USPTO: Using AI To Bolster IP and Save Money
One of the most daunting tasks in managing intellectual property is the ability to quickly provide the most precise answers to the most general questions. We have turned to Artificial Intelligence and Machine Learning (AI/ML) algorithms to augment our examiners toolset and provide precisely that! Our team used natural language processing (NLP) to deconstruct patent application responses (called Office Actions) and create enriched citations to make future research easier and faster for stakeholders. This approach used design thinking from the user perspective to understand the needs of various stakeholders and the myriad of data variables required in order to deliver exactly what was asked for, e.g. user-centric answers. The NLP model proved to be quicker and more accurate than the prior work of dozens of experts. NLP saved millions of dollars in enriched citation implementation.
We also piloted AI/ML in our classification efforts. Our current, manual classification service is comparatively slow (days) and costly. Our new AI/ML algorithms have been “trained” to classify patent and non-patent documents with CPC symbols and C* symbols in hours at a tenth of the cost. This service contains a feed-back, human-in-the-loop to verify and validate accuracy of results. “AutoClass” provides seamless integration into routing and search functions with significant cost savings once fully adopted. This new, smarter routing system can save time (pendency) and millions of dollars for the agency and its fee payers.
Finally, our pilot AI/ML search function is becoming a valuable component for performing prior art searches. New AI/ML capabilities embedded in the Patents End to End (PE2E) Search product can improve the process of finding prior art, and thus, foster innovation for entrepreneurs and innovators. While searching through vast amounts of data and documents, examiners are challenged to uncover all of the relevant prior art. To address this challenge, we undertook a comprehensive strategy that included extensive market research and rigorous testing of a wide range of concepts with hundreds of examiners. Powerful examiner feedback and ideas helped drive the development and testing processes. User centered design (UX) was a key component that fueled the effort. Now, easy-to-use AI/ML capabilities can be part of PE2E Search in the browser, so examiners will be able to find similar art for the concepts created. There are also new ways to access documents during search, and new methods to review documents. Clearly, such new capabilities can help empower examiners and add value to their vital work. The machine learning algorithms capture feedback and “learn” as you use it, so it gets MORE useful and more powerful as you go!
The lessons learned through our AI/ML efforts include:
- Start with a use case that ties hard numbers to a material impact – act now & be bold!
- Ensure the results you are providing are useful and constructive (to users – NOT you)
- Remember AI/ML supplements human intelligence – it does NOT replace it!
- Confirm with experts that results are reasonable before using a feedback loop
If you have any comments or questions, don’t hesitate to contact me at: