June 11, 2021
written by: Kasumi Kanetaka
A question that is often asked is why there are so many lawyers? The question is often paired with questions about why lawyers spend so much time on seemingly insignificant tasks and why they are always busy and have a poor work-life balance.
Although this problem is not unique to the legal industry, it tends to be especially prominent for lawyers because of the level of detail required by the job. Most tasks require a plurality of checks to ensure there are no errors. For example, in patent prosecution, filing responses to Office Actions issued by the U.S. Patent and Trademark Office and preparing preliminary amendments for new patent applications often require many detailed steps.
In preparing responses to each Office Action, the focus is mainly on amending the claims and/or arguing against the Office Action to overcome objections and rejections. When amending claims, word choice and phraseology are substantially important, and many checks are performed to ensure USPTO formality requirements are met. For example, it is important to check claim identifiers, antecedent basis, and grammar to ensure that the claims conform to drafting arguments. In the arguments, each sentence and paragraph will be reviewed to ensure that the discussion and reasons presented are concise, yet persuasive thoughts fully addressing the points raised in the Office Action.
Even so, each case is unique with different steps required in the drafting of the documents. Since each case may require different approaches, prioritization of steps may vary. To effectively prioritize these steps and work efficiently, practitioners are increasing looking to Artificial Intelligence to handle some of the steps for patent prosecution.
For instance, there is a service that provides AI-based patent drafting software which automates and improves patent drafting for practitioners. There is also software that can recognize “terms” used in the specification, claims, and drawings, throughout an application so that the same terms will be autosuggested and updated consistently throughout the application. In addition, there is software that can perform automated renumbering of claims when constructing a claim set.
Law firms also developing AI-based drafting software in-house. Examples include AI patent drafting tools, which use machine learning and artificial intelligence to consistently draft high-quality patent applications. At Oblon, we have created, in-house, an AI model using Python to predict § 101 rejections by merging two USPTO data sets: USPTO PatentView and the USPTO Office Action Research Dataset. Oblon’s AI tool predicts § 101 rejections with a 90% confidence level.
In addition to supporting drafting applications, artificial intelligence can assist in drafting office action responses and preliminary amendments. Checking claim identifiers and addressing antecedent basis issues are perfect for automation because these tasks are repetitive and require consistency. However, artificial intelligence is capable of so much more than simple automated tasks. For instance, it may be possible to use machine learning to learn a writer’s habits of writing and to auto-suggest words (verb tense, punctuations, etc.) to the writer while arguments are being drafted.
However, many substantive matters, such as word choice in claim amendments and development and flow of arguments, will still require decisions of human beings for the time being. Other tasks also still require a human. For example, to increase chances of allowance, practitioners often conduct interviews with Examiners to understand how the Examiners read and interpret claim language. Because each Examiner has unique personalities and preferences, it is important and advantageous to conduct Examiner interviews to develop arguments in Office Action responses. Although artificial intelligence may someday be able to interface with Examiners, for the moment human interactions and communication are necessary.
For the foreseeable future, it seems that the use of artificial intelligence in the legal field will remain as an assistive tool helping practitioners perform tasks such as patent drafting while leaving certain more abstract tasks to be performed by the practitioner. This application of artificial intelligence to the legal field will help lawyers handle transactional work such as patent prosecution in a much more efficient and effective manner.