{"id":"W4412708207","doi":"10.1016/j.aap.2025.108178","title":"Assessing the efficacy of pre-trained large language models in analyzing autonomous vehicle field test disengagements","year":2025,"lang":"en","type":"article","venue":"Accident Analysis & Prevention","topic":"Traffic and Road Safety","field":"Engineering","cited_by":4,"is_retracted":false,"has_abstract":false,"ca_institutions":"","funders":"National Research Council Canada; National Highway Traffic Safety Administration; National Research Council","keywords":"Disengagement theory; Field (mathematics); Computer science; Risk analysis (engineering); Poison control; Control (management); Data science; Applied psychology; Engineering; Artificial intelligence; Psychology; Business; Medicine; Environmental health","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006046569,0.000122923,0.0002582542,0.0003530943,0.0000912345,0.00006281213,0.0002176955,0.00006537662,0.00006325562],"category_scores_gemma":[0.00008654927,0.0001021895,0.0002638049,0.001272117,0.000009653522,0.0003327031,0.00006635945,0.0001797732,0.00000211696],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007711125,"about_ca_system_score_gemma":0.00001807746,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00009151523,"about_ca_topic_score_gemma":0.001660704,"domain_scores_codex":[0.9989203,0.00009743108,0.0004551124,0.0001821114,0.0001439185,0.0002011523],"domain_scores_gemma":[0.9992653,0.0002917344,0.00008554962,0.000312731,0.00002298297,0.00002171985],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.000009048735,0.0002145696,0.5022503,0.00001862029,0.0008959707,0.000003306772,0.001193855,0.4683518,0.001060931,0.0002556782,0.00007746379,0.02566847],"study_design_scores_gemma":[0.0003736358,0.000006648434,0.6200342,0.00006016756,0.0006028632,5.828525e-8,0.0004333818,0.3777133,0.0005701126,0.0001238301,0.000008939094,0.0000728228],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8747321,0.000490918,0.1233026,0.00007259486,0.00005844495,0.0001779556,4.83574e-7,0.00009138617,0.001073524],"genre_scores_gemma":[0.9992691,0.0000510654,0.0002222807,0.0000170632,0.00002009929,0.00001497145,0.00003825055,0.000009476499,0.0003576376],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1245371,"threshold_uncertainty_score":0.4167166,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.009146477039634901,"score_gpt":0.2989159307987034,"score_spread":0.2897694537590685,"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."}}