In a stimulating experiment, OpenAI’s Deep Research was put to the test against established legal research AI platforms Lexis+AI, Westlaw Precision AI, and vLex’s Vincent AI at an AI Smackdown organized by the Southern California Association of Law Libraries. The focus was on a legal question concerning the standards for appealing class certification in California, which presented complexities due to the state’s “death knell” doctrine—mandating immediate appeals on denials of class actions.
Deep Research distinguished itself in the contest by effectively recognizing and emphasizing the death knell doctrine, unlike some of its commercial counterparts. Lexis+AI omitted mention of this critical issue, and while Westlaw Precision AI did touch upon the need for immediate appeal, it risked confusion. Only vLex’s Vincent AI supported a detailed response that highlighted the issue.
An intriguing aspect of Deep Research’s operation is its running commentary during the research process, which quickly zeroed in on this pivotal issue. Without access to paywalled resources, unlike its competitors, Deep Research relied solely on public sources, including Casetext, a site yet available for free despite its acquisition by Thomson Reuters.
The results prompt speculation about the potential of OpenAI’s tool with comprehensive access to legal databases. Deep Research provided an extensive analysis of the issue, detailing the specific conditions under which the death knell doctrine applies, the procedural framework for appeals in such cases, and the standards of review applied by appellate courts.
This exploration not only highlights the capabilities of Deep Research in handling complex legal inquiries but also raises questions about the future landscape of legal research should public-facing AI tools gain access to proprietary legal information.
For a comprehensive examination of how Deep Research handled the challenge, the full memo produced during this test can be viewed here.