{"id":"W4400047099","doi":"10.1093/jla/laae003","title":"Large Legal Fictions: Profiling Legal Hallucinations in Large Language Models","year":2024,"lang":"en","type":"article","venue":"The Journal of Legal Analysis","topic":"Artificial Intelligence in Law","field":"Social Sciences","cited_by":192,"is_retracted":false,"has_abstract":true,"ca_institutions":"Institute on Governance","funders":"","keywords":"Typology; Hallucinating; Profiling (computer programming); Psychology; Criminology; Law; Political science; Social psychology; Sociology","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004897761,0.0001277861,0.0003113869,0.0007305808,0.0006213278,0.0004121588,0.000535322,0.00009008325,0.0005336605],"category_scores_gemma":[0.0003382771,0.00009112238,0.0004335746,0.003033713,0.0001501958,0.001784098,0.000057799,0.0006588643,0.00004518895],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002539833,"about_ca_system_score_gemma":0.0003846796,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.00698277,"about_ca_topic_score_gemma":0.05767747,"domain_scores_codex":[0.9974256,0.0005361529,0.000695395,0.0001567718,0.0007586377,0.0004274125],"domain_scores_gemma":[0.9988409,0.0003875919,0.0002005047,0.0002023954,0.0002629838,0.0001055786],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00004014147,0.0002469022,0.00109173,0.0000127613,0.0008475963,0.0002488239,0.04555769,0.05787428,0.0004744795,0.8919711,0.0007470509,0.0008875162],"study_design_scores_gemma":[0.0003354718,0.0001633551,0.0006665798,0.0002452633,0.003701441,0.00007421943,0.2267784,0.5971668,0.002058098,0.02178692,0.1463411,0.0006822521],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7368293,0.00593226,0.1778313,0.01909184,0.001200211,0.0003624153,0.00008768706,0.0001362968,0.05852874],"genre_scores_gemma":[0.9941667,0.0002305715,0.000317516,0.000131694,0.0005693408,0.000004052925,0.00000405322,0.0000128652,0.004563183],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8701841,"threshold_uncertainty_score":0.9996298,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03341118742895829,"score_gpt":0.3685634204669446,"score_spread":0.3351522330379862,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}