{"id":"W4406184544","doi":"10.1002/aaai.12208","title":"On the reliability of Large Language Models to misinformed and demographically informed prompts","year":2025,"lang":"en","type":"article","venue":"AI Magazine","topic":"Ethics and Social Impacts of AI","field":"Social Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto; University of Ottawa; Dalhousie University","funders":"","keywords":"Reliability (semiconductor); Computer science; Psychology; Physics; Quantum mechanics","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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00186411,0.00007598243,0.0001434967,0.00007263078,0.0003179171,0.00007156549,0.0002113746,0.0001205438,0.00004175163],"category_scores_gemma":[0.004073821,0.00005180353,0.00005097397,0.0004650149,0.0002770902,0.0002132793,0.00007520807,0.0002091093,0.000008090471],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002926698,"about_ca_system_score_gemma":0.0002959175,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0006387934,"about_ca_topic_score_gemma":0.005023142,"domain_scores_codex":[0.9990438,0.00009924496,0.0001923994,0.0001155164,0.0002988719,0.0002502049],"domain_scores_gemma":[0.9984901,0.0009005516,0.00004792914,0.00020354,0.0002585203,0.00009937128],"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":[0.00007714324,0.00007990295,0.0008015323,0.00004418,0.00001981329,8.029241e-7,0.03874185,0.00001892274,0.0001600351,0.9415925,0.01653445,0.001928877],"study_design_scores_gemma":[0.001184366,0.0003964996,0.06116337,0.0003213045,0.00005244668,1.950151e-7,0.01199651,0.0006544541,0.0003660097,0.8400051,0.08347437,0.0003853835],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7273962,0.00003351122,0.0001629278,0.09816863,0.00007142029,0.0004656431,0.00001525275,0.00003333228,0.173653],"genre_scores_gemma":[0.9836649,0.00007208563,0.0001679868,0.0141429,0.00002222246,0.00001026099,0.000001904999,0.000003547257,0.001914186],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.2562686,"threshold_uncertainty_score":0.4877039,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01959875229797699,"score_gpt":0.3544841793404421,"score_spread":0.3348854270424652,"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."}}