Category: experiment

  • Stanford Will Augment Its Study Finding that AI Legal Research Tools Hallucinate in 17% of Queries, As Some Raise Questions About the Results

    Stanford University released a study on generative AI legal research tools from LexisNexis and Thomson Reuters, finding that they deliver hallucinated results more often than claimed. The study has faced criticism for comparing different products and not testing one of Thomson Reuters’ AI platforms. Thomson Reuters has now made the product available to the researchers, […]

  • Artificial intelligence used to replicate Brown v. Board of Education oral arguments

    Brown v. Board of Education was decided 70 years ago. There are no existing audio recordings of the arguments in which Thurgood Marshall, the then-NAACP chief counsel, opposed school segregation, nor are there recordings of Chief Justice Earl Warren reading the opinion from the bench. Jerry Goldman, the founder of Oyez, utilised AI to replicate […]

  • I stumbled upon LLM Kryptonite – and no one wants to fix this model-breaking bug

    The challenges and potential risks of integrating large language models (LLMs) into various applications exist. The author describes encountering a widespread issue in which different AI-powered chatbots produced incoherent and endless responses to a simple prompt. After contacting the respective providers, the author received confirmation that the issue was resolved across multiple LLMs, indicating a […]

  • A Legal Risk Taxonomy for Generative Artificial Intelligence

    This paper presents a taxonomy of legal risks associated with generative AI (GenAI) by analysing existing and potential legal claims. The methodology used in this paper involves analysing (1) the legal claims filed in existing lawsuits and (2) the reasonably foreseeable legal claims that may be filed in future lawsuits. The paper identifies seven most […]

  • Exploring the Nexus of Large Language Models and Legal Systems: A Short Survey

    There is a significant shift in natural language processing tasks in the legal sector, thanks to the development of Large Language Models (LLMs) and Artificial Intelligence (AI). LLMs are now proving exceptionally useful in the legal domain, offering unique advantages and presenting various challenges. This survey examines the relationship between LLMs and the legal system, […]

  • The responsible machine problem

    The legal system faces challenges in extending its principles to machines. Liability, accountability and agency are complex issues attributing to entities without consciousness or intent. As technology advances, the legal system must evolve to address the unique challenges posed by integrating machines into societal roles traditionally held by humans. AI systems are becoming increasingly capable […]

  • Why Large Language Models Like ChatGPT Treat Black- and White-Sounding Names Differently

    The co-author of a paper on auditing large language models for race and gender bias is Julian Nyarko, a Stanford Law School Professor and also an Associate Director of Stanford HAI. The paper discusses how popular large language models handle specific queries involving first and last names indicative of race or gender. According to the […]

  • 16 Changes to the Way Enterprises Are Building and Buying Generative AI

    Generative AI proved to be a significant success with consumers in 2023, with sales exceeding one billion dollars in record time. The enterprise revenue opportunity in 2024 is expected to be even more critical. However, last year, most enterprise engagement with Generative AI was limited to only a few obvious use cases and shipping “GPT-wrapper” […]

  • Legal sentence boundary detection using hybrid deep learning and statistical models

    The authors propose leveraging Deep learning models, including LegalBERT and CaseLawBERT, to detect sentence boundaries in English legal texts. After evaluating various deep learning models, the Convolutional Neural Network Model (CNN) is the top-performing model.  A hybrid architecture that combines the strengths of both models, achieving a 4% improvement in the F1-score. (The F1 score […]

  • Will Large Language Models Really Change How Work Is Done?

    Utilising Large Language Models (LLMs) effectively within an organisational context is more challenging than most people assume. The most popular and practical applications of LLMs include manipulating existing information, customer interactions, knowledge management, and software engineering. However, LLMs have not yet demonstrated that they can satisfactorily perform all the tasks knowledge workers execute in any […]