{"id":"W4411619707","doi":"10.3390/ai6070138","title":"Architectural Gaps in Generative AI: Quantifying Cognitive Risks for Safety Applications","year":2025,"lang":"en","type":"article","venue":"AI","topic":"Human-Automation Interaction and Safety","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; University of Alberta","keywords":"Generative grammar; Cognition; Computer science; Risk analysis (engineering); Psychology; Artificial intelligence; Business; Neuroscience","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":"codex-gemma-dda1882f352a","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0001543974,0.00009946214,0.0001424812,0.0001949639,0.0001761339,0.00002790313,0.00008705954,0.00007388331,0.001046023],"category_scores_gemma":[0.00006412678,0.00009832547,0.00007319324,0.0002166342,0.00004788555,0.00006095234,0.00001933639,0.0002328731,0.0002839223],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005438741,"about_ca_system_score_gemma":0.00003931802,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001089442,"about_ca_topic_score_gemma":0.0004772235,"domain_scores_codex":[0.9990978,0.0001198335,0.0002986993,0.0002560034,0.00005319598,0.000174454],"domain_scores_gemma":[0.999195,0.0004445529,0.00006372601,0.0001476352,0.0001197915,0.0000292385],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.001039886,0.0005073079,0.01597848,0.00008196082,0.0004077386,0.00000516651,0.0169865,0.0004242855,0.00105874,0.6664124,0.02802526,0.2690722],"study_design_scores_gemma":[0.006807204,0.0001611783,0.4571228,0.0002268195,0.0001014803,0.00002130169,0.006966358,0.008130467,0.002356648,0.01944046,0.4980337,0.0006316086],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.05947946,0.0002093508,0.8204293,0.01192777,0.001190235,0.002131298,0.0002239946,0.0002096981,0.1041989],"genre_scores_gemma":[0.9872418,0.000004059894,0.0004508474,0.00603805,0.00009550341,0.001019163,0.0001176979,0.000009684568,0.005023165],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9277624,"threshold_uncertainty_score":0.9998671,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.08266320799307564,"score_gpt":0.4800894348919881,"score_spread":0.3974262268989125,"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."}}