{"id":"W4360796975","doi":"10.1101/2023.03.16.23286914","title":"Learning to fake it: limited responses and fabricated references provided by ChatGPT for medical questions","year":2023,"lang":"en","type":"preprint","venue":"medRxiv","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":32,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University; Montreal Children's Hospital; Université de Montréal; Cegep de Saint Hyacinthe; Centre Hospitalier Universitaire Sainte-Justine","funders":"","keywords":"Quality (philosophy); Quartile; Psychology; Medical education; Interpretation (philosophy); Minor (academic); Computer science; Medicine; Humanities","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.001410974,0.0002277959,0.0004347539,0.0003234975,0.0002284536,0.00006104868,0.0001784593,0.0005003354,0.00007877201],"category_scores_gemma":[0.01858574,0.000202005,0.00007377055,0.0003289199,0.0000899061,0.00003440341,0.0001737284,0.0008372946,0.00008842789],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00009216189,"about_ca_system_score_gemma":0.0008818551,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001589534,"about_ca_topic_score_gemma":0.0004981442,"domain_scores_codex":[0.9976536,0.0002254732,0.0006578424,0.0006534931,0.0004413091,0.00036828],"domain_scores_gemma":[0.9971479,0.001396557,0.0001662398,0.00033944,0.0004420686,0.0005077559],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.003195023,0.0007269996,0.3774101,0.004213348,0.0004340169,0.00008341213,0.03990458,0.00009607858,0.007624899,0.0005861063,0.1518084,0.413917],"study_design_scores_gemma":[0.0007824961,0.007757477,0.2193653,0.01645939,0.001097386,0.0001440332,0.03168324,0.03319239,0.03557721,0.01314572,0.6378655,0.002929855],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.875334,0.0003910501,0.001819924,0.1196987,0.000859028,0.001471309,0.00004125495,0.0003075603,0.00007717783],"genre_scores_gemma":[0.9862542,0.001149092,0.0007250478,0.001470175,0.0005052125,0.001025083,0.0004388946,0.00005373995,0.008378518],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4860571,"threshold_uncertainty_score":0.9896811,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2205691592056989,"score_gpt":0.4621102855995091,"score_spread":0.2415411263938102,"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."}}