{"id":"W7103204945","doi":"","title":"What's In My Human Feedback? Learning Interpretable Descriptions of Preference Data","year":2025,"lang":"","type":"article","venue":"ArXiv.org","topic":"Explainable Artificial Intelligence (XAI)","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Canadian Institute for Advanced Research; Open Philanthropy Project; National Science Foundation","keywords":"Preference; Diversity (politics); Preference elicitation; Preference learning; Work (physics)","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006859833,0.00166815,0.0007228374,0.001561192,0.0006197551,0.00218192,0.001074366,0.001553223,0.003885813],"category_scores_gemma":[0.04852922,0.0004162579,0.0007806442,0.001227016,0.001212457,0.005603911,0.001445259,0.002323275,0.001895226],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001257366,"about_ca_system_score_gemma":0.001109322,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005762369,"about_ca_topic_score_gemma":0.01388505,"domain_scores_codex":[0.9922013,0.00473958,0.0003646098,0.001515611,0.0009171588,0.0002616783],"domain_scores_gemma":[0.9757136,0.01519587,0.00166329,0.004039199,0.002855518,0.0005325759],"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":[0.003352661,0.0005799874,0.1565993,0.002754587,0.0004693004,0.0006913964,0.0091564,0.0361217,0.03227331,0.02565112,0.08382836,0.6485218],"study_design_scores_gemma":[0.000328648,0.00101769,0.05622274,0.0009778035,0.0002336621,0.001262995,0.005816895,0.6585189,0.02582613,0.1305639,0.1187532,0.0004775392],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2960616,0.004265046,0.6521695,0.007737801,0.0006788662,0.0004481153,0.01561845,0.007725903,0.01529477],"genre_scores_gemma":[0.8431315,0.0006481139,0.1399841,0.001898363,0.0001811135,0.0003718686,0.009057983,0.0006397585,0.00408713],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006859833,"threshold_uncertainty_score":0.03627872,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1960479997698643,"score_gpt":0.3504283725767246,"score_spread":0.1543803728068603,"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."}}