Crediting Artificial Intelligence
By Terry Tush
Mike Schuster envisions a possible future where the U.S. Patent Office receives hundreds if not thousands of patent applications for inventions created by artificial intelligence (AI). He believes that day may be closer than many believe.
Computer-generated technologies have already been patented, and more applications are on the way. Up to now, however, applicants have hidden AI’s role in inventions because of uncertainty regarding how the Patent Office will treat computer inventors. That’s why it’s necessary to set standards regarding AI and patent law, says the Oklahoma State University professor of legal studies, and his recent research could assist in making more informed decisions.
Examples of AI-created inventions include bullet-train technology and communications systems. A software package called the Invention Machine created a system controller, and the individuals who uploaded background information into the AI received the patent. But Schuster and many others are asking: Who should have received the patent? Does the patent belong to the AI itself? Or to the creator of the AI?
“That raises a host of interesting legal questions,” said Schuster, who conducts research in intellectual property law. “You have to ask, are we going to grant patents for a non-human created invention? If so, who owns them?”
Those are some of the questions he addresses regarding AI and patents in an article forthcoming in the Washington and Lee Law Review.
Schuster began his research by filing a Freedom of Information Act request with the U.S. Patent Office seeking information on guidelines regarding how to treat patent applications claiming a non-human inventor. The office responded that it had no such policies.
With no guidance and finding little literature on the subject, he thought about the following questions: “If we are going to grant these patents, who should own them? Should it be the original coder of the software or the software company that employed the coder? What about the party in industry who is actually paying to use the AI to create new inventions that it’ll employ in commerce? Or should it be the engineer who looks at the AI’s output and recognizes a great idea? And what about the attorney who recognizes a patentable invention? These are all viable candidates for ownership.”
“I assumed the Patent Office was going to grant these patents and wondered to whom we should allocate these rights to maximize economic efficiency,” he said.
Why is Schuster’s research important to the average family in the U.S.? He identifies one type of AI (out of many) that can be used to create new products called genetic algorithms, which mimics evolution for new and improved technologies. Genetic algorithms are able to evaluate many sets of invention characteristics and discard poor performing sets as it goes.
“The process continues until it either identifies a system that satisfies a particular goal, say a microwave oven that will cook a chicken in two minutes, or after X many iterations of the genetic algorithm process,” Schuster said.
Some decisions on AI inventions and patent law need to be decided before the process moves along much further, he said.
“I think the time is near to address whether we’re going to grant patents for computer-created inventions and who might own these patents,” he said. “It is certainly an issue in business that is coming to the fore as we speak. That being said, it’s better to make an informed decision that we can live with for the next several generations, as opposed to making a bad decision hastily.”
Read It
Schuster’s research was selected as an “Editor’s Pick” in the April issue of IPPro Patents. To read it, visit okla.st/AIresearch.