As attorneys, we’ve all taken Legal Research and Writing. This is where we first encountered Westlaw and Lexis, using these sites to delve into case law for various assignments while chasing Westlaw and Lexis points in the hopes of getting a free TV or iPod (back when that was a thing). Professors always emphasized the critical process of tracing the history of a case and determining if the case is still considered “good law.”
With the rise of generative AI over the past year, it’s unsurprising that lawyers are turning to this advanced technology for legal research. However, there’s a growing concern: the blind acceptance of AI-generated content. This issue was highlighted in U.S. Supreme Court Chief Justice John Robert’s 2023 year-end report on the federal judiciary. A notable incident involved the Second Circuit reprimanding a New York attorney for submitting a brief with reference to a non-existent case, hallucinated by ChatGPT, without proper verification.
This alarming trend has led U.S. Patent and Trademark Office Director Kathi Vidal to release a memorandum on February 6, affirming that the current mechanisms are sufficient to address both the proper and improper use of AI in legal settings. Vidal promises more detailed guidance in the coming months. The memo acknowledges the potential of AI to increase accessibility and reduce costs in front of the PTAB and TTAB. However, it also raises concerns about the misuse of AI, drawing parallels with issues previously seen in federal courts, such as citing irrelevant sources.
Vidal’s memo emphasizes that existing USPTO rules, which require parties to maintain the integrity of proceedings, apply equally to AI-generated submissions, saying those rules “apply regardless of how a submission is generated … simply assuming the accuracy of an AI tool is not a reasonable inquiry.”
Nearly a year since the commercial debut of generative AI, it’s clear that while generative AI offers innovation, it also has limitations, such as the tendency to produce factually incorrect information (“hallucinations”). For lawyers, generative AI should be viewed as an aid, not a substitute for thorough legal processes. Unbridled reliance on generative AI is comparable to citing a case without Shepardizing – a significant oversight. Please check out our webinar for more information on the best practices for the use of AI in both litigation and non-litigation settings.
Given that the USPTO recognizes generative AI’s potential to make legal processes more accessible and cost-effective, the USPTO does not seem inclined to ban generative AI usage (assuming that is even possible). Thus, the direction of upcoming guidance is a guessing game. Perhaps the USPTO may consider stricter penalties for those misusing AI-generated content. This may be a step in preventing its misuse in the legal profession.