{"id":"W7116210972","doi":"","title":"From Isolation to Entanglement: When Do Interpretability Methods Identify and Disentangle Known Concepts?","year":2025,"lang":"","type":"article","venue":"arXiv (Cornell University)","topic":"Explainable Artificial Intelligence (XAI)","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"International Max Planck Research School for Advanced Methods in Process and Systems Engineering; International Max Planck Research School for Environmental, Cellular and Molecular Microbiology; McGill University","keywords":"Interpretability; Disjoint sets; Independence (probability theory); Linear subspace; Affect (linguistics); Feature (linguistics); Isolation (microbiology)","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":["metaepi_narrow","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.001157224,0.0005048531,0.0005748885,0.0004914764,0.0005899325,0.0007836015,0.00182123,0.000243095,0.0009695581],"category_scores_gemma":[0.0003820082,0.0006324943,0.0002429346,0.002085852,0.0004674773,0.002106267,0.002508459,0.0003981895,0.0002730111],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007130648,"about_ca_system_score_gemma":0.0002063208,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001929436,"about_ca_topic_score_gemma":0.000439556,"domain_scores_codex":[0.9949454,0.001165739,0.0006911148,0.002268655,0.0001693684,0.0007597284],"domain_scores_gemma":[0.9962489,0.0009074727,0.0002801255,0.001708292,0.0003878527,0.0004673266],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0007254488,0.0009728313,0.08203181,0.0001937505,0.0006941627,0.0002644444,0.03721694,0.009648282,0.008869007,0.6768087,0.002380432,0.1801942],"study_design_scores_gemma":[0.00071206,0.0003032913,0.005290638,0.0004286958,0.000353382,0.00000177331,0.00769425,0.658641,0.01979227,0.2947275,0.01109367,0.000961469],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3020915,0.0004439334,0.6919024,0.0006697922,0.001478708,0.0006509634,0.00003523831,0.0001049959,0.002622466],"genre_scores_gemma":[0.9810504,0.0001283977,0.01508114,0.0006031136,0.0001027059,0.000003435327,0.000008606678,0.00001908436,0.003003104],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.6789589,"threshold_uncertainty_score":0.9999437,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06445692341868224,"score_gpt":0.2979039952585593,"score_spread":0.2334470718398771,"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."}}