chevron-down Created with Sketch Beta.

Landslide®

June/July 2024: Transactions

Perspective: Inventions—Not Trees in the Forest

Steven Caltrider

Summary

  • Inventorship requires human recognition and appreciation of a complete and operable invention; only a person can have an original conception of an invention and be an inventor.
  • The USPTO inventorship guidance focuses on whether the human contribution to an invention is significant enough to qualify for a patent when AI also contributed.
  • By introducing “significant contribution” as the bar for whether there is an invention, the guidance conflates obviousness and conception.
Perspective: Inventions—Not Trees in the Forest
DonLand via Getty Images

Jump to:

One of the most famous questions in philosophy that is apropos to modern intellectual property (IP) law is this: If a tree falls in a forest and no one is around to hear it, does it make a sound? The question is apropos to modern IP law because of parallels to the current discussion of inventorship of artificial intelligence (AI)–assisted inventions. The IP corollary is if an AI machine generates all the elements of a claim and no one is around to recognize or appreciate that there is a complete and operable invention, is it an invention? Unlike the debate of the tree in the woods, the answer to the IP question is a clear and resounding no. More importantly, a careful study of this question also informs how the question of inventorship should be addressed under current law and the U.S. Patent and Trademark Office (USPTO) Inventorship Guidance for AI-Assisted Inventions.

The starting point of the analysis is the U.S. Constitution. Article 1, Section 8 provides that Congress has the power “[t]o promote the progress of science and useful arts, by securing for limited times to authors and inventors the exclusive right to their respective writings and discoveries.” Section 100 of the Patent Act defines inventor as follows: “The term ‘inventor’ means the individual or, if a joint invention, the individuals collectively who invented or discovered the subject matter of the invention.” Section 115 sets out the inventor’s oath and declaration, which provides in part that “such individual believes himself or herself to be the original inventor or an original joint inventor of a claimed invention in the application.” Section 116 sets out the conditions for joint invention of two or more persons. Section 256 provides for the correction of a named inventor (“Whenever through error a person is named in an issued patent as the inventor, or through error an inventor is not named in an issued patent . . . ”). Collectively these laws make clear that an inventor is a person, but they do not fully answer the pertinent question of the day: Who is an inventor in an AI-assisted discovery? Or stated differently, is an AI-assisted discovery not an invention because it lacks a legally recognized inventor?

Who Is an Inventor?

Inventorship is “a question of who actually invented the subject matter claimed in a patent.” And “[c]onception is the touchstone of inventorship, the completion of the mental part of invention.” It defines the “legally operative moment of invention.” Conception is “the originating of something in the mind.” As such, it is an inherently human activity. It is a non sequitur to suggest that a computer can legally “conceive” of an invention. Only a person can have an original conception of an invention and be an inventor.

Decades of law provide a solid foundation to understand conception and therefore what constitutes an invention under U.S. law. Conception within the meaning of the patent law is “the formation in the mind of the inventor of a definite and permanent idea of the complete and operative invention as it is thereafter to be applied in practice.” Notably, the inventor must recognize or appreciate the invention to find conception. In other words, a person who generates a composition, process, or any embodiment but does not recognize or appreciate doing so is not an inventor. There is no conception under the law. The details of the case law are informative. In Invitrogen Corp. v. Clontech Laboratories, a gene was modified using a technique called “random mutagenesis.” The technique makes random changes to the gene to discover a modified gene having certain desired properties. In Invitrogen, roughly 100 mutants were prepared randomly and sequenced, i.e., the chemical structures of the mutants were determined. However, the court found that merely sequencing the mutant gene was not enough, and that there was no conception until the inventor appreciated and recognized the invention—i.e., the law requires the inventor to recognize which genes out of those randomly generated have the desired properties.

This bears directly on AI-assisted inventions. The distinction between an AI-assisted invention and genes randomly generated is a difference in the research tool. AI-computing power is far more advanced and efficient in identifying putative operable embodiments. But, until such operability is recognized and appreciated, there is no conception. Without conception, there is no invention. Only when a skilled person looks at the AI output and recognizes and appreciates that the embodiment is complete and operable is there a conception. The degree of further testing necessary for recognizing and appreciating that the embodiment is complete and operable will, of course, depend on the facts. A new chemical entity may require testing whereas a new axle may be recognizable without testing by the skilled artisan. In both circumstances, however, the “legally operative moment of invention” is when the skilled artisan recognizes that the AI-assisted output represents a complete and operable invention.

An embodiment generated by an AI algorithm is not an invention. If the embodiment resides within the computer and/or on the printer and no one looks at it, then there is no conception and therefore no invention. Only when the inventor recognizes and appreciates that the AI-output represents a complete and operable invention is there conception and a legally recognized invention under the law. The proper test for answering the question “who is an inventor” is to assess who conceived of the claimed invention.

USPTO Inventorship Guidance

Most respectfully, the USPTO gets off on the wrong foot by looking to the body of law directed to a different question than who legally conceived of the invention and therefore is the inventor. The USPTO relies heavily on the Pannu v. Iolab Corp. decision. The Pannu decision is directed to the question of what contribution is sufficient for someone to be a joint inventor and sets out factors that determine whether a person made a significant contribution as a joint inventor. The apt question for the USPTO guidance is who legally conceived of the invention—a related but different issue. The guidance also relies on the Solvay v. Honeywell case to dismiss the role of recognition and appreciation. Respectfully, Solvay is not a conception case based on recognition and appreciation. The question before the court was whether those in the U.S. who received the invention previously conceived in Russia could be an inventor. They could not, as there was no original conception, and therefore they were not “another inventor” under § 102(g)(2). An AI-assisted invention is not a previously conceived invention under the law.

USPTO Examples on Inventorship

Along with the guidance, the USPTO published two examples as being instructive to applicants. The examples are helpful to understanding the guidance and advancing the discussion of this important issue. Example 1, which is directed to a remote control (RC) car, illustrates the challenges of applying Pannu to the question whether there is a conception of the invention and therefore an inventor. It conflates obviousness and conception because of the focus on a “significant contribution” to identify an inventor. Example 1 provides using the output of the AI computer:

Ruth and Morgan reviewed the output of Puerto5 [the AI algorithm] and agreed the design should work for the RC car. However, recognition and appreciation of the invention without any significant contribution to the conception of the claimed invention is not sufficient for inventorship. Therefore, their contribution amounts to recognizing the problem and asking an AI system to solve the problem. For these reasons, Ruth and Morgan’s contribution was not a significant contribution to the conception of the invention.

This is incorrect as a matter of law. As importantly—like the current disarray in § 101 jurisprudence—it conflates conception with considerations regarding obviousness. Conception is completed when “Ruth and Morgan reviewed the output of Puerto5 and agreed the design should work for the RC car.” Some inventions may require wet-lab confirmation to validate. Some may be “reviewed” and “agreed” by simply applying skill in the art. The analysis of “significant contribution” that requires something more is akin to requiring a flash of genius or some sort of inventive step to find conception of an invention, which is not the law. On these facts, the RC car may be obvious because there was nothing more than the publicly available AI tool (Puerto5) that generated the axle as its output. But, the law recognizes an original conception at “the legally operative moment of invention” when it was “reviewed” and “agreed.”

Supporting Innovation

Stepping back from the law, is it good public policy to expand the Pannu factors to, in essence, serve as a substantive ground of patentability? The USPTO guidance introduces “significant contribution” as the bar for whether there is an invention, not just whether a contribution is sufficient to be a joint inventor. For a few reasons, finding an inventor at the moment of a legally recognized conception and leaving § 102 and § 103 as the substantive test for patentability reflects better IP policy supporting innovation.

First, the framework of the statute does not provide for inventorship to be a substantive issue of patentability (beyond requiring a human inventor(s)). Section 100 defines invention, inventor, and joint inventor. Section 101 addresses patentable inventions and notes that “[w]hoever invents or discovers . . . may obtain a patent therefor, subject to the conditions and requirements of this title.” Section 101 does not set a bar over which a putative inventor must exceed. In contrast, the conditions for patentability are found in § 102 and § 103. Thus, there is no clear statutory basis to find an invention unpatentable for want of an inventor making a “significant contribution.” To the contrary, the statute expressly provides that a “discovery” is entitled to a patent. Part and parcel of discovery is recognizing and appreciating a complete and operable invention.

Second, identically claimed embodiments should be treated the same. Consider two patent applications both claiming an embodiment. One patent applicant discovered the embodiment through laborious trial and error (or random mutagenesis). The other patent applicant discovered the embodiment using AI. The law should not penalize the applicant who utilized the most up-to-date tools and techniques to discover the embodiment. In some respect, doing so is the antithesis of the very incentive to innovate that the patent system is intended to foster. Furthermore, applying the law in this way runs counter to the goal of transparency and certainty. It creates a hidden “land mine” for an investor or prospective licensee. A prospective licensee has no way to ascertain from the face of the patent whether AI was used and if used to what extent the putative inventors made a significant contribution. The reminder of the duties owed to the USPTO will not solve this problem as the role of AI and whether there is a significant contribution will only be finally resolved in litigation. The test of “significant contribution” like “something more” under § 101 introduces more uncertainty in a system that thrives best with predictability and certainty.

Finally, the lessons learned from the last decade or so with § 101 litigation support not conflating the question of conception with the substantive assessment of patentability. Creating another issue that can only be finally resolved through discovery and litigation will increase costs as well as uncertainty. The cost and uncertainty will chill prospective inventors and investors and push the patent system further out of reach for some.

A reliable and durable patent right is the proven basis for American prosperity. The USPTO and the courts should be mindful not to increase the costs and uncertainty in the patent system when it is not clearly dictated by statute or precedent.

©2024. Published in Landslide, Vol. 16, No. 4, June/July 2024, by the American Bar Association. Reproduced with permission. All rights reserved. This information or any portion thereof may not be copied or disseminated in any form or by any means or stored in an electronic database or retrieval system without the express written consent of the American Bar Association or the copyright holder.

    Author