The Implications of Artificial Intelligence on Patent Law
Today powerful computers perform computationally intensive processes similar to the cognitive functions we associate with the human mind, such as learning and problem solving. This is sometimes referred to as Artificial Intelligence (“AI”). As AI plays an ever increasing role in the development of potentially patentable inventions questions of inventorship and novelty arise.
In August of this year an AI system known as DABUS (“Device for Autonomous Bootstrapping of Unified Sentience”) was identified as the inventor on patent applications filed in the U.S. Patent and Trademark Office, the U.K. Intellectual Property Office, and the European Patent Office (EPO). The applications were for a beverage container and a light that flashes in a rhythm that is difficult for humans to ignore. Although many patent applications have been filed in the past for AI systems, these may be the first with an AI system named as the inventor.
DABUS was created by Dr. Stephen Thaler, President and CEO of Imagination Engines. It is a “creativity machine” that uses a system of neural networks to generate new ideas by altering the network interconnections without human intervention. These idea generating modules are monitored by a second system of neural networks that detects critical consequences and, if they are determined to be novel, important, useful, or valuable, reinforces them.
The America Invents Act defines an inventor as “the individual or, if a joint invention, the individuals collectively who invented or discovered the subject matter of the invention.” U.S. patent law only recognizes individuals as inventors, not businesses or machines. Moreover, inventorship is determined by “the formation in the mind of the inventor of a definite and permanent idea of the complete and operative invention.” The question therefore arises – can DABUS be the inventor, i.e., did it conceive the inventions.
An associate professor of law at the University of Missouri School of Law, Dennis Crouch, has noted that computers have assisted in the development of inventions for some time, and he suggests that the “inventor” is the person who instructed the computer to do the work and who recognized that a solution to a problem had been developed by the computer. However, today AI systems are capable of developing their own code and languages that are more efficient means of communication than human language. This enables AI, not a human programmer or logic developer, to create new programming that draws new conclusions, and that identifies and exploits an inventive apparatus or method.
Whether AI can be an inventor also raises a question important to the issues of obviousness and enablement, namely whether the programmer or the AI is the hypothetical person of ordinary skill in the pertinent art used to determination whether a purported invention would have been obvious. Because AI can sort and store vast databases and access information at speeds far beyond human capabilities, AI can recognize patterns that lead to invention more readily than humans. Eventually the hypothetical person of ordinary skill in the art may be deemed to be a person using AI.
Who is the person of ordinary skill also impacts the determination of whether a patent specification enables its claims, i.e., whether it includes sufficient detail to enable a person of ordinary skill to practice the invention without undue experimentation. If the AI is the person of ordinary skill, then significantly less information may be required to be disclosed to enable the claims.
As AI continues to advance, Congress, the PTO, and the courts will have to resolve the difficult question of whether AI can be considered an inventor, and if not, who among those responsible for the AI are the inventors.