{"id":"W4403482299","doi":"10.2196/60827","title":"The Comparative Sufficiency of ChatGPT, Google Bard, and Bing AI in Answering Diagnosis, Treatment, and Prognosis Questions About Common Dermatological Diagnoses","year":2024,"lang":"en","type":"article","venue":"JMIR Dermatology","topic":"Social Media in Health Education","field":"Social Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Preprint; Medical diagnosis; Medicine; Dermatology; World Wide Web; Computer science; Pathology","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":[],"consensus_categories":[],"category_scores_codex":[0.0002357625,0.0001286281,0.0003316225,0.0001385473,0.0005854488,0.00009538973,0.0001184941,0.0001620401,0.000007455271],"category_scores_gemma":[0.0004107826,0.00009866921,0.00003282703,0.0004648001,0.001049889,0.000185878,0.00003990648,0.0001759793,0.000004476967],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001487651,"about_ca_system_score_gemma":0.0002631821,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004106024,"about_ca_topic_score_gemma":0.01424751,"domain_scores_codex":[0.9983158,0.0005848458,0.0003643338,0.0002501122,0.0001162085,0.000368721],"domain_scores_gemma":[0.9931121,0.006550247,0.00008517279,0.00009512752,0.00004819201,0.0001091939],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.000006768744,0.0001682451,0.8683791,0.00009519007,0.00001543034,0.00001783039,0.07499309,0.000001478314,0.0000127846,0.04460518,0.0002446889,0.0114602],"study_design_scores_gemma":[0.0004160744,0.0001934863,0.8699771,0.0004470365,0.00004827918,0.00008613963,0.02525762,0.00047707,0.0004764127,0.004381131,0.09794791,0.0002917143],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9659256,0.01730158,0.000006597226,0.01428727,0.0006988298,0.001329929,0.000006228712,0.0000754371,0.0003684661],"genre_scores_gemma":[0.9851269,0.01096457,0.00005264892,0.00008402084,0.00006729885,0.003689938,0.000003575833,0.000007338439,0.000003705033],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.09770322,"threshold_uncertainty_score":0.7950445,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.08202554106877735,"score_gpt":0.445074765340222,"score_spread":0.3630492242714446,"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."}}