{"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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.08351144,0.001512645,0.001343141,0.002832906,0.001526267,0.004262725,0.00193771,0.003148966,0.002902916],"category_scores_gemma":[0.5731789,0.0008566839,0.0007437213,0.001397825,0.003474711,0.006316244,0.005120323,0.003912771,0.002061769],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00160555,"about_ca_system_score_gemma":0.001785344,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006584829,"about_ca_topic_score_gemma":0.004216863,"domain_scores_codex":[0.89294,0.08720475,0.003330659,0.008161074,0.007069316,0.001294236],"domain_scores_gemma":[0.2033786,0.7396044,0.01831776,0.02125053,0.01540305,0.00204565],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.009941157,0.002045722,0.4739398,0.003775104,0.001482052,0.001974866,0.07609916,0.08664495,0.02093897,0.0080504,0.0222111,0.2928967],"study_design_scores_gemma":[0.0004381786,0.002623945,0.1211442,0.00164943,0.0005994551,0.001598516,0.01674641,0.7965745,0.02011167,0.02593505,0.01192764,0.000651086],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9292315,0.0008300295,0.05471992,0.001764436,0.0004215324,0.0004770176,0.001130872,0.003045216,0.008379593],"genre_scores_gemma":[0.9833111,0.0001072654,0.01362917,0.0005364153,0.00007983983,0.0002055809,0.000917072,0.0002774478,0.0009361135],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.08351144,"threshold_uncertainty_score":0.441656,"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."}}