AI use in the patent application process

I came across this interesting blog article relating to the US Patent and Trademark Office's (USPTO) use of AI in the patent application process.

In most countries, when a patent application is filed it is "examined" by the patent office to check that it meets the relevant requirements for grant. One of these requirements is that the invention claimed in the patent application is new. In order to determine if an invention is new, a patent examiner will conduct a search of documents that were publicly available at the time the patent application was filed in order to ascertain if any documents disclose the invention. If the examiner cannot find any so called "prior art" documents, the invention is considered to be new.

Of course, the examiner cannot possibly review every single document ever published. Therefore, even if no relevant prior art is found by a given examiner it does not mean that the invention is new, but merely means that the examiner could not find any relevant prior art. Should relevant prior art come to light in the future, say when the patent application has been granted, this newly discovered prior art can be used to invalidate the patent. Such newly discovered prior art may well come to light during litigation and so can be highly problematic for patent owners. It is therefore important for those that use the patent system that patent offices conduct thorough searches for relevant prior art.

I was encouraged then to see that the USPTO are expanding their use of AI when searching for prior art documents during the patent application process. In particular, the USPTO are using a tool referred to in the article as "Similarity Search". The article notes that "a patent examiner provides input, including a patent specification, to the “Similarity Search” feature" and "the feature then uses AI models to identify and, within seconds, output U.S. and foreign patent references similar to the patent application being examined". The patent examiner can therefore combine their own search with that of Similarity Search, increasing the robustness of the search process and reducing the risk that relevant prior art is not identified prior to grant of the patent.

The addition of Similarity Search should help improve confidence in the USPTO’s work, and provide further legal certainty to applicants. It will be good to see other patent offices around the world moving towards incorporating AI in their examination process too. 

To learn more about Marks & Clerk's AI expertise, please read our latest AI Report here.