{"id":"W4409362112","doi":"10.1609/aaai.v39i26.34987","title":"Evaluating Mathematical Reasoning Beyond Accuracy","year":2025,"lang":"en","type":"article","venue":"Proceedings of the AAAI Conference on Artificial Intelligence","topic":"Mathematics Education and Teaching Techniques","field":"Social Sciences","cited_by":10,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Computer science; Management science; Artificial intelligence; Engineering","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":[0.02455784,0.003421987,0.001456868,0.004718421,0.001164436,0.006459234,0.003582394,0.003667388,0.007608385],"category_scores_gemma":[0.151347,0.000816071,0.002319398,0.002380515,0.00211629,0.008779842,0.005423318,0.004356566,0.003320084],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002839947,"about_ca_system_score_gemma":0.004050849,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005621327,"about_ca_topic_score_gemma":0.01048504,"domain_scores_codex":[0.9637626,0.01912989,0.00279159,0.004558715,0.008658589,0.001098622],"domain_scores_gemma":[0.8920812,0.07613218,0.003824175,0.01525786,0.01084559,0.001858935],"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":[0.003288087,0.001450863,0.03972095,0.006646371,0.002205947,0.0003180552,0.002137268,0.2374054,0.01359633,0.02587029,0.1025237,0.5648368],"study_design_scores_gemma":[0.0005694729,0.001216781,0.01365688,0.001035111,0.0006068972,0.0003024963,0.0009966183,0.8379625,0.04089003,0.05275679,0.0497604,0.0002461305],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3927853,0.01668323,0.4284207,0.00715702,0.001942443,0.001372198,0.02905534,0.0698866,0.05269727],"genre_scores_gemma":[0.7200148,0.001067948,0.2299341,0.001492961,0.0002329064,0.0006638132,0.03874089,0.003021808,0.0048307],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.02455784,"threshold_uncertainty_score":0.1298758,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1780688572621074,"score_gpt":0.4662585735755815,"score_spread":0.2881897163134741,"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."}}