News
Practice Areas
Technologies
- Artificial Intelligence
- Computer Science & Software
Are AI systems patent-eligible?
A ruling from the U.S. Court of Appeals for the Federal Circuit, which hears all patent-related appeals, raises questions about the eligibility of artificial intelligence (AI) systems for patent protection. The system in the case ran into trouble with the so-called Alice test (named for the case where it originated) for patent eligibility.
Invalidating visualization patent
The case involves four patents held by AI Visualize that generally related to the visualization of medical scans using a dedicated, low-bandwidth web portal. The patents were part of the same patent family. They generally covered systems and methods for users to review three-dimensional virtual views of a volume visualization dataset (VVD) on an internet-connected computer without needing to transmit or locally store the entire VVD.
The patentee sued Nuance Communications for infringement, and Nuance moved to dismiss the case. It argued that the patents’ claims weren’t eligible for protection and therefore were invalid. After the trial court granted the motion based on Alice, AI Visualize appealed.
Transforming required
Alice is a two-part test. In step one, the court determines whether a claim is directed to a patent-ineligible concept, such as an abstract idea. If not, the inquiry ends. If so, the court then reviews whether the claim includes elements that transform the concept into a patent-eligible application of the concept.
The appeals court agreed with the trial court that the patents’ claims covered an abstract idea. It has previously explained that the steps of obtaining, manipulating and displaying data — particularly at a high level of generality — are abstract concepts. The claims here covered such “functionally oriented” steps as storing data, accepting user requests to view the data and checking the location of the necessary data. In other words, the court said, the claims were for converting data and using computers to collect, manipulate and display the data.
The court rejected AI Visualize’s argument that the claims weren’t abstract because they required the creation of “on the fly” virtual views at a client computer. This creation, it found, was achieved by manipulating part of the existing VVD, so the “creation” was just abstract data manipulation.
It was irrelevant that the specification contained language about how virtual views provided a technical solution to a technical problem. The claims themselves didn’t explain how to create frames or virtual views, let alone in a way that would meaningfully support a technical solution to an existing technical problem.
The appeals court also affirmed the trial court’s finding that the claims didn’t cover something “significantly more” than the abstract idea. It noted that routine, conventional or well-known claim elements don’t sufficiently transform the claim.
AI Visualize contended that the creation of virtual views adequately transformed the claims to a patent-eligible concept. The court reiterated, however, that creating a virtual view is itself an abstract idea. Moreover, technology already existed to present richer views.
The appeals court also wasn’t convinced that creation of virtual views “on demand” transformed the claims into something more. It found that the patentee didn’t make sufficient factual allegations to show that the claims involved unconventional technology or a concrete application of the abstract idea of virtual view creation. Merely reciting an abstract idea performed on a set of generic computer components, the court concluded, doesn’t produce an inventive concept.
Learning lessons
The ruling in this case demonstrates some of the challenges that AI systems may face when attempting to obtain patent protection — and the vital role of patent drafting. Patentees must include details that clarify how their invention provides a technical solution to a technical problem or involves a transformative application of what otherwise would be an abstract idea or ideas.
© 2024
