The recent update to Stanford’s research on AI legal research tools revealed concerning hallucination rates in Westlaw’s AI-powered tool. The study, titled “Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools,” faced previous criticism for seemingly unequal comparisons. The updated report, available here, incorporated data from Thomson Reuters’ AI-Assisted Research on Westlaw Precision, highlighting the distinct challenges in ensuring AI reliability.
In response, Thomson Reuters emphasized its lower hallucination rates based on internal tests and expressed interest in collaborating with Stanford to develop AI benchmarks, critical for advancing legal tech’s accuracy. For further details, refer to the complete article.