Artificial intelligence exists on a wide-ranging spectrum. On one end, grammar and spell check can detect and correct grammatical errors and typos in written text. On an opposite end, generative artificial intelligence such as ChatGPT, Bard, DALL-E 2, Midjourney and Claude can create complex images and text. As Generative AI rapidly becomes a tool used in everyday life, the United States Patent and Trademark Office (USPTO) has now found itself at the center of a debate on whether artificial intelligence can be listed as an “inventor” on a patent.

Since the release of DALL-E 2 and ChatGPT last year, many have used generative AI for fun like generating a photo-realistic image of the family dog or creating a poem about an arbitrary subject. However, generative AI recently passed a Bar exam and roasted comedians, raising questions about whether white collar professionals and artists should fear for their jobs. It comes as no surprise then that companies and individuals are beginning to leverage generative AI to create novel and potentially patentable concepts.

Pharmaceutical company Novartis has created a generative model called JAEGER for creating novel molecules. JAEGER, like most AI, requires input from humans to generate such molecules. Corey Salsberg, Vice President and Global Head of Intellectual Property at Novartis, contended at a recent U.S. Senate hearing that “whether AI can or should be awarded a patent is not the right question at this time. … The right question is whether AI is even inventing.” Salsberg points out that artificial intelligence is currently better thought of as lab tools, likening it to a calculator or microscope. Although JAEGER can generate novel molecules, Salsberg states that generating is not the same as inventing where a human needs to “define the problem, guide the solution, and recognize, appreciate and synthesize the results.”

But what if the entire inventive process occurs within an artificial intelligence model? While testifying at the same Senate hearing, Ryan Abbott, Professor of Law and Health Sciences at the University of Surrey School of Law, said we are already there. Abbott demonstrated a “fractal container” created by an AI called DABUS, which was built by Stephen Thaler. DABUS is trained on general knowledge and is designed to combine simple ideas and concepts into more complex ideas. In Professor Abbott’s example, DABUS identified a beverage container based on fractal geometry that would have increased surface area, thereby improving heat transfer, the opposite of a thermos, and be easy to grip.

The United Kingdom Patent Office, the UK counterpart to the USPTO, stated that the fractal container was “new, inventive, and useful” which are the major requirements to obtain a patent in the UK. A UK Supreme Court case is currently pending to determine whether a patent can be awarded to Stephen Thaler for the fractal container along with a different invention created by DABUS because DABUS is listed as the inventor on the application. DABUS been awarded a patent in South Africa, and applications for DABUS inventions are pending in other countries (including China, Japan and Israel, among others).

The United States has refused to consider the container for patentability with DABUS disclosed as the inventor. Stephen Thaler, with Abbot among his representation, challenged the USPTO’s decision in the U.S. Court of Appeals for the Federal Circuit. The Federal Circuit rejected the challenge and held that only a natural person could be listed as an inventor on a patent. The U.S. Supreme Court declined to review the ruling, leaving the Federal Circuit’s decision as the last word, for now.

Abbott stated during the Senate hearing that under the Copyright Act, companies can be listed as authors, so it is not so unusual for non-human entities to be listed for doing what is traditionally a human’s work. Abbott clarified that “no one is arguing that the AI should own the patent,” but he said “it is a bedrock principle of property law that you own property made by your property…. [I]f I have a 3D printer make a physical beverage container; I own that container. There’s no reason why I should be any less entitled to the intellectual property.”

Regardless of whether or not generative AI is currently capable of inventing patentable subject matter, it is not a stretch to imagine that it will be able to soon. According to Senator Chris Coons, chairman of the Senate Judiciary Subcommittee on Intellectual Property, the United States needs to ensure that innovation generated by AI systems remain in the United States or risk “incentivizing innovators to turn to other countries with more favorable laws to protect their AI-generated inventions.” In addition to laws regulating AI generally, the United States will need legislation clarifying rules for inventorship and whether and how to accommodate software that has meaningfully contributed to conception of an inventive concept. With the recent Senate hearing on artificial intelligence and patents, along with two public meetings earlier this year being held by the USPTO called “A.I. Inventorship Listening Sessions,” the United States is in the process of determining the course of action that it will take. Only time will tell whether it is a hybrid system where humans and artificial intelligence are listed as co-inventors, only artificial intelligence are inventors or only humans are inventors. There also is a chance there will be no need for a patent system once AI is fully capable of inventing.

[View source.]