The Landscape and Impact of AI as an Inventor: Is AI the next Edison?

Aaron Ostler – On August 5, 2022, the Federal Circuit Court of Appeals released Thaler v. Vidal, an important decision concerning patent law with implications for future research and development (R&D) practices. Stephen Thaler develops and runs artificial intelligence (AI) systems that generate patentable inventions. Thaler submitted a patent application to The United States Patent and Trademark Office (PTO) for a system called the Device for the Autonomous Bootstrapping of Unified Science (DABUS). However, the PTO rejected the patent because Thaler listed the AI as the inventor, prompting this litigation. The court’s holding relied on the Patent Act’s definition of an inventor as an “individual,” coupled with precedent finding an individual to typically be a natural person. Thaler plans to file a writ of certiorari.

This decision matters because of the impact on research and technology in AI’s evolving landscape. Under patent law, the inventor must have contributed to the conception of the invention, which requires having a definite and permanent idea of the complete and operative invention. Without the ability to register the patent to the AI, the two options for patenting an invention could be to credit the invention to whoever contributed the most or to the software developer. If the current law requires a definite idea of the complete invention, then it is unlikely that many inventions would allow for the greatest contributor or the software developer to qualify as the inventor, as AI progresses to the extent it was used in Thaler. To address this, either the law will need to adapt to allow the greatest contributor or software developer to be listed as the inventor, or it would need to expand the definition of an individual to include artificial intelligence systems. Continued AI development may be stunted by the Patent Act’s limitations on who qualifies as an inventor. The inability to register patents for AI-invented systems will likely dissuade companies from using AI as a principal mechanism for invention because of an inability to exclude others from using the final product. As a result, the opportunity for efficient technological development by AI inventorship would diminish in favor of human inventorship that does not implicate patent concerns.

While AI has been an integral aspect of inventorship for years, the technology landscape is growing increasingly saturated with AI-generated inventions that many believe will eventually become pervasive throughout a range of industries. As the use of AI systems in inventorship increases, there is a developing need to protect AI work under patent law. The pharmaceutical industry is highly incentivized by the ability to exclude others from the R&D yields. The inability to patent pharmaceuticals principally from AI systems may result in companies forgoing the AI systems and opting for human inventorship to maintain the ability to patent the drugs produced. Pharmaceutical development has long utilized AI alongside human ingenuity, however, AI may eventually play such a significant role that there will not be a “person” that can be classified as the inventor. The AI system does not need to be the inventor by default; the person with the largest contributor to the final product of the software developer behind the AI system could be credited with the patent if the law changes. However, each of these introduces its own issues, such as deciding who gets the patent when there are multiple software developers working on different aspects of the algorithm. In response to an evolving landscape of technological invention, companies should begin considering how their approaches to development and patent creation will mesh with AI. 

Whether the court will recognize AI as an individual for inventorship purposes remains to be seen, but English common law may provide support for Thaler’s argument: early treatment of corporations as persons in English common law was favored by Blackstone for its “advantage [to] the public.” Treatment of corporations as persons, or categorizing corporations as legal persons, has longstanding roots in United States legal history, with many corporations having been afforded constitutional protections like freedom of speech and religion that are also afforded to individual citizens. However, while corporations are treated as persons in many aspects, they cannot be the inventor on a patent because a corporation does not contribute to the invention’s subject matter or have ingenuity. Rather, the employee, or inventor, does. AI may not fail in this respect if the AI system is capable of being the principal contributor to the subject matter of the final product and the contribution qualifies as ingenuity. Patents are granted to reward and incentivize the allocation of time and resources to technological advancements. Further, there are early signs that AI can have ingenuity. For example, DALL-E 2, an AI text-to-image technology, can generate images and artwork from the input of text.

Thaler has applied DABUS for patents in numerous other countries similarly listing DABUS as the inventor. The UK denied DABUS a patent on technical groups because the registered inventor was not a natural person but found no substantive problem. The patent would have been granted if a natural person were the inventor. The European Patent Office (EPO), however, noted that there is no previous case law preventing AI from being an inventor, which establishes the possibility that an AI inventor could receive a European patent if AI was considered a person. One IP attorney, Peter Finnie, has advocated for a carved-out classification under the law for electronic persons that would allow AI to be considered an inventor—similar to the legal carveout for corporations as legal persons. Conversely, South Africa granted DABUS its first patent, where the law does not define what constitutes an inventor. A legally certain answer to the question of whether AI is an inventor in the United States may rest upon Thaler’s writ of certiorari being granted. If not, the question may need to be placed on hold until Congress can address it through legislation. Either way, as AI continues to be further integrated into R&D, there will likely be no shortage of novel issues emerging in the technology landscape.

Leave a Reply

Your email address will not be published. Required fields are marked *