{"id":"W4412072331","doi":"10.2196/70176","title":"Reducing Hallucinations and Trade-Offs in Responses in Generative AI Chatbots for Cancer Information: Development and Evaluation Study","year":2025,"lang":"en","type":"article","venue":"JMIR Cancer","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":13,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Preprint; Psychology; Data science; Computer science; World Wide Web","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":[0.0005587026,0.00007588978,0.0001378918,0.0003126858,0.0001048423,0.00002543415,0.00002021374,0.0000539741,0.00002465816],"category_scores_gemma":[0.0001465648,0.00007233587,0.000008222481,0.0002987856,0.00002219842,0.0002362542,0.00001044721,0.00009825987,4.270117e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005360344,"about_ca_system_score_gemma":0.001245274,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001669202,"about_ca_topic_score_gemma":0.01242218,"domain_scores_codex":[0.9991705,0.00005913168,0.0003607012,0.0001556977,0.0001295783,0.0001244305],"domain_scores_gemma":[0.9995738,0.0001013257,0.00005511097,0.00006965052,0.0001652253,0.00003483313],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.0003026118,0.0001285715,0.2312184,0.0001673273,0.00001953018,4.208486e-7,0.07498731,0.0003105347,0.0001072936,0.0001333779,0.0005968193,0.6920278],"study_design_scores_gemma":[0.0005580388,0.0001206066,0.9632029,0.0004744181,0.00002957048,8.624041e-7,0.01489082,0.009726703,0.003534077,0.0002904169,0.007070663,0.0001009276],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.977192,0.001351023,0.0001225935,0.01850631,0.0002481689,0.002489715,0.000004734496,0.000008390154,0.00007705689],"genre_scores_gemma":[0.9928961,0.0001912992,0.0002802424,0.00123963,0.00006952479,0.005121628,0.00001400227,0.000003885085,0.0001836767],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7319845,"threshold_uncertainty_score":0.6931871,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1862239589163959,"score_gpt":0.5187098469781782,"score_spread":0.3324858880617823,"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."}}