Context: The impact of Artifical Intelligence on the patent system that has made more headline news than any other is the question of whether AI systems can be inventors, with conflicting decisions in different jurisdictions. But that is not the only important question in that regard.
What’s new: Yesterday (April 30, 2024), the United States Patent & Trademark Office (USPTO) launched a consultation (regulations.gov) seeking comments over the next 90 days on “Considerations for the Impact of AI on Prior Art” and “Considerations for the Impact of AI on the Knowledge of a PHOSITA” (Person Having Ordinary Skill In The Art).
Direct impact: The USPTO will evaluate the responses with a view to its own guidelines as well as potential recommendations for legislative change, a possibility that is explicitly raised. The questions at issue are not only relevant to the patent prosecution process, but also to infringement litigation, such as when the availability of AI influences the knowledge that a person of skill in the art is assumed to have and, therefore, claim construction.
Wider ramifications: AI-“generated” policy implications pose a challenge to the convergence of the world’s major patent systems that has been a trend in recent decades. The USPTO is trying to position itself as a thought leader among the world’s leading patent offices, but also hopes to potentially learn from other jurisdictions (the 14th question relates to “laws or practices in other countries” that may address any of the questions at hand). The European Patent Office (EPO) should probably do more in this field, as its focus still appears to be largely on inventorship and patent-eligibility questions (EPO AI webpage, last updated two years ago). The European Commission’s DG GROW and patent policy experts in the EU member states should also consider prioritizing the impact of AI on patent law over the (not really useful) proposed EU regulation on standard-essential patents.
First part of the consultation: impact of AI on prior art
Obviously, patent law was not originally created with the possibility of AI-generated prior art in mind, so there is no explicit requirement that prior art be made by humans. But the USPTO is mulling whether AI-generated prior art should be ineligible or treated differently, and whether someone who submits AI-generated prior art should make a disclosure.
Subquestions 4b and 4c basically relate to areas in which AI-generated prior art tends to have shortcomings (at least based on current technologies):
b. How should the fact that an AI-generated disclosure could include incorrect information ( e.g., hallucinations) affect its consideration as a prior art disclosure?
c. How does the fact that a disclosure is AI-generated impact other prior art considerations, such as operability, enablement, and public accessibility?
Hallucinations are also at issue in various AI lawsuits (typically in parallel to copyright infringement claims), the most recent one having been filed yesterday by eight newspapers belonging to a hedge fund (April 30, 2024 ai fray article).
Operability, enablement and public accessibility can be adversely affected by AI generating vast quantities of material, some of which may simply not make technical sense.
The most fundamental concern that the USPTO raises in the first part is that the sheer quantity of AI-generated prior art could make it impossible to obtain patents or, which is another extreme, all of that abundance (which includes a lot of material that doesn’t even make technical sense, though it is grammatically correct) may make it impossible to find relevant prior art:
5. At what point, if ever, could the volume of AI-generated prior art be sufficient to create an undue barrier to the patentability of inventions? At what point, if ever, could the volume of AI-generated prior art be sufficient to detract from the public accessibility of prior art ( i.e., if a PHOSITA exercising reasonable diligence may not be able to locate relevant disclosures)?
Second part of consultation: impact on PHOSITA (i.e., non-obviousness considerations and claim construction)
In that part, the USPTO raises interesting questions about whether the ability of AI to identify patterns in text material across different fields of technology could mean that basically every field of art is “analogous” for the purposes of potentially combining references. But the first question is whether the P in PHOSITA means that the focus should be on what a person (a human being) does. And should there be the assumption that such a person makes use of AI tools, the USPTO would like to know how it can determine what tools the PHOSITA should be assumed to utilize.
In the end, even claim construction in infringement proceedings (beyond just an invalidity defense) can be impacted by AI now.
These are all important and complicated quesitons. The purpose of this article was not to discuss each of them in detail, but to provide a bird’s-eye view. Again, here’s the USPTO’s request for comments.
Responses are accepted until July 29, 2024, and ip fray will report on interesting submissions.