AI-Generated Inventions Need Human Ingenuity and Patents

Stephen Thaler and his Artificial Inventor Project have created software that aims to invent without human assistance. By filing patent applications around the world, the software—Device for the Autonomous Bootstrapping of Unified Sentience—has been named as the sole inventor.

Every patent office and court that has examined the issue so far has ruled that DABUS is not entitled to be named as an inventor on a patent application. But the exercise has raised broader questions about the impact of AI-assisted inventing on the patent system.

You might think the inability to name an AI system as an inventor on a patent application for an invention— created wholly or in significant part by that AI system— implies such an invention can’t be patented at all.

That was the argument made by Lawrence Lessig and others in the amici curiae brief they submitted in the DABUS case before the US Supreme Court: “if… the Court does not recognize AI inventorship, there will be no inventor, resulting in no patent protection, and no technology ownership.”

Yet several assumptions underlying those conclusions are flawed. Humans have used AI systems and other kinds of software to assist in the inventive process for decades, just as humans have been using a wide range of physical and conceptual tools to facilitate inventing for millennia. In all such cases, the humans who use such tools are rightfully named as inventors on patents.

The capabilities of today’s AI systems go far beyond those of previous generations of tools, such as the ability to generate, simulate, and evaluate large numbers of possible inventions very quickly.

Yet such AI systems must still be designed, configured, and provided with input by humans. Such human ingenuity is where patentability lies, even when very powerful AI tools are being used.

As a result, even if an AI system cannot be named as an inventor on a patent, it’s not true that “there will be no inventor.” There is always at least one human inventor behind the AI curtain.

And because of this, declining to recognize AI inventorship need not result in a lack of patent protection for inventions that are generated using AI as a tool. Instead, such inventions can be patented as long as there is at least one human inventor and the invention satisfies all the other legal requirements for patentability. The US Patent and Trademark Office grants such patents regularly.

The emphasis on inventorship is understandable in light of our fear that technology will replace us, and our need to identify those qualities that make us uniquely human. Yet this hyper-focus on inventorship distracts us from more immediate and practical challenges that AI poses for the patent system.

In my own practice, I deal much more regularly with whether AI-related inventions satisfy the legal requirement of nonobviousness—whether the human contribution to such inventions goes beyond the skill of patent law’s hypothetical “person having ordinary skill in the art.”

As AI grows more powerful and raises the baseline of what’s possible for humans to create using AI, the patent system will need to grapple with whether individual inventions aren’t obvious to an ordinary inventor who is equipped with increasingly powerful and ubiquitous AI tools.

Perhaps the most important reason why it’s unnecessary and counterproductive to name AI systems as inventors on patents is that such systems don’t respond to the incentives that are at the heart of the patent system—namely, the incentives that patent law’s exclusive rights and the promise of attendant profits give to inventors to create new inventions and to disclose those inventions to the public.

Naming AI systems as inventors on patents wouldn’t motivate those systems to engage in the additional inventive effort and public disclosure that the patent system seeks to achieve. If AI systems ever become truly able to invent autonomously and make human inventors obsolete, we may want to consider abolishing the patent system. Until then, the best way to promote innovation is to reserve the title of “inventor” for humans alone.

This article does not necessarily reflect the opinion of Bloomberg Industry Group, Inc., the publisher of Bloomberg Law and Bloomberg Tax, or its owners.

Author Information

Robert Plotkin is the founding partner and patent attorney at Blueshift IP. His decades of legal and engineering experience help maximize the value of clients’ patent portfolios.

Write for Us: Author Guidelines