AI is coming for our jobs. AI is going to make our jobs easier. You’ve probably heard both, given that AI has dominated the technology conversation for several years now. This holds true across industries and across professions, and patent practitioners are no exception.
Since our founding in 2018, Patent Bots has been a pioneer of AI in patent proofreading and patent drafting software. In fact, AI is used across the suite of Patent Bots tools, albeit not necessarily in the ways you’d expect. These days, when someone says AI, it’s likely they are referencing generative (gen) AI. However, AI has long been used for its non-generative capabilities, with generative AI only coming onto the scene within the last decade.
So what exactly is the difference between non-generative and generative AI and what role does each play in automated patent drafting?
Non-Generative AI in Patent Drafting Tools
Non-generative AI refers to artificial intelligence systems that analyze, classify, or make predictions about existing data without creating new content. Put another way, this type of AI focuses on understanding and processing information that already exists.
As an example, Patent Bots uses non-generative AI to parse the text of patent claims. This occurs with a custom-trained machine learning model for grammatical decomposition. The drafting tools in Patent Bots also allow you to generate text, particularly for some of the more administrative sections of your patent application, such as the claim summary or abstract, using non-generative AI. In this case, the models do not generate new, original content but rather put together the appropriate language based on the content already in your application.
Of course the uses of non-generative AI extend beyond automated patent drafting. Patent Bots uses this type of AI for its analysis and other tools as well. Take the Art Unit Predictor. Classification is a common use for non-generative AI, and with this solution, Patent Bots trained machine learning models with data to classify Art Groups and highlight the ones to which the patent application is most likely to be assigned.
Generative AI in Patent Drafting Tools
Generative AI refers to artificial intelligence systems that create new, original content based on patterns learned from training data. Rather than just analyzing or classifying existing information, this type of AI produces outputs that didn't previously exist.
Patent Bots offers a generative AI option in our patent drafting software for drafting the title, background and abstract. In this case, the output generated is wholly new content. Understanding that law firms have varying degrees of comfort with gen AI, this feature can be enabled or disabled at the firm level within the Patent Bots platform.
As with non-generative AI, Patent Bots also employs generative AI for tasks outside of automated patent drafting. This includes PatentPlex, which provides easy access to patent family trees, patent history, and more. For ease of analysis, the PatentPlex tool includes a Generative AI Assistant that allows you to explore the patent and its prosecution history. PatentBots securely submits the question, the text of the patent, and the prosecution history to a generative AI agent. Patent Bots then displays the answer, leveraging generative AI to help users more quickly hone in on the information needed.
Choosing the Right Tool AI Tool For Automated Patent Drafting
With so many AI solutions, both generative and non-generative, available in the market, how do you know what type of AI tool is best for each type of patent drafting task?
Patent Bots founder and patent attorney Jeff O’Neill co-authored a piece in Landslide Magazine from the American Bar Association to dive into just this topic. His takeaways:
Different Tools for Different Tasks: Patent drafting can be supported by various text generation and AI tools, each suited to different parts of the application depending on the need for precision and complexity.
Strengths and Limitations: Generative AI tools excel at generating natural-sounding, general-purpose text, making them useful for tasks like writing abstracts or background sections. However, their tendency to produce "average" text makes them unreliable for sections requiring exact language, such as claims or detailed invention descriptions.
Rules-Based Tools Offer Precision: For highly structured and repetitive tasks like generating claim summaries, both non-generative AI and generative AI tools can be effective, if they are trained to follow strict legal formatting and ensure consistency.
Human Expertise is Irreplaceable: Drafting patent claims and detailed descriptions still requires the critical thinking, legal reasoning, and domain expertise that only human practitioners can provide.
Want to experience AI patent drafting tools - both generative and non-generative - for yourself? Start your free two-week Patent Bots trial today.