|

During "The AI Patent Boom and How it's Impacting the Scope of Eligibility" panel at Legalweek 2020 in New York, lawyers discussed the technology and case law impacting patent applications concerning artificial intelligence.

To be sure, while questions concerning nonhuman inventors on patent applications was the main focus, it isn't the only pressing issue in AI patent law.

Patent offices are also determining "whether patents that are so abstract in nature and are almost human activities processed by a smart computer should be patentable" and what specific aspects of machine learning should be patentable, said Square intellectual property and legal operations director Kirupa Pushparaj.

However, Pushparaj noted Europe has recently taken a rigid view that a computer can't be listed as an inventor on a patent application, despite ongoing advancements in machine learning.

Specifically, last summer AI-powered computer DABUS grabbed the IP and tech world's attention when it was listed as the inventor in patent applications for a food container and signaling device. The patent filings were notable because in most patent offices across the globe, only humans can be inventors listed on a patent application.

"DABUS was a test case, the first we know of where artificial intelligence with no human being was listed as the inventor on a patent application," said Covington & Burling of counsel Gregory Discher. "The bottom line is that the European Patent Office refused the application. It essentially boiled down to a requirement that requires the naming of a human inventor."

He explained the European Patent Office requires an inventor to have a family name and address, which DABUS didn't have. Despite DABUS's patents being rejected, the panelists said advancements in deep learning makes the prospect of more computer-generated inventions likely.

"One paradigm shift of machine learning versus AI at large is I think machine learning is a paradigm shift in tech in that humans are no longer programming this," Discher explained.

"It's a fundamental shift and approach to software at large that people are not programming this thing, the software effectively programs itself outside of human intervention. That segues into what did the human invent? What did the human do," he added.

Discher said AI will need a legal standing similar to corporations to perhaps avoid a fate similar to the DABUS patent filing.

"We have not created an AI legal construct, we've created legal constructs for corporations," he noted.