{"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":"codex-gemma-dda1882f352a","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.000528822,0.0001603971,0.0004365582,0.0005853169,0.0001001862,0.00003977416,0.00008423044,0.0002461963,0.001120808],"category_scores_gemma":[0.001477077,0.0001214621,0.0001059321,0.001340464,0.0001971009,0.0001267705,0.00001904008,0.000389477,0.0000792997],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001534541,"about_ca_system_score_gemma":0.001763649,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008042576,"about_ca_topic_score_gemma":0.0005882004,"domain_scores_codex":[0.9979862,0.0001342529,0.0006452378,0.0003903333,0.0006313476,0.0002126552],"domain_scores_gemma":[0.9983733,0.0005528773,0.0001420855,0.0002837457,0.0002802442,0.0003677155],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"qualitative","study_design_scores_codex":[0.0009258695,0.003851858,0.02270549,0.002415129,0.0007547861,0.00002872948,0.0722746,0.00002352157,0.0003643194,0.02253629,0.03320001,0.8409194],"study_design_scores_gemma":[0.0009307156,0.0130776,0.1435526,0.01896139,0.004892137,0.0003242088,0.3483027,0.2683905,0.008429176,0.007247815,0.1840386,0.001852552],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9669396,0.003974319,0.00138342,0.02520104,0.001075298,0.0007872503,0.00001087997,0.00008998065,0.0005382427],"genre_scores_gemma":[0.9961745,0.0004567601,0.0007318804,0.0008060912,0.0004585767,0.0002109218,0.0001455919,0.00001936397,0.0009962913],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8390669,"threshold_uncertainty_score":0.9997923,"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."}}