{"id":"W4415301335","doi":"10.18280/isi.300807","title":"Scalable Human Oversight for Aligned Large Language Models: A Hybrid Framework for Intent Fidelity","year":2025,"lang":"","type":"article","venue":"Ingénierie des systèmes d information","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Scalability; Fidelity; Key (lock); Scale (ratio); Scope (computer science); High fidelity","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"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":[0.009877136,0.001547731,0.001476109,0.001062796,0.0008268665,0.002249341,0.002782245,0.001832079,0.00203513],"category_scores_gemma":[0.04137694,0.0007490277,0.001023933,0.0007124898,0.002910224,0.00550168,0.006157402,0.003896312,0.000811347],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001584634,"about_ca_system_score_gemma":0.002879583,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004268256,"about_ca_topic_score_gemma":0.004975613,"domain_scores_codex":[0.9911658,0.004968065,0.0003674688,0.001904114,0.001164912,0.0004296025],"domain_scores_gemma":[0.9770444,0.01281061,0.002250511,0.005212167,0.001796723,0.0008856299],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0006913577,0.0003835477,0.007932728,0.0003029055,0.0001794173,0.0002839931,0.00124671,0.57397,0.01520506,0.0373148,0.00654018,0.3559494],"study_design_scores_gemma":[0.00002722126,0.00007526022,0.0002514049,0.00001402143,0.00001153749,0.00004653353,0.0000454181,0.9732788,0.001961759,0.02341484,0.0008552098,0.00001799807],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01599964,0.0002914414,0.9788785,0.0004156992,0.00003286094,0.00008567079,0.00008649411,0.003580948,0.0006288282],"genre_scores_gemma":[0.6318614,0.0001753713,0.364274,0.0004974991,0.0001197599,0.0001960097,0.0004936714,0.0006396135,0.001742614],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.009877136,"threshold_uncertainty_score":0.0522359,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02382805834801389,"score_gpt":0.2797846948369074,"score_spread":0.2559566364888936,"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."}}