The ongoing evolution of artificial intelligence in the legal profession was front and center at a recent Southern California Association of Law Libraries meeting, where a panel of law librarians assessed the performance of three major AI-driven legal research platforms: Lexis+ AI, Westlaw Precision AI, and vLex’s Vincent AI. The analysis aimed to determine how these platforms stack up when providing direct responses to legal inquiries, and the findings highlighted each platform’s distinct strengths and limitations. Interested readers can find further details here.
The meeting involved the evaluation of AI responses to three legal questions concerning California state law or federal law within the Ninth U.S. Circuit Court. Each AI platform’s effectiveness was assessed based on six specific factors: accuracy, depth, citation of primary and secondary sources, presentation format, and iterative capabilities.
In addressing a question about interlocutory appeal certification timeframes, all three platforms accurately cited the governing statute. Lexis+ AI and Westlaw Precision AI provided similar, well-integrated answers, while Vincent AI showed slight variations in the authorities it cited, though it succeeded in offering a novel perspective on the subsequent query concerning calculation methods.
The nuanced differences became more pronounced with a query about the California Reproductive Leave Loss for Employees Act. Here, Vincent AI diverged significantly by identifying a potential regulatory provision relevant to a private right of action, although the panelists noted this conclusion derived from an implicit interpretation, underscoring AI’s unpredictable nature.
Regarding the standard of appealing class certification, the platforms identified the abuse of discretion standard accurately. Differences arose in relating to the “death knell” doctrine. Vincent AI delivered a stand-out comprehensive answer and presented iterative options, allowing users to modify and regenerate queries, which was praised for its user-control flexibility.
In sum, the panelists concluded that while these AI platforms offer valuable starting points for legal research, they remain far from foolproof. Verifying AI-generated information against traditional legal resources remains a crucial step, as highlighted by panelist Tanya Livshits, who stressed AI’s role as a “tool” rather than a catch-all solution.
The thorough review also pointed to a lack of transparency regarding AI models used by these vendors, prompting calls for increased clarity on their technological underpinnings. As the legal community continues to adapt to these advancements, caution and user oversight remain paramount in utilizing AI-based research tools effectively.