Journal of Empirical Legal Studies最新文献

Legal practice has witnessed a sharp rise in products incorporating artificial intelligence (AI). Such tools are designed to assist with a wide range of core legal tasks, from search and summarization of caselaw to document drafting. However, the large language models used in these tools are prone to “hallucinate,” or make up false information, making their use risky in high-stakes domains. Recently, certain legal research providers have touted methods such as retrieval-augmented generation (RAG) as “eliminating” or “avoid[ing]” hallucinations, or guaranteeing “hallucination-free” legal citations. Because of the closed nature of these systems, systematically assessing these claims is challenging. In this article, we design and report on the first preregistered empirical evaluation of AI-driven legal research tools. We demonstrate that the providers' claims are overstated. While hallucinations are reduced relative to general-purpose chatbots (GPT-4), we find that the AI research tools made by LexisNexis (Lexis+ AI) and Thomson Reuters (Westlaw AI-Assisted Research and Ask Practical Law AI) each hallucinate between 17% and 33% of the time. We also document substantial differences between systems in responsiveness and accuracy. Our article makes four key contributions. It is the first to assess and report the performance of RAG-based proprietary legal AI tools. Second, it introduces a comprehensive, preregistered dataset for identifying and understanding vulnerabilities in these systems. Third, it proposes a clear typology for differentiating between hallucinations and accurate legal responses. Last, it provides evidence to inform the responsibilities of legal professionals in supervising and verifying AI outputs, which remains a central open question for the responsible integration of AI into law.

Patent law encourages inventors to label their products as “patented,” to mark their legal status and potentially secure monetary damages from infringing competitors. We examine whether such labels might have a separate and direct impact on consumers, by affecting how they view patented products and influencing what they purchase. We develop and conduct two experiments to isolate the impact of patent status on consumer behavior. In an online randomized experiment, we demonstrate how increasing the salience of patent status heightens consumers' beliefs that products are innovative and well made. We also reveal consumers' surprisingly sophisticated understanding of the patenting process and what being patented means. Despite this informed perspective, consumers are not more inclined to buy patented products. To determine if these results hold in a real-world setting, we conduct a field experiment at a small retail pharmacy chain. Using scanner data spanning over 4 years, we find no evidence that consumers respond to increased patent salience. Our collective results suggest that while consumers view patented products as more innovative and well made, these positive attributes do not necessarily translate into heightened purchasing behavior. Our research suggests that patent marking might often serve only a legal, rather than a marketing, function. This has implications for patent policy, including the relevance of patent salience in damage analyses and litigation strategies.

Choi, S. J., J. M. Erickson, and A. C. Pritchard. 2024. “Paying for Performance? Attorneys' Fees in Fraud Class Actions.” Journal of Empirical Legal Studies 21, no. 4: 899–926. https://doi.org/10.1111/jels.12402.

The title of the article should be “Paying for Performance? Attorneys' Fees in Securities Fraud Class Actions.” The word “securities” was inadvertently omitted from the title.

We apologize for this error.

Since the United States Supreme Court's 2012 decision in Mayo v. Prometheus announced a new legal test for patent-eligible subject matter, policymakers, and scholars have vigorously debated the decision's impact on molecular diagnostics innovation. Molecular diagnostics serve as the cornerstone of personalized medicine and its promise of treatments with fewer side effects and better outcomes for patients. This article contributes to the presently thin evidence base on the impact of Mayo by using data on patent applications, examinations, and grants from 2010 to 2019 to comprehensively trace the effects of the test and subsequent related developments. Using descriptive data as well as a difference-in-difference (DID) design we evaluate the extent to which the decision was followed by one of three expected outcomes: a decline in patent quantity (“retrenchment”); increase in patent prosecution “toughness”; and applicant “adaptation” with respect to submitted claims. We find substantial support for our toughness and adaptation hypotheses, but not our retrenchment hypothesis: molecular diagnostic patenting did not decline in aggregate, though there is some evidence of a decline, relative to a control, in the number of diagnostic patent applications and grants associated with small, U.S.-based firms. These results suggest that molecular diagnostic patents are harder to get but they are still being applied for and granted, with their narrowed scope making them less likely to block follow on innovation.

Which judicial leaders are more successful in achieving consensus? This article examines the impact of the leadership of presiding justices on consensus formation on the Norwegian Supreme Court where cases are distributed randomly to two parallel decisional panels. We hypothesize that presiding justices with certain characteristics (e.g., gender and chief justice), when in charge of the decision-making process, are more willing and better able to forge consensus, which could lead to greater respect for courts and the rule of law. We account for a variety of characteristics of the justices, as well as several conditions under which the cases were decided. The results confirm that both chief justices and female justices, when operating as the presiding justice of the panel, are significantly more likely to steer the case towards a unanimous decision as compared to their fellow justices. Legal academics serving as presiding justices had no discernable impact on consensus formation. The results provide evidence outside the American context that chief justices and women justices have the ability to achieve greater consensus. As such, diversity and appointments have consequences for judicial leadership and for consensus formation on a court.

This paper introduces a novel approach to measure a patent's economic value by examining whether the patent's disclosure leads to rejection of another pending US patent application. This approach considers the use of the patent by the US Patent and Trademark Office (USPTO) in office action rejections on the grounds of novelty or obviousness, as well as its citation as an X or Y reference in a European Patent Office (EPO) search report, which provides analogous information. Unlike conventional citation metrics widely employed by economists, the novel metric is arguably more closely tied to private value, as it is centered on the examiner's rejection of another patent application based on the patent in question. As this metric is not directly influenced by potential strategic behavior by patent lawyers or patent applicants, it is more directly tied to private value. This study evaluates how patents used in novelty and obviousness rejections, and comparable EPO information, correspond to common measures of private value—specifically patent renewal, the assertion of a patent in litigation, the number of independent patent claims, and classification in multiple technological subject matters. We examine rejection data over a defined time period for US patents issued from 1999 to 2007 and then link value data to these patents. A similar analysis is also conducted for EPO search reports. Our findings reveal that rejection uses, as well as X and Y EPO search citations, independently exhibit positive correlations with all of these value measurements. Moreover, when information from both the USPTO and EPO concerning a given patent is combined, further insights about that patent's value are obtained. We also find that rejection uses provide unique insights into additional measures of private value such as the patent being listed in the Orange Book or as a Standard Essential Patent. Accordingly, these rejection use metrics provide another independent tool for evaluating a patent's value.

Over the last three decades, plea bargaining has been adopted by many jurisdictions worldwide. However, a comprehensive account of both its adoption as well as its use is still missing. We survey 174 jurisdictions, finding that 101 allow plea-bargaining. For 52 jurisdictions, we also compute plea-bargaining rates, as the percentage of convictions imposed through plea bargaining over all criminal convictions. Relying on this novel dataset, we find that Muslim-majority populations and the French and Scandinavian legal origins are associated with lower probabilities of formalizing plea bargaining, while democracies are associated with higher probabilities. The Spanish and Socialist legal origins, a looser regulation of the procedure, and jury trials are associated with higher plea-bargaining rates, while higher income levels correlate with higher plea-bargaining rates only up to a certain point.