Tests how AI assistance changes law students’ speed and work quality across six realistic legal assignments.
Legal AI research
Papers, evidence and connections across jurisdictions.
A statutory-survey evaluation shows how model errors, retrieval failures and incomplete human answer keys interact.
A three-month patent-lawyer experiment separates better AI-assisted work from retained professional skill.
A direct human–model comparison separates resistance to prompt changes from agreement with ordinary speakers in insurance interpretation.
Singapore case-citation retrieval tests the value of stating the legal principle for which authority is sought.
Graded relevance judgments for drafting-oriented clause retrieval.
A field study of AI adoption and support in legal aid, drawing on practitioners’ reported experiences.
A preregistered, manually labelled evaluation of proprietary legal RAG products.
Observed judicial AI use and a judge experiment examine the risks of generating reasons after choosing a decision.
Compares human and model workflows through simulated business-formation cases, extending evaluation beyond final answers.
Arabic legal evaluation using Saudi sources and translated English-language benchmark tasks.
Compares models, junior lawyers and legal-process outsourcers on finding and locating contractual issues.
A law-student experiment compares AI highlighting and summaries, showing why interface choices matter for legal work.
Indian legal understanding across eight tasks, with translation between English and nine Indian languages.
Separates externally checkable answers from contradiction-based hallucination estimates.
Groups legal tasks into memorisation, understanding and application.
Chinese criminal case retrieval with expert judgments of substantive and procedural relevance.
Transforms existing legal annotations into retrieval queries with character-span targets.
Combines four corpora and evaluates adapted multilingual and monolingual encoders.
Re-examines GPT-4's bar-exam headline through comparison populations, score construction and evaluation transparency.
Uses tax-law problems and synthetic statutes to separate knowledge of legal text from applying rules to facts.
LegalBench: A Collaboratively Built Benchmark for Measuring Legal Reasoning in Large Language Models
Connects task evaluation to six legally recognisable reasoning categories.
A common framework for evaluating legal classification and entity recognition across eleven datasets and 24 languages.
Expert-annotated questions about the meaning of merger agreement provisions.
Korean judicial datasets for classification, outcome prediction and summarisation.
Examines group disparities across four legal datasets, exposing differences that overall accuracy can conceal.
Unifies seven English datasets with accessible evaluation code and comparable metrics.
Expert-written summaries at three lengths for long collections of civil-rights litigation documents.
A legal training corpus and study of privacy and toxicity filtering grounded in legal context.
Pairs document-level entailment judgments with supporting evidence spans.
Expert-supervised clause annotations across multiple commercial contract types.
An Indian Supreme Court corpus for studying judgment prediction and the passages used to explain it.
A Chinese criminal-case dataset linking court-written facts to charges, statutory provisions and prison terms.