Petition for Writ of Certiorari filed in DABUS AI-as-Inventor Case

By Aaron Gin --

Dr. Stephen Thaler, Ph.D., a computer scientist and inventor, has petitioned the Supreme Court of the United States to consider the question of whether the Patent Act restricts the definition of an "inventor" to human beings.  The petition represents an opportunity for the Court to clarify whether an artificial intelligence (AI) can be an inventor under U.S. patent law.

The primary reasons provided by Thaler for the Court to grant certiorari include:  1) describing a conflict between the lower court decisions and the text of the Patent Act; 2) pointing out conflicts with prior Supreme Court jurisprudence; and 3) a policy argument highlighting the importance of patentability of AI-generated inventions in the U.S. and worldwide.

Thaler is the creator of DABUS (Device for the Autonomous Bootstrapping of Unified Sentience), which is a combination of multiple artificial neural networks.  The various neural networks of the DABUS system were each programmed to represent discrete concepts such as "temperature", "enjoyment", or "survival."  DABUS was then trained under external supervision from a human trainer to form short consequence chains, such as taking a drink at an appropriate temperature results in enjoyment.  Subsequently, during an unsupervised activity, DABUS autonomously extended the consequence chains into longer, more-complex chains that resulted in positive outcomes.

Based on this architecture and training, Thaler claimed that DABUS autonomously developed two novel concepts:  "Neural Flame," an emergency beacon that flashes in a desired pattern to attract the attention of rescuers, and "Fractal Container," a beverage container that improves grip and better regulates heat transfer to increase user enjoyment.  These concepts formed the basis of two patent applications U.S. Application Nos. 16/524,350 and 16/524,532.  Thaler filed the patent applications with the USPTO, listing the sole inventor as "DABUS" under "given name" and "Invention generated by artificial intelligence" under "family name" in an Application Data Sheet.  Thaler also included a "Statement on Inventorship," which described some of the functional aspects of DABUS and argued why the AI should be considered as an inventor under the Patent Act.

The USPTO issued a Notice to File Missing Parts to amend the ADS.  Through several rounds of petitions, Thaler asked the USPTO to vacate the Notice and provided further arguments.  However, the USPTO pointed to numerous references to an inventor as a "person" in Title 37 of the Code of Federal Regulations and the MPEP definition of "conception" as "the complete performance of the mental part of the inventive act," and dismissed Thaler's request for reconsideration in a final written decision.

Thaler challenged the USPTO's decision in a suit filed against the USPTO in the Eastern District of Virginia.  The District Court judge granted summary judgment for the defendant based on the plain statutory language of the Patent Act (35 U.S.C. §§ 100 and 115) and Federal Circuit (Univ. of Utah v. Max-Planck-Gesellschaft, 734 F.3d 1315 (Fed. Cir. 2013)) authority.  Thaler appealed to the Federal Circuit, which affirmed the District Court in an August 2022 decision.  In brief, the CAFC agreed that an inventor must be a human being, barring AIs from inventorship.

Through his counsel, Thaler filed the Petition on March 17, 2023.  The case was docketed on March 21, 2023 as Supreme Court No. 22-919 and captioned "Stephen Thaler, Petitioner v. Katherine K. Vidal, Under Secretary of Commerce for Intellectual Property and Director, United States Patent and Trademark Office, et al."  A response from the Director of the USPTO is due by April 20, 2023.

For additional information, please see:

• Supreme Court Docket No. 22-919
• Stephen Thaler v. Vidal; Petition for Writ of Certiorari

Comments

[Max Drei]

To "conceive" an invention it is not enough to generate a design for an article of commerce. One has to i) assess the design and "conceive" that it might indeed be an embodiment of a patentable inventive concept, and then ii) output in readable form what that conceived concept is.

But what if, in the near future, an AI not only outputs a product design but also a set of claims fitting to the design, and oven-ready for filing at the USPTO. Has the AI then done enough to be entitled to be recognised as having "conceived" an invention?

[skeptical]

Mr. Max Drei,

It is beyond trivial for an AI to ALSO perform an assessment, as well as to output in readable form.

Why do you strain so much to deny the non-human aspect of an innovation without a human being the innovator?

At the minimum in the US Sovereign, our laws protect innovation by humans - under the foundation of the Lockean exchange of Quid Pro Quo.

As in the parallel case of the Naruto (colloquial recalling here: https://en.wikipedia.org/wiki/Monkey_selfie_copyright_dispute ) the mere presence of creativity is not necessarily recognized by a legal grant of intellectual property.

I am genuinely perplexed by the level of effort to NOT understand the legal foundations that control.