Applying Artificial Intelligence to the Patent System
Some straightforward uses of AI would be to assist the patent office in determining whether applications for patents should be granted or denied. Specially designed AI software could assist patent office examiners by finding relevant prior art, summarizing patent disclosures for efficient review, and helping to interpret difficult patent claim language. The increase in examination productivity could benefit users of the patent office by reducing fees, processing time, and error rates. Some less apparent possibilities for AI could provide new services to society. For example, AI could serve the public by summarizing the teachings of patent documents and automatically answering technical questions using information contained in millions of patent documents.
While general-purpose AI techniques have many productive uses, AI techniques tailored to the problem domain have greater promise. Developing AI capabilities specifically optimized for the patent system will require significant research and experimentation. Carnegie Mellon University has launched a new center to help develop this nascent field. The Center for AI and Patent Analysis is an interdisciplinary initiative with a twofold mission. It conducts original research on AI tools and techniques to serve various users of the patent system. It also seeks to define a long-term research agenda to focus the AI research community on collaboratively and incrementally solving the greatest challenges for patent analysis.
Document Type: Research Article
Publication date: August 1, 2019
Technology and Innovation, edited and published by the National Academy of Inventors, is a forum for presenting information encompassing the entire field of applied sciences, with a focus on transformative technology and academic innovation. Regular features of T&I include commentaries contributed by the United States Patent and Trademark Office (USPTO) and in-depth profiles of Fellows of the National Academy of Inventors in every issue.