{"id":"W4416996294","doi":"10.1007/s43681-025-00838-x","title":"Toward ethical AI through Bayesian uncertainty in neural question answering","year":2025,"lang":"en","type":"article","venue":"AI and Ethics","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"HealthForceOntario","funders":"","keywords":"Interpretability; Bayesian probability; Bayesian inference; Inference; Artificial neural network; A priori and a posteriori; Calibration; Question answering","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00659778,0.000564774,0.001048919,0.001608248,0.0008832276,0.003324001,0.001862673,0.00268559,0.003441004],"category_scores_gemma":[0.04339514,0.0007367936,0.0009987049,0.001281539,0.002283186,0.007866592,0.003255032,0.005315971,0.0005628802],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001720927,"about_ca_system_score_gemma":0.001400312,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005752949,"about_ca_topic_score_gemma":0.00589543,"domain_scores_codex":[0.9965024,0.002354667,0.0001342115,0.0004567243,0.0003910676,0.0001608717],"domain_scores_gemma":[0.9688167,0.02812281,0.0007386344,0.0008956577,0.001044065,0.0003820609],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0003013768,0.0002736003,0.004762156,0.0004290953,0.0002395387,0.0001298774,0.001886411,0.2027911,0.001880881,0.5623908,0.01149306,0.213422],"study_design_scores_gemma":[0.00001122973,0.000009800126,0.0002555582,0.00002851364,0.00001918984,0.00001790659,0.00007694127,0.5276681,0.000390901,0.4702534,0.001257437,0.00001096574],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02773197,0.0009871862,0.9575053,0.007444353,0.00007187794,0.00006328082,0.0002530342,0.0003757062,0.005567309],"genre_scores_gemma":[0.7900302,0.0007866109,0.2035187,0.00123738,0.0003242999,0.0001782171,0.0006828796,0.0001111751,0.003130501],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.00659778,"threshold_uncertainty_score":0.03489286,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05108168716002209,"score_gpt":0.3555435753322798,"score_spread":0.3044618881722577,"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."}}