{"id":"W3008170323","doi":"10.1162/coli_a_00380","title":"The Limitations of Stylometry for Detecting Machine-Generated Fake News","year":2020,"lang":"en","type":"preprint","venue":"Computational Linguistics","topic":"Misinformation and Its Impacts","field":"Social Sciences","cited_by":15,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Stylometry; Misinformation; Computer science; Artificial intelligence; Authorship attribution; Natural language processing; Machine learning; Computer security","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":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0005680899,0.0001300425,0.0001871174,0.00008917998,0.0008673275,0.0002204032,0.0003275859,0.0001395365,0.00000750722],"category_scores_gemma":[0.05329722,0.000115857,0.000119844,0.0002716814,0.0001487149,0.00001897999,0.0001173297,0.0002722195,0.000008482636],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000774161,"about_ca_system_score_gemma":0.001052556,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001336405,"about_ca_topic_score_gemma":0.000293646,"domain_scores_codex":[0.9985673,0.0001092018,0.0005279487,0.0001532811,0.0004529956,0.0001893166],"domain_scores_gemma":[0.9941486,0.003269622,0.0005125175,0.0001263265,0.001815709,0.0001272173],"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":[0.00004680807,0.0000466591,0.00009318096,0.0001469031,0.0001535077,7.251857e-7,0.02641992,0.291219,0.000007884362,0.6214941,0.01104962,0.04932171],"study_design_scores_gemma":[0.0003947165,0.00006016215,0.0006117737,0.00006572065,0.00008187872,2.813873e-7,0.005816284,0.4005699,0.00004925616,0.1584112,0.4336226,0.0003161525],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.008279372,0.0009793306,0.8614061,0.006623032,0.00960237,0.002392926,0.001719757,0.0003443576,0.1086527],"genre_scores_gemma":[0.9480206,0.0001774191,0.04916722,0.0004120571,0.001450832,0.000009214691,0.0005013081,0.00001774547,0.0002436051],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9397413,"threshold_uncertainty_score":0.9546773,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1728446409505819,"score_gpt":0.3757561247219955,"score_spread":0.2029114837714136,"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."}}