{"id":"W4416036241","doi":"10.18653/v1/2025.emnlp-main.743","title":"The discordance between embedded ethics and cultural inference in large language models","year":2025,"lang":"","type":"article","venue":"","topic":"Language and cultural evolution","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Inference; Semantics (computer science); Natural language; Language model","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":[0.03630032,0.0005342514,0.001593531,0.002058083,0.002765825,0.007514073,0.003616331,0.004055283,0.01015569],"category_scores_gemma":[0.2786509,0.001084918,0.001120671,0.001815583,0.01205238,0.0191053,0.005616757,0.006124078,0.0007218111],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001863251,"about_ca_system_score_gemma":0.001546054,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00385244,"about_ca_topic_score_gemma":0.003593616,"domain_scores_codex":[0.9766887,0.01828766,0.0007289942,0.002031795,0.001778043,0.0004848587],"domain_scores_gemma":[0.6417027,0.3228909,0.006844012,0.02133484,0.005083433,0.002144147],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0001769751,0.00007460831,0.009643963,0.0001817527,0.0002528038,0.000298423,0.008446171,0.01748476,0.0004203068,0.9411331,0.00173457,0.02015254],"study_design_scores_gemma":[0.00003809889,0.00001612947,0.002365046,0.00003484165,0.000038213,0.0001218224,0.0006327437,0.04536276,0.0001885384,0.9504468,0.0007294333,0.00002568082],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6011968,0.001117268,0.3001383,0.02644264,0.0001855872,0.0001290783,0.0004389785,0.0003334584,0.07001793],"genre_scores_gemma":[0.9840031,0.0001634879,0.01392871,0.0004781365,0.0001248675,0.00006113147,0.0001094748,0.0001207592,0.001010437],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.03630032,"threshold_uncertainty_score":0.1919767,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03718999158423501,"score_gpt":0.3888402992393538,"score_spread":0.3516503076551188,"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."}}