{"id":"W4394647506","doi":"10.1162/tacl_a_00670","title":"Scope Ambiguities in Large Language Models","year":2024,"lang":"en","type":"preprint","venue":"Transactions of the Association for Computational Linguistics","topic":"Topic Modeling","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University; National Research Council Canada; Mila - Quebec Artificial Intelligence Institute","funders":"Fonds de Recherche du Québec-Société et Culture","keywords":"Scope (computer science); Computer science; Linguistics; Cognitive science; Psychology; Programming language; Philosophy","routes":{"ca_aff":true,"ca_fund":true,"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":[0.0006891214,0.0001648432,0.0002673357,0.0002101269,0.0001119064,0.0001275178,0.0007808955,0.0002018387,0.000002375405],"category_scores_gemma":[0.0004853776,0.0001590203,0.0002747108,0.0002517656,0.00001693702,0.00003994545,0.0001996416,0.0005301927,0.000003907081],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004270563,"about_ca_system_score_gemma":0.0004234529,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001127005,"about_ca_topic_score_gemma":0.000102703,"domain_scores_codex":[0.9983252,0.00007481411,0.000554899,0.0003342984,0.0004977028,0.0002130634],"domain_scores_gemma":[0.9981161,0.0005063651,0.0003448258,0.0003790976,0.0006267565,0.00002690401],"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":[0.000002047634,0.0000484877,0.00003005559,0.0001754659,0.00005735332,3.552575e-7,0.00122223,0.7876428,0.000001341837,0.2103863,0.0001253349,0.0003082244],"study_design_scores_gemma":[0.0001687564,0.000006501507,0.00008083854,0.0001368833,0.00003736827,2.406774e-7,0.00003763238,0.6484197,0.00003762421,0.3507294,0.0002464954,0.00009857232],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.001163186,0.0003863148,0.9907159,0.0006545199,0.003805688,0.0004511294,0.0006563903,0.0001107464,0.002056109],"genre_scores_gemma":[0.9117383,0.00001330506,0.08667691,0.00006985012,0.0002250666,0.00005485447,0.00004403695,0.00001927292,0.001158358],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9105752,"threshold_uncertainty_score":0.648466,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02432477583018319,"score_gpt":0.2884349640007543,"score_spread":0.2641101881705711,"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."}}