{"id":"W4411035583","doi":"10.1080/29974100.2025.2503343","title":"Evaluating firearm examiner testimony using large language models: a comparison of standard and knowledge-enhanced AI systems","year":2025,"lang":"en","type":"article","venue":"Journal of Psychology and AI","topic":"Digital and Cyber Forensics","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"York University; New York University Shanghai","keywords":"Computer science; Forensic engineering; Natural language processing; Engineering","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":[0.0005632032,0.00008805495,0.0003151428,0.0001194448,0.00006258927,0.00006955316,0.0001777861,0.0000649627,0.000001338593],"category_scores_gemma":[0.00004508527,0.00006821377,0.00003611175,0.0001870424,0.0000777911,0.0003519059,0.00009305255,0.0001917559,3.422118e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001058366,"about_ca_system_score_gemma":0.00007221696,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000004436046,"about_ca_topic_score_gemma":0.000004371335,"domain_scores_codex":[0.9991343,0.00007536538,0.000384236,0.0001421019,0.0001246621,0.0001393292],"domain_scores_gemma":[0.9992836,0.00008609412,0.000224333,0.0001357886,0.0002194224,0.00005072548],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000553509,0.001453778,0.01854138,0.0006546754,0.0005902259,0.0001007186,0.02450573,0.007645387,0.03803805,0.08079702,0.01343432,0.8136852],"study_design_scores_gemma":[0.004345328,0.001674482,0.005715766,0.001354131,0.0001283001,0.0003206191,0.001149066,0.9536552,0.002497874,0.02751637,0.001323968,0.0003188838],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6596945,0.007304956,0.3294176,0.0002756185,0.0004557864,0.00005118153,0.00000390852,0.000008430577,0.002788046],"genre_scores_gemma":[0.9950197,0.00002760686,0.004536864,0.0002390859,0.00003977597,5.87972e-7,2.714277e-7,0.000003236973,0.0001329232],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9460098,"threshold_uncertainty_score":0.2781677,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06091395159346706,"score_gpt":0.4294073973453022,"score_spread":0.3684934457518351,"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."}}