What if an artificial intelligence algorithm designs a new drug or identifies a novel drug target? Who invented the subject matter: the algorithm, the scientist who relies on the algorithm, or both?

The biopharmaceutical industry relies heavily on intellectual property protections to safeguard its investments. One issue commonly facing the industry is inventorship, i.e., who among the several teams of scientists involved in a project should be credited as an inventor. Inventorship has been considered “one of the muddiest concepts in the muddy metaphysics of the patent law.” Indeed, in the biopharma industry, inventorship disputes may be even more complex given the unpredictable nature of drug discovery and development.

Now, with the advent of AI-assisted biopharma innovation, the inventorship issue may have reached a new level of complexity. Biopharma companies across the globe are expanding their research and development capabilities to include the use of AI in, for example, drug discovery, clinical trials, and personalized medicine.

This expansion has led some to ask whether an AI-machine or algorithm must be credited with a portion of inventorship. We will explore the laws, regulations, and decisions made by Congress, the USPTO, and U.S. courts to gain insight into how inventorship should be determined in the context of AI-assisted biopharma innovation.

Only Natural Persons Can Be ‘Inventors’

Many advances in the biopharma space occur through the collaborative efforts of multi-disciplinary teams. Now, more often, those teams can include AI-assisted algorithms and screening methodologies.

Should this non-human contribution be granted inventor status? Congress, through the America Invents Act, defined “inventor” as “the individual or, if a joint invention, the individuals collectively who invented or discovered the subject matter of the invention.” Congress further specified that “[w]hen an invention is made by two or more persons jointly, they shall apply for patent jointly and each make the required oath.”

The USPTO, clarifying the requirements for the inventor oath, included several anthropomorphic requirements such as gender and sentience: “identify the inventor … by his or her legal name … [and] include a statement that the person executing the oath … believes the named inventor … to be the original inventor.”

As for the courts, when determining the scope of patent eligible subject matter, the U.S Supreme Court relied on congressional committee reports and testimony stating “a person may have invented a machine … which may include anything under the sun that is made by man.” Indeed, courts have clarified that only natural persons can be inventors and not corporations. Thus, through interpretation and analogy, AI cannot meet these requirements and therefore cannot qualify as an inventor under current U.S. patent laws, regulations, and legal precedent.

Conception of Novel AI-Assisted Biopharma Invention

Conception is the touchstone to determining inventorship.” It is the “formation, in the mind of the inventor, of a definite and permanent idea of a complete and operative invention.” It requires a nexus between the invention and the inventor’s belief in the invention. While AI might assist in identifying a molecule or target, it is difficult to conclude that the AI system appreciates a complete and operative invention.

In the biopharma industry, a scientist could, theoretically, identify a target, define that target in an AI-algorithm that incorporates the company’s proprietary training data as well as data from the public domain, and review the results for all compounds that meet set parameters for that target (e.g., safety, efficacy, pharmacokinetics, pharmacodynamics, etc.).

The reverse may also occur where a scientist selects a scaffold molecule, identifies target parameters, and uses an AI-algorithm to screen substituted variants to that molecule to identify potential lead compounds.

It is highly unlikely, however, that an AI-algorithm could identify an optimized target or molecule without some human intervention because the algorithm would, currently, be limited by the amount of clean training data and human expertise to develop the algorithm.

Despite advances in the industry, the above scenarios are still iterative and require substantial human intuition. Under either scenario, it is the scientist that identifies and first appreciates the problem and possible solutions to that problem developed by AI. And it is the scientist who would develop, test, and ultimately appreciate the utility of the result.

Accordingly, even though AI-assistance can greatly streamline the biopharma discovery and development processes, those efforts should not lead to an inventive contribution. The fact that a scientist utilizes AI assistance should not undermine the inventive nature of the work. Indeed, as recognized under 35 U.S.C. § 103, “[p]atentability shall not be negative by the manner in which the invention was made.” The same should hold true for inventorship.

Author Information

Frank A. DeCosta III, Ph.D, is a partner in Finnegan’s Washington, D.C., office. He currently serves as leader of the firm’s litigation section and has significant experience in patent litigation, client counseling, and providing opinions related to electrical and computer technology, specifically artificial intelligence and machine learning, consumer electronics, software, medical devices, and information systems.

Sanya Sukduang is a partner in Finnegan’s Washington, D.C., office. Sukduang has a dynamic practice litigating matters concerning Abbreviated New Drug Applications (ANDA) challenges for brand drug manufacturers, diagnostic methods, biological products and medical devices.

Jorge F. Gonzalez is an associate in Finnegan’s Washington, D.C., office. He practices patent litigation with a focus on chemical and pharmaceutical technologies, including Hatch-Waxman litigation related to Abbreviated New Drug Applications (ANDA).