Experts warn AI inventorship debate will only heat up

Panellists from an AIPLA webinar on emerging technologies said that refusing to grant a patent to an AI inventor has the potential to devalue inventorship and could have serious implications for companies that invest heavily in new technologies.

The discussion, on May 13, came on the heels of this year’s decision by the EPO to deny patent applications to an AI system named DABUS (Device for the Autonomous Bootstrapping of Unified Sentience) because a machine is not a human inventor. The EPO also argued that a machine cannot receive a patent because it has no “legal personality” and therefore cannot own a patent right.

Gareth Jones, vice president of intellectual property at Benevolent AI, said that patents are an important part of his company’s business model, which relies on licensing to major pharmaceutical companies so that they can do late-stage research.

His company uses artificial intelligence (AI) to sift through data on the human body, and then uses that data to find new ways to treat diseases. He said there could be serious ramifications if companies like his are not allowed to patent AI creations.

“The fundamental value of our model is the exclusivity the patent provides. It would have an impact on our revenue if we can’t protect the R&D generated by AI,” he said.

Richard Kennedy, partner at Venner Shipley in the UK, added that many companies are currently using AI as part of the creative process. He said that in most cases AI is just a tool of invention rather than an inventor by itself. The DABUS patents are an extreme example of inventions created without human interference, he added.

 “I should mention there is debate whether AI inventorship is happening at all. The EPO said that for the foreseeable future, AI is just a tool, and others are saying AI will soon replace humans. We want to fix any gaps in the law about this, so it’s worth asking if humans and AI need the same protection,” he said.

Ryan Abbott, professor of law and health sciences at the University of Surrey in the UK, said that while some technologies are tools in the creative process, some emerging technologies are now replacing human beings.

“However ambiguously we define the creative process, there is an act of invention. The way we will invent in the future will be by using inventive AI, and the way to incentivise that is to have patent protection,” he said.

“The philosophical issue here is about conception of innovation, and if we have patent law to encourage innovation it is important we recognise the contribution to the inventive output. Listing a human inventor instead of the AI devalues inventorship.”

Abbott added that in its justification of its ruling, the EPO did not dispute that DABUS wasn’t inventive. This creates legal problems for companies and research institutions that invent with particularly creative AI because they would not have a human being who could claim inventorship.

“It may not be a problem to lie and say a person invented something if there was only one person involved, but it’s increasingly going to be more difficult to find somebody’s name to put down as the inventor and not have disputes about that. In the US it’s a criminal offence to put down an incorrect name,” he said.

Granting patent rights to a machine raises concerns for obviousness. A “skilled person” in the art could never have the equivalent amount of information stored in their brain as an AI with an almost unlimited capacity for data. Abbott argued that AI will naturally raise the bar for obviousness, and in 20 years’ time the test for obviousness will change.

“At a certain point we will have AI general intelligence, or very advanced intelligence, that will be widely used in R&D, and we want to encourage that sort of thing in areas like drug research.

“At that point the ordinary skilled person in the art will also be an AI, so what is non-obvious to the new AI-skilled person will also have to be non-obvious to the AI inventor,” he said.

Jones at Benevolent AI agreed and added: “The level of the skilled person will change once AI becomes more of a commodity and you are no longer need to patent things everyone has access to.”

Kennedy said there is a risk that as companies begin replacing humans with machines, they might feel compelled to keep their discoveries as trade secrets if they feel they cannot get a patent for an AI-generated invention.

“To counterbalance that, if you can list an AI as an inventor and patents are available at the press of button, is there a danger of getting swamped with patents for AI inventions? It is a difficult balance to strike,” he said.  

Striking the right balance will be difficult for legislators and policy makers in the future. However they decide to move forward will have lasting implications not only for IP, but for the way humans and robots invent.