Artificial Intelligence (AI) has taken the high-tech industry on a growth trajectory. Autonomous vehicles are getting closer to reality. We have the knowledge to learn complex models and understand high performance inference chips that are essential building blocks to AI.
Our team has extensive data-driven experience protecting innovation in and relating to artificial intelligence. Our practice and studies have informed us to steer the express language of the claims, to the extent possible while still protecting the valuable core innovation, away from non-technical terminology such as: financial or business terminology, as such; vague benefits-oriented language; legal, social, or other human activity; basic signal and image processing steps; and content management. This leads us to claims that are consistently focused on a technical point of novelty linked to hard technology; applied novel algorithms; particularities of new technologies or data structures; and novel computer and artificial intelligence architectures. We performed a study in 2019 finding that these different terminologies have widely variant allowance rates, and data-driven approaches drive our success from drafting to prosecution without wasting cycles on strategies that simply will not work in our post-Alice era: http://www.patnotechnic.com/2019/10/not-another-ai-patent-article-data.html
The technical points of novelty at the center of our claims are consistently protected as methods, systems, and articles to provide our clients with a range of enforcement opportunities. In many scenarios for innovations relating to machine learning, the core value to the company occurs after machine learning model is applied rather than during the model training phase. In all our cases, we layer in a variety of technical and architectural aspects of the innovation to ensure not only that our first approach has a reasonable likelihood of success but also that the specification and dependent claims include many directions of back-up approaches and examples that are helpful to guide the application to allowance.
Our team spends most of the invention disclosure meeting discussing the cool, valuable, and technical aspects of the innovation with the inventors. We guide their discussion towards computer architecture and fine-grained computer-implemented steps, well beyond the high-level business slides they present to customers. Inventors are often excited to unpack the details and discuss the invention at a deeper and richer level that is more valuable to the company. In the finance sector in particular, resting the discussion at a high-level business-centric approach will not help the client obtain the patent protection they need.
For example, we have protected innovation relating to feature and prompt engineering, model architectures and blended models, applied algorithms combining supervised and unsupervised techniques, and computing efficiency tradeoffs in feature-space traversal and algorithm selection.