A Look from Silicon Valley at How Patent Practitioners View Thaler
Editor’s note: This article is the third in a series of four posts about AI and inventorship, developed as part of Professor Colleen V. Chien’s AI and the Law class at Santa Clara University School of Law. The first blog post describes the Thaler v. Hirshfeld appeal pending before the Federal Circuit and related appeals. The second blog describes the technology behind DABUS. This third blog presents the results of a survey that was sent to patent practitioners across the Bay Area about their perceptions of the importance of AI inventorship.
Carson Swope is a rising third-year student at Santa Clara Law School currently working at Cruise as an Intellectual Property Intern, and Cosette Cornwell is a Graduate Fellow of Santa Clara Law School.
On June 6th, the U.S. Court of Appeal for the Federal Circuit will hear oral arguments in Thaler v. Vidal. The Court will decide whether an AI system comprised solely of source code can qualify as an “inventor” under the Patent Act. If so, then an AI system would be entitled to a patent for any useful, non-obvious, and novel idea that it generates. It may seem counterintuitive to treat a piece of code, existing within a matrix of wires and chips, as an “inventor” — just like a human being, but without the beating heart and brain tissue — but that is exactly what Dr. Stephen Thaler is arguing before the court.
To gauge sentiment in Silicon Valley about the importance of this case, we circulated a questionnaire answered by over 200 intellectual property practitioners (ongoing), 65 of whom indicated that they were patent practitioners. The survey yielded three primary insights into how practitioners who work in the computer, mechanical, and material sciences in Silicon Valley view the issue of AI inventorship. It was conducted pursuant to an institutional review board (IRB) review and its informed consent requirements.
First, a policy change to AI inventorship does not seem to be of major concern to practitioners. Only 18% of the 65 respondents reported that they felt “very concerned” about the issue and only one respondent reported feeling “extremely concerned.” In contrast, more than half of respondents answered that they were “very,” if not “extremely” concerned about the impact of changes to subject matter eligibility on innovation at their companies. Still, a majority of respondents registered at least some amount of concern over the impacts of AI inventorship on innovation at their companies, with 18% reporting they felt “very concerned” about the issue, 25% “somewhat” concerned, 28% “a little concerned,” and 28% reporting no concern at all.
Second, the survey revealed that AI inventorship is yet to be an active issue in the day-to-day work of patent prosecutors. When asked if they ever had to consider whether an AI should be listed as an inventor, only 4 out of 65 respondents replied in the affirmative. Of those 4, one of them clarified that the issue was “more theoretical than actual” and another explained that AI inventorship “was discussed as [a] possible topic for future consideration,” but not “immediately relevant to any current applications.” Thus, it appears that virtually none of the respondents have actually decided whether to name an AI as an inventor on a specific patent, though at least a few have considered it prospectively as something they may do in the future.
Third, a substantial share of respondents are following Thaler’s case with a moderate to high degree of interest, with 31% agreeing “somewhat” and 9% agreeing “strongly” that the issue of AI inventorship would impact innovation in their industry. Electronics, IT, mechanical practitioners, and “other” practitioners (N= 52) were much more likely than biopharma practitioners (N= 13) to indicate agree, strongly or somewhat, that AI inventorship would impact innovation.
Degree of agreement among survey respondents to the question: How the issue of AI inventorship is decided will impact innovation in my industry (by focal technology) (N= 65).
These survey responses seem to reflect a consensus view among patent attorneys that AI inventorship is an interesting and important topic but too speculative at this point to require an overhaul of long-standing patent law. In a comment submitted to the USPTO in August 2019, the American Intellectual Property Law Association (AIPLA), a national bar association of approximately 12,000 members, made that case.
The association argued firmly that current U.S. law should not change to include AI systems within the definition of “inventor.” First, AIPLA distinguished between what it calls “inventive AI,” where AI makes an inventive contribution to the claimed patent as an inventor, which is what Dr. Thaler claims DABUS has done, and “AI inventions,” which refers to all innovations that incorporate AI, and represents a much broader and better understood use of AI. Second, AIPLA argued that while a few test cases such as DABUS exist now, “there is not yet enough information available to know whether the claimed subject-matter is truly ‘inventive AI.’” “Thus, until there are more examples of and more understanding around ‘inventive AI,’ it is premature to consider a change to the existing law.”
Regardless, as AI continues to make breakthroughs — from DeepMind’s AlphaFold2, which is poised to predict the shape of nearly every protein known to science, to OpenAI’s DALL-E 2’s image generator — it may be sooner than either the AIPLA or the Court of Appeal thinks that truly “inventive AI” will arrive, if it hasn’t already.