Step One in Fixing the U.S. Patent System – AI Searches
Recently, the U.S. Patent and Trademark Office (USPTO) started using AI tools to help find “prior art” to filed patent applications. Prior art refers to publications (patents, scientific articles, webpages, etc.) that pre-date the filing of a patent application. The idea is to help examiners find the best, most relevant prior art, resulting in (a) faster examination times, (b) better quality examination, and (c) reduced doubt of the validity of an issued patent (i.e., “stronger patents”).
In one proposal, AI searching tools would be applied to filed patent applications BEFORE examination begins. A patent applicant would receive the search report after the patent application is filed and be given an opportunity to revise patent claims that were submitted or even withdraw the patent application for a partial refund of patent filing fees. In such a pre-examination system, an automated patent search would certainly help speed up patent examination by giving an initial assessment of patent filings before an examiner even looks at the application. However, the USPTO is only using the AI searching tool to help examiners find the best prior art during examination.
AI patent searching can add “Contextual Understanding” to searches that typically are only “term” searches where an examiner looks for specific key terms in prior art. Generative AI models can interpret the intent behind a query, allowing for more nuanced searches. AI patent searching can also add “Concept Exploration” to searches that help examiners explore patent landscapes by generating patent summaries or identifying patents related to a given technology or field.
As a patent attorney with 25 years of experience, I know that patent quality depends heavily on the quality of the search performed by examiners. For some important patents, clients ask for patent searches to be done before the application is prepared and filed. The search results are then provided to the examiner. These clients want the examiner to have the very best prior art in front of them to help in identifying the best claims possible for the patent. An AI pre-examination search for all applications can help with three key points of criticism about the patent system – speed up a slow process, improve examination quality, and strengthen issued patents.
As explained by the USPTO, a patent examiner provides input, including a patent specification, to the “Similarity Search” feature. 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.
View referenced article
