Photo: Shutterstock

 

Back in 1955, John McCarthy coined the term "artificial intelligence" to represent the science of developing intelligent machines. The following year, McCarthy established AI as a field when he organized the Dartmouth Conference to operate under the "conjecture that every aspect of learning or any other feature of [human] intelligence can in principle be so precisely described that a machine can be made to simulate it." While AI as a term and science may not be new, certain legal issues surrounding the patenting of AI inventions certainly is at its infancy.

What are "AI Inventions"?

The term AI inventions is an umbrella that covers two categories: inventions that utilize AI and inventions that are created by AI. The first category—inventions that utilize AI—broadly covers inventions of software and/or hardware used to run the AI. These inventions relate to AI algorithms, collection, storage, and use of training data (i.e., input data into the AI), hardware used to execute the AI algorithms or operate with training data, and applications of AI (i.e., uses for the data output by AI). In short, inventions that utilize AI "include anything under the sun that is made by man," subject to exclusions for laws of nature, physical phenomena, and abstract ideas, of course.

The second category—inventions that are created by AI—on the other hand, encompasses solutions to problems identified by the AI itself or those identified to the AI by a human. One of the few examples available today are the solutions identified by Device for the Autonomous Bootstrapping of Unified Sentience (DABUS), a Creativity Machine. Its creators describe DABUS as the result of the constant combining and detaching of multiple disconnected neural networks that, eventually, coalesce into structure representing complex concepts. These concepts connect with other concepts to represent the anticipated consequence of any given concept in a manner reminiscent of what can be considered a stream of consciousness. Through this process, DABUS "invented" a new type of beverage container and a flashing device used to attract attention, both of which are subject of patent applications filed with the U.S. Patent and Trademark Office (USPTO), the European Patent Office (EPO), and the U.K. Intellectual Property Office (UKIPO). All of these applications have since been rejected because the current patent law of the three jurisdictions limits inventors to "natural persons."

Challenges in Patenting AI Inventions

Each category of AI Inventions carries its own patentability challenges. Inventions that utilize AI, for example, are patentable assuming the claims meet all the necessary requirements of patentability, i.e., novelty, non-obviousness, and sufficient written description/enablement. In fact, DABUS, discussed above, is itself patented.

Fig. 1, U.S. Pat. No 5,659,666 for DABUS. Fig. 1, U.S. Pat. No 5,659,666

Applicants seeking to patent such inventions will face the ever-evolving patent-eligibility limitations in their applicable jurisdiction. For example, in the U.S., applicants must navigate agency and case law developments to avoid claiming patent-ineligible abstract ideas, especially where the claims are directed to algorithms or specific uses of what may be perceived as an existing technology.

Inventions that are created by AI, on the other hand, present additional challenges. U.S. patent law, for example, defines the act of invention as the mental step of conception. (Per Townsend v. Smith, conception is the "formation in the mind of the inventor of a definite and permanent idea of the complete and operative invention." According to the USPTO, only a "natural person" can engage in the mental step of conception and, thus, only a "natural person" can be named an inventor.

The inquiry does not end here, however. A possible mechanism to secure patent protection in these circumstances may come down to how one views a solution to a problem that is created by AI. The premise is rather simple: A "natural person" developed the algorithm/wrote the code that created the AI. The "natural person," or the AI itself, then identified a problem which, in turn, led to the AI to derive the solution. The question as to whether the solution is patentable, under the current legal rubric, is whether the "natural human" or the AI invented the solution? Did the "natural person"—one that created the AI—invent the solution, because the solution could not have been identified by the AI but-for the "natural person" and the AI is merely reducing the solution to practice? Or, on the other hand, is the solution the result of the AI's "organic" growth and maturation, which is outside of the control or direction of the "natural person"? The current state of the law does not provide a clear answer.

Protecting Inventions that are Created by AI

The current challenges to patenting inventions that are created by AI do not mean that such inventions should be left unprotected. It is possible that legislative changes may be needed as the full scope of AI's capabilities become better understood. In the meantime, however, trade secret protection may be a viable option where it is determined that patent protection is unavailable. The invention can be maintained as a trade secret so long as its public disclosure can be prevented. Knowledge of the invention can be limited to reduce the breadth of internal disclosure and employees "in the know" can be subject to non-disclosure agreements or other restrictions to reduce the likelihood of misappropriation. Of course, public disclosure may not be preventable, especially where the invention, or its use, is detectable outside of the organization. When this happens, the long-term viability of the trade secret may be jeopardized by the possibility of independent invention by another outside the organization or reverse engineering.

Another option is to file applications where both human and AI conceived different parts of the invention. Because; in this case; a "natural person" has conceived the invention, the USPTO should not reject these applications for lack of inventorship.

We are only beginning to scratch the surface of the various legal implications that AI will have on patent protection. While the current law permits patenting of inventions that utilize AI, significant challenges exist when it comes to patenting inventions that are created by AI. The good news is that these challenges are not insurmountable and can, and likely will change as more and more AI applications are being filed and AI takes a more prominent position on the center stage.

 

This article reflects only the present personal considerations, opinions, and/or views of the authors, which should not be attributed to any of the authors' current or prior law firm(s) or former or present clients.

Eugene Goryunov is a partner in the Intellectual Property Practice Group in the Chicago office of Haynes and Boone and an experienced trial lawyer that represents clients in complex patent matters involving diverse technologies. He has extensive experience and regularly serves as first-chair trial counsel in post-grant review trials (IPR, CBMR, PGR) on behalf of both Petitioners and Patent Owners at the USPTO.

David L. McCombs is a partner in the Intellectual Property Practice Group in the Dallas and Washington, D.C. offices of Haynes and Boone and is primary counsel for many leading corporations in inter partes review (IPR) and is regularly identified as one of the most active attorneys appearing before the Patent Trial and Appeal Board (PTAB).

Dina Blikshteyn is of counsel in the Intellectual Property Practice Group in the New York office of Haynes and Boone. Dina's practice focuses on post grant proceedings before the U.S. Patent and Trademark Office, preparing and prosecuting domestic and international patent applications, as well as handling trademark and other IP disciplines.

Jonathan Bowser is of counsel in the Intellectual Property Practice Group in the Washington, D.C. office of Haynes and Boone. He is a registered patent attorney focusing on patent litigation disputes before the Patent Trial and Appeal Board (PTAB) and federal district courts.

Raghav Bajaj is a partner in the Intellectual Property Practice Group in the Austin office of Haynes and Boone. His practice focuses on patent office trials before the Patent Trial and Appeal Board (PTAB), including inter partes review (IPR) and covered business method (CBM) review proceedings, representing both petitioners and patent owners.

Angela Oliver is an associate in the Intellectual Property Practice Group in the Washington, D.C. office of Haynes and Boone. She focuses her practice on patent appeals before the U.S. Court of Appeals for the Federal Circuit and post-grant proceedings before the U.S. Patent and Trademark Office.