{"id":"W4390617165","doi":"10.2196/51148","title":"Pure Wisdom or Potemkin Villages? A Comparison of ChatGPT 3.5 and ChatGPT 4 on USMLE Step 3 Style Questions: Quantitative Analysis","year":2024,"lang":"en","type":"article","venue":"JMIR Medical Education","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":50,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Test (biology); Cheating; Medical education; United States Medical Licensing Examination; Psychology; Scale (ratio); Mathematics education; Computer science; Medical school; Medicine; Social psychology; Cartography; Geography","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01914618,0.0005215533,0.0008610988,0.002645435,0.000583149,0.001516132,0.0009392519,0.0009988567,0.002144255],"category_scores_gemma":[0.1700535,0.0003032387,0.0007959411,0.001462719,0.001258527,0.002105861,0.002778965,0.001214066,0.000909851],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001306272,"about_ca_system_score_gemma":0.0009716331,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001580466,"about_ca_topic_score_gemma":0.002820626,"domain_scores_codex":[0.9838397,0.008527066,0.00202123,0.001320576,0.00368293,0.0006085113],"domain_scores_gemma":[0.7569168,0.2006484,0.01589628,0.004342244,0.01938077,0.002815607],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.003549865,0.001139957,0.6156456,0.003112641,0.0004521458,0.000599077,0.1296042,0.00218291,0.008738033,0.001703716,0.008878203,0.2243936],"study_design_scores_gemma":[0.0001156544,0.001954027,0.9317054,0.000625665,0.0001934583,0.0004727496,0.03893436,0.00976299,0.004660162,0.001225843,0.01018572,0.0001640678],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9926662,0.0001717928,0.002786565,0.0001914891,0.00004459931,0.0006057196,0.0006216742,0.0001278858,0.002784116],"genre_scores_gemma":[0.9856079,0.0001851355,0.008707568,0.0002688517,0.00004424132,0.002048997,0.001136512,0.0000777141,0.001923022],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01914618,"threshold_uncertainty_score":0.1012558,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.09793939183542286,"score_gpt":0.5100299225747479,"score_spread":0.412090530739325,"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."}}