Intellectual property professionals have been using computer tools in researching and preparing patent applications for many years. For example, lawyers use word processing tools to draft and edit almost all documents, and they routinely use search engines and Internet resources for research. Naturally, there are no rules prohibiting the use of such computer tools in the patent application process, so long as they comply with applicable statutes and regulations. Similarly, there are no rules specifically regulating the use of AI in researching or preparing patent applications and related materials. Nevertheless, professionals who choose to utilize AI must be careful to comply with existing rules that they may violate through irresponsible use.
In October 2023, President Biden issued an Executive Order on the Safe, Secure, and Trustworthy Development and Use of AI. Pursuant to that order, the United States Patent and Trademark Office (USPTO) issued guidance in April 2024 to guide the responsible use of AI tools in practice at the USPTO. In the guidance, the USPTO expressed its commitment to maximizing the benefits of AI while recognizing the need to “mitigate the risks that may arise from the use of AI in practice at the USPTO.” In this article, we outline some potential use cases for AI in the patent application process, as well as the benefits and risks associated with such use.
Using AI in document creation
Advances in AI tools designed specifically for use in the patent application process have the potential to revolutionize the work of IP practitioners. Large-scale language models (LLMs) such as ChatGPT are a special class of generative AI tools that use language processing algorithms to understand and generate language. LLMs have the potential to provide tremendous value to IP practitioners, as they can train software to understand and follow specific formatting, style, and structure requirements for specific document types. In fact, there are already more than a dozen AI-powered tools designed specifically for writing patent applications.
Recently developed LLMs such as davinci, Rowan, and others promise to reduce the amount of work required to draft patent applications and responses to patent office actions. These tools can significantly improve the efficiency of document drafting by providing actionable answers within seconds of entering prompts. The accuracy and usability of an AI system's output depends on several factors, including the quality of the system's training data, the complexity of the task, and the appropriateness of the user's input. According to a 2023 report by McKinsey, the legal field in particular has great potential to realize productivity gains through the adoption of AI. In its latest guidance on the use of AI, the USPTO recognizes that newly developed generative AI tools “have the capacity to draft technical specifications, generate responses to patent office actions, prepare and respond to abstracts, and even draft patent claims” with minimal human input. By reducing the time and amount of work required to prepare these documents, IP practitioners should be able to reduce the cost of their services and improve access to patents for inventors overall.
While the potential benefits of responsible use of AI are great, the risks associated with irresponsible use of such tools are equally great. Attorneys are under no obligation to disclose that they used AI to draft documents they submit to the USPTO. However, USPTO rules require a human to sign nearly all submissions to the agency. Under 37 CFR 11.18(b)(1), the signature certifies that all statements are true to the best of a party’s own knowledge and that the party made a reasonable investigation under the circumstances. Because AI systems are known to omit, misrepresent, or falsify information, practitioners should review the accuracy of all documents generated using AI. Unlimited reliance on AI tools does not constitute reasonable investigation, and failure to correct mistakes made by AI tools may result in disciplinary action. To eliminate the risk of submitting false statements, practitioners should independently verify the accuracy of AI-generated text.
Use of AI in research
Practitioners are increasingly using AI tools for various research-related tasks, including prior art searches and predictive analysis of examiner behavior. USPTO examiners are already using AI-powered tools such as More Like This Document and Similarity Search to conduct prior art searches during application review. Both IP practitioners and patent examiners can use AI to reduce the time required to identify similar characteristics of previously registered patents and pending applications. AI-powered research tools have the potential to improve the accuracy of prior art searches and reduce the average time cost.
As with document drafting, patent practitioners should exercise caution when using AI tools to conduct research. Many AI systems retain user-entered data for training AI models or sharing with third parties. Thus, when practitioners enter client information into an AI-powered search tool to search for prior art, they should carefully review the system's data privacy policy to ensure that sensitive client data is not being shared. The USPTO updated its rules in May 2021 to address emerging concerns related to the disclosure of client information. The added rule, 37 CFR 11.106(d), states:[a] “Practitioners must use reasonable efforts to prevent inadvertent or unauthorized disclosure of, or unauthorized access to, information relevant to the representation of a client.” To avoid the risk of confidentiality breaches when using AI for research, practitioners should closely examine systems' data policies and act with caution regarding the type of information shared.
Using AI to Access the USPTO System
In addition to applications in document preparation and research, AI can be used to interact directly with the USPTO’s online services. For example, AI tools can be used to assist with document submission, auto-filling forms, or uploading information to the USPTO servers. Automating certain mechanical aspects of patent filing and maintenance with AI has the potential to reduce errors and save time by facilitating redundant and predictable tasks with minimal human involvement. While AI’s powerful and ever-expanding capabilities make it an attractive tool for practitioners who want to work more efficiently, USPTO.gov user policies can impose substantial risks for practitioners who use these tools directly with the USPTO’s online resources. Thus, practitioners should exercise extreme caution to ensure compliance when using AI tools that interact directly with the USPTO’s online services.
The USPTO website provides resources and services for researching, filing, and managing patent applications. To access many of the USPTO's online services, practitioners must first register and verify an account on USPTO.gov. This account is exclusive to the individual who registered the information and cannot be shared with other users, including AI systems. Practitioners who use AI to access USPTO's online services should be careful not to allow unauthorized access to the AI system. For example, only registered users can access the USPTO's Patent Center tools for submitting specifications and other application materials. Thus, practitioners should not use AI to upload and submit documents directly within the Patent Center. Providing an AI tool with access to the Patent Center or other services that require a USPTO.gov account violates USPTO policy and may result in the cancellation of a user's account. To avoid account cancellation and possible civil and criminal penalties, practitioners who use AI to interact with USPTO's online services should always be mindful of the type and scope of access provided to the AI system. Practitioners should generally refrain from using AI to perform tasks that require a USPTO.gov account.
Conclusion
AI has the power to transform the patent prosecution process by increasing efficiency, improving research capabilities, and facilitating the automation of various tasks. Responsible use of AI in the patent prosecution process will benefit society at large by empowering examiners and practitioners to perform prior art searches more quickly and accurately. Furthermore, advances in LLMs designed specifically for patent prosecution materials could significantly reduce the average time cost of preparing specifications, responses to examiner actions, and other documents. Additionally, AI can be used to facilitate the completion of repetitive and predictable tasks, such as collecting data and automatically filling forms. As these systems are trained on large sets of available data, the speed, accuracy, and overall usefulness of AI are constantly improving. Practitioners who leverage the vast capabilities of AI can improve the quality and efficiency of their services to their clients.
However, practitioners using AI must always keep in mind the various risks associated with its use.
The USPTO’s April 2024 guidance outlines many ways in which existing rules and policies apply to actual AI use at the Patent Office. When using AI to draft documents for submission to the USPTO, users should always review the documents and independently verify the accuracy and relevance of the AI-generated text. When using AI to conduct prior art searches or other forms of research, practitioners should be cautious about inputting confidential client information and carefully review the tool’s data policies to prevent inadvertent disclosure. Similarly, practitioners should be cautious when using AI tools to interact with the USPTO’s online services. Many of the tools and services on USPTO.gov are accessible only to registered users. AI tools cannot become registered users and should not be used to access resources reserved for registered users. The risks above are not an exhaustive list. Practitioners should always independently assess the risks of implementing new technology.
AI tools are already being used by patent examiners and IP practitioners to perform a variety of tasks within the patent prosecution workflow. As the power of these algorithms grows over time, we expect AI tools to become increasingly performant, accurate, and efficient. Practitioners must strive to implement AI responsibly to leverage its full potential, while recognizing the enduring need for human governance in risk mitigation.
This article was co-written by Gavin Dacier, 2024 Summer Associate.