{"id":"W4376133361","doi":"10.1126/sciadv.abq0701","title":"Judging facts, judging norms: Training machine learning models to judge humans requires a modified approach to labeling data","year":2023,"lang":"en","type":"article","venue":"Science Advances","topic":"Explainable Artificial Intelligence (XAI)","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"ca_institutions":"Schwartz/Reisman Emergency Medicine Institute; Vector Institute; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Government of Canada; Canadian Institute for Advanced Research; Microsoft Research; University of Toronto; University of Southern California","keywords":"Normative; Computer science; Artificial intelligence; Object (grammar); Machine learning; Normative model of decision-making; Core (optical fiber); Psychology; Cognitive psychology; Epistemology","routes":{"ca_aff":true,"ca_fund":true,"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":["metaepi_narrow","sts","scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.004739987,0.0003842688,0.0004251263,0.001274979,0.002184348,0.001274124,0.007407859,0.00005846654,0.000003930686],"category_scores_gemma":[0.00128713,0.0003692225,0.00005928101,0.008020204,0.0003164703,0.0108549,0.003503838,0.0003598129,0.0002356686],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001776569,"about_ca_system_score_gemma":0.0003637055,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003075294,"about_ca_topic_score_gemma":0.000157706,"domain_scores_codex":[0.9935908,0.0001039498,0.0006483115,0.002319955,0.001481024,0.001856001],"domain_scores_gemma":[0.9967476,0.0002965694,0.0001902674,0.00194998,0.0002558297,0.0005597852],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000006896636,0.00002851234,0.00009185137,0.00002544947,0.000005190952,0.00001506172,0.0205963,0.821354,0.02527298,0.02125992,0.00004389518,0.1113],"study_design_scores_gemma":[0.00007739698,0.00009094371,0.00003388009,0.0001149821,0.00000487958,0.000007264922,0.006735488,0.9697468,0.01097747,0.009734775,0.001932397,0.0005437683],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.07012616,0.0002894,0.9210641,0.00160939,0.0006319886,0.0005391418,0.00001440916,0.001179633,0.004545735],"genre_scores_gemma":[0.8691677,0.00004611898,0.1293977,0.0007013685,0.0001100862,0.00007063526,0.00001805523,0.00003369341,0.0004545987],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.7990416,"threshold_uncertainty_score":0.999876,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2543949435950519,"score_gpt":0.357115698066272,"score_spread":0.1027207544712201,"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."}}