Patentability of AI-assisted Inventions
DBL’s Managing Partner Tom Dunlap recently wrote an interesting article about the current state of artificial intelligence (AI) inventorship. Mr. Dunlap detailed the patenting efforts surrounding DABUS, an AI system developed by Dr. Stephen Thaler that autonomously created certain inventions. In Thaler v. Vidal, The Federal Circuit held in no uncertain terms that only a natural person—not an AI—can be named as an inventor to a patent in the United States.[1] However, the Federal Circuit left open the question of “whether inventions made by human beings with the assistance of AI” could be patentable.[2]
Dr. Thaler appears to be more interested in pushing the boundaries of inventorship to include artificial intelligence rather than working within the existing inventorship framework. He deliberately chose to list DABUS as the sole inventor on the patent applications.[3] He also asserted that he did not contribute to the conception of the inventions.[4] As the Federal Circuit hinted, Dr. Thaler may have had more success if he listed himself as the inventor rather than the AI.
In this article, I explore reasons why there may be a better case for the patentability of AI-assisted inventions.
The person interacting with an AI could be found to have conceived of an AI invention.
Conception has been called “the touchstone of inventorship.”[5] It “is the formation in the mind of the inventor, of a definite and permanent idea of the complete and operative invention. . . .”[6]
If an AI cannot be an inventor because it is not a natural person, then it likely follows that an AI cannot be capable of conception. An AI surely does not have a “mind” like a natural person. An AI may instead be thought of as merely a tool that outputs results to be interpreted by people.
Therefore, for an invention created by AI, the only possible entity that could conceive of the invention would be a person who interacts with the AI. The first person who sees and appreciates an idea that is output by an AI as an invention would logically be the first person to have conceived the invention. Effectively, the person would have “discovered” the invention.[7]
Additionally, AI that exists today generally does not create things spontaneously. People request the assistance of an AI to accomplish certain tasks by training the AI, optimizing its parameters, and in some cases providing generative prompts.[8] All of these contributions would further support that a person should properly be credited with conceiving the output of an AI.
Inventions are patentable no matter how much effort is required.
That a person would have made little effort outside of merely appreciating that something output by an AI is an invention makes no difference to the patentability of the invention. 35 U.S.C. §103 provides that “[p]atentability shall not be negated by the manner in which the invention was made.”[9] In fact, many famous inventions were discovered by pure luck. For example, Percy Spencer discovered the idea of a microwave oven when he was standing near an active magnetron (a device for generating microwaves) and noticed that a chocolate bar melted in his pocket.[10]
Private AI output is not prior art.
The output from an AI also could not be considered prior art. Generally, 35 U.S.C. §102(a) provides that “A person shall be entitled to a patent unless—(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.”[11] As long as an AI system does not publicly share its inventions, its output could not qualify as prior art that could anticipate or render obvious the invention.[12]
Allowing a person to patent an AI-assisted invention would promote progress.
The Progress Clause of the Constitution provides that the purpose underlying the Patent Act is “to promote the progress of science and the useful arts.”[13] Although the Federal Circuit dismissed Dr. Thaler’s constitutional arguments about allowing AI inventors to promote progress as speculative,[14] much has changed in the last year. AI systems like ChatGPT and Dall-E have exploded in popularity.[15] The emergence of the internet provides a useful Constitutional comparison. The internet did not exist when the First Amendment was enacted, but online speech is covered by the First Amendment.[16] The risks with not adapting the law to account for AI or AI-assisted inventorship are more tangible than ever.
[1] Thaler v. Vidal, 43 F.4th 1207, 1213 (Fed. Cir. 2022).
[2] Id.
[3] Id. at 1209-1210.
[4] Id.
[5] Ethicon, Inc. v. United States Surgical Corp., 135 F.3d 1456, 1460 (Fed. Cir. 1998) (quotations omitted).
[6] Id.
[7] See 35 U.S.C. 100(a) (defining invention as an “invention or discovery”).
[8] Prompt engineering is a new field in the AI industry. See https://github.com/dair-ai/Prompt-Engineering-Guide.
[9] 35 U.S.C. § 103.
[10] https://www.discovermagazine.com/technology/5-inventions-that-were-discovered-by-accident.
[11] 35 U.S.C § 102(a).
[12] 35 U.S.C. §§ 102, 103.
[13] U.S. Const. art. I, § 8, cl. 8.
[14] Thaler, 43 F.4th at 1213.
[15] https://www.mckinsey.com/featured-insights/mckinsey-explainers/what-is-generative-ai
[16] Reno v. Am. Civ. Liberties Union, 521 U.S. 844, 870 (1997).
[View source.]
