 2020 (here) came in response to two patent applications on inventions created by an AI system called “Dabus”, in the context of the Artificial Inventor Project. Dabus is also known as the Creativity machine, which was developed by Dr. Stephen Thaler, who is named as the applicant and assignee in the patent applications. The Artificial Inventor Project has filed patent applications via the Patent Cooperation Treaty in various countries including the US, UK, Germany, and China.
The applicant (Dr. Thaler) referred that the Creativity machine is programmed as a series of neural networks trained with general information to independently create. It was the machine, not a person, who recognized the novelty and salience of the inventions at stake.
The application was listing a single inventor (Dabus) and the family name “invention generated by artificial intelligence”.
The main argument of the USPTO is similar to that of the EPO: US patent law refers to inventors as humans, individuals, or persons. The term “inventor” therefore means the individual who invented the subject matter of the invention. The patent statues preclude a broad interpretation where “inventor” could be construed to cover machines. This view was confirmed by the Federal Circuit that (albeit referring to inventorship in the context of corporations) explained that patent laws require the inventor to be a natural person (see University of Utah v. Max-Planck-Gesellschaft zur Forderung der Wiessenschaften e.V., 734 F.3d 1315 (Fed. Cir. 2013), here: “[t]o perform this mental act, inventors must be natural persons“).
Also, the Manual of Patent Examining Procedure explains that inventorship requires “conception”. And “conception” is defined as a mental act, that is the formation in the mind of the inventor of the idea of the invention. Reference to “mental” and “mind”, again, points to a natural person.
The Office has also explained that inventorship has long been a condition for patentability, as naming an incorrect inventor is a grounds for rejection.
Last, the Office has refused to enter into any policy considerations on the advantages of supporting allowing AI as inventor, as in any case “they do not overcome the plain language of the patent laws”.
Is the end of the story? As the UKIPO said, further debate is needed. It will be interesting to see what local patent offices in other countries will say, especially after a Chinese court held AI-written articles protected by copyright – see here – although a human element still appears necessary.