Generative artificial intelligence (AI) systems, such as ChatGPT, can output new content based on user input. If such new content forms part of a new “invention” (i.e., part of a patent claim), does the AI system need to be listed as an “inventor”? There are currently two schools of thought for answering this question. The below article explores this further.
It may all come down to “conception” – that is, whether an AI system can mentally conceive and contribute to an invention. Conception is important because, at least under U.S. patent law, an inventor is one who contributes to the conception of at least one claim element of a given patent claim (i.e., an invention). Assuming that an AI system is able to “conceive” (which is arguable with respect to generative AI), then it is possible that an AI system could “contribute” to a given claim element and satisfy the requirement for an “inventor.”
Thus, according to one school of thought, if an AI system can be considered to have conceived of an idea that becomes part patent claim element, then, the AI system needs to be listed as an inventor.
However, according to a second school of thought, an AI system is merely a tool that is incapable of “conception” outside of human input, such that the AI system would NOT need to be listed as an inventor.
The below article explores these schools of thought: the case for and against the requirement of listing an AI as an inventor based on aspects of conception, inventorship, and joint inventorship under U.S. patent law.
1. The Case For Listing an AI as an Inventor: where an AI System is considered to be able to “Conceive” of an Invention
In August 2022, the U.S. Court of Appeals for the Federal Circuit affirmed the USPTO’s decision that U.S. Patent law requires that a patent application list at least one human inventor. See Thaler v. Vidal, 43 F.4th 1207 (Fed. Cir. 2022) (cert denied). That is, an artificial intelligence (AI) system (e.g., such as DABUS, as was the case in Thaler) cannot be listed as the sole inventor on a patent application. The Federal Circuit made clear, however, that it did not address at the time the topic of human-and-AI joint inventorship, i.e., “Moreover, we are not confronted today with the question of whether inventions made by human beings with the assistance of AI are eligible for patent protection.” Id. (Original emphasis).
Thus, from the Thaler decision, we know that an AI inventor cannot be listed alone, but does an AI system need to be listed as a joint inventor where there was human-and-AI interaction for the development of an invention?
Under one school of thought, where an AI system is believed to be able to “conceive,” then the answer would be “yes.” After all, U.S. Patent Law requires all inventor(s) to be listed. Thus, if an AI system is, in fact, an “inventor” (by virtue of contributing to the conception of a claim element), then such an AI system must be listed. If it is not listed, then presumably any patent issued by virtue of listing human inventor(s) only (and without the AI system inventor) could presumably be at risk of invalidity (or at risk of being rendered unenforceable as the case may be).
For example, in a scenario where a human used an AI system to generate inventive subject matter that was added to a patent claim (i.e., an invention), then it is arguable that the AI system is at least a joint inventor. In such a hypothetical, joint inventorship would need to be considered based on the type or otherwise “quality” of conception contributed to a claim. Recently, on May 2, 2023, the Federal Circuit in HIP, Inc. v. Hormel Foods Corp. found that a putative inventor must “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.” Appeal 2022-1696 (Fed. Cir. May 2, 2023) (citing Pannu v. Iolab Corp., 155 F.3d 1344, 1351 (Fed. Cir. 1998) (emphasis added). In that case (which was not AI related), the subject patent involved claims to recite a method of making precooked meat. The patent identified only individuals at Hormel as inventors, though they collaborated with an individual at HIP. The putative inventor (at HIP) asserted he was a joint inventor of one feature of one of the lengthy independent claims. The Federal Circuit considered the assertion and the specification of the patent to conclude that the alleged contribution (use of a particular type of oven) was not qualitatively significant to warrant joint inventorship status. In particular, the purported contribution — which related to preheating meat in an infrared oven — was minuscule “in view of the whole invention.”
Thus, applying this to AI joint inventorship questions, a few scenarios can arise with respect to AI-and-human interaction, each of which turns on whether the AI system and/or human made a quality contribution to the conception of the invention:
- In a scenario where an AI system contributes to a claim element that is “significant in quality,” and where a human contributes to the “conception” of a claim element that is “insignificant in quality” (as in the HIP case), such scenario would cause the sole inventor to be the AI system, which, accordingly to current law is not possible. See Thaler v. Vidal.
- If the hypothetical for case 1 (above) were reversed (with the human making a qualitatively significant contribution/conception and the AI system making a qualitatively insignificant contribution/conception), then the listing of the human only would be feasible under current U.S. jurisprudence.
- Finally, in the hypothetical where both the AI system and a human make contributions that are “significant in quality,” then both would need to be listed, at least according to the first school of thought.
Under this school of thought, it is possible that the USPTO may require applicants to notify, with a patent filing, whether any AI system was used to generate a portion of the invention. This would be similar to the Copyright Office’s recent “duty” to disclose AI-generated works or portions thereof. In particular, the Copyright’s guidance on AI-generated works describes a “duty” to inform the Copyright Office when a work generated by a generative AI (e.g., Midjourney) is sought to be copyrighted. That is, the Copyright’s guidance states that: “applicants have a duty to disclose the inclusion of AI-generated content in a work submitted for registration and to provide a brief explanation of the human author’s contributions to the work.” For more on this topic, see this article describing a case involving a comic book that included both AI-generated works (images) and human-generated works (e.g., text) and the Copyright Office’s treatment of the same. See U.S. Copyright Office Partially Allows Registration of Work having AI-generated Images (“Zarya of the Dawn”).
2. The Case Against Listing an AI as an Inventor: where AI is merely a tool of a human inventor
The second school of thought assumes that because any given AI system cannot “conceive,” then such AI system need not be listed as an inventor pursuant to current U.S. patent law.
Following this logic, and in view of the Thaler decision, because an AI system cannot be an “inventor” under current patent laws (35 U.S.C. §§ 101 and 115), then such laws do not require the listing of an AI system as an inventor. Only the human needs be listed, even if the human used the AI system to develop an invention. Under this view, the human is deemed to have used the AI system merely as a tool. Moreover, according to this theory, the AI system would have been unable to produce any output without the input, setup, or otherwise direction of the human operator.
Said another way, even if an AI system were hypothetically able to contribute to the conception of an invention (which was accepted as true in the Thaler decision, see id. at fn. 2), because it was determined that an AI system could not ultimately qualify as an “inventor” then if follows that any such conception or contribution would be irrelevant to the question of inventorship. The same would presumably hold true for the joint inventorship question, too.
For these reasons, in a hypothetical where a natural person uses an AI system (e.g., as a tool) to develop a patentable invention, then such a natural person would follow current patent laws by listing himself or herself as an inventor and satisfying the requirements of Section 115 but would not need to list the AI system as a joint-inventor or somehow fulfill Section 115 with respect to the AI system. For example, the Thaler decision informs that Section 115 relates to natural persons only. See id. at 1211 (“In describing the statements required of an inventor when applying for a patent, the statute consistently refers to inventors and co-inventors as ‘individuals.’”) (citing 35 U.S.C. § 115). Accordingly, if an AI system cannot be an “inventor” per Thaler, then an AI system would not be subject to the requirements of “inventorship.” For example, a given AI system would not be listed as an inventor or be required to adhere to the oath or declaration strictures for inventorship required by Section 115.
Thus, according to this school of thought, where the AI system cannot be an inventor or “conceive,” then it is considered acceptable not to list the AI system as an inventor, and the AI system is treated merely as a tool in an invention discovery to aid the human inventor.
The issue of inventorship, especially joint inventorship, has been described by at least one court as “one of muddiest concepts in the muddy metaphysics of the patent law.” Mueller Brass Co. v. Reading Industries, Inc., 352 F. Supp. 1357, 1372 (E.D. Pa. 1972).
Add to that the topic of AI inventorship, and the question becomes “muddier” still. Time will tell as to which one of the above schools of thought will win out in the courts. But this author believes that the answer lies in how the topic of “conception” is defined and whether an AI system can, in fact, conceive of an invention or at least a portion thereof.
Subscribe to get updates to this post or to receive future posts from PatentNext. Start a discussion or reach out to the author, Ryan Phelan, at email@example.com (Tel: 312-474-6607). Connect with or follow Ryan on LinkedIn.