{"id":"W4412888277","doi":"10.18653/v1/2025.findings-acl.695","title":"When Detection Fails: The Power of Fine-Tuned Models to Generate Human-Like Social Media Text","year":2025,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"National Research Council Canada","funders":"Ministère de la Défense Nationale","keywords":"Social media; Computer science; Power (physics); Data science; World Wide Web; Physics","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0130193,0.001740482,0.001273988,0.001728454,0.00119933,0.003199364,0.001896675,0.003272279,0.00137646],"category_scores_gemma":[0.1008463,0.0006836154,0.0009630917,0.0007749917,0.002182639,0.00566299,0.002455344,0.004455468,0.001707008],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001092569,"about_ca_system_score_gemma":0.001102301,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003628989,"about_ca_topic_score_gemma":0.003580287,"domain_scores_codex":[0.9888831,0.005343806,0.0004813916,0.003227004,0.001571277,0.0004934141],"domain_scores_gemma":[0.9160057,0.0634865,0.00346441,0.01413222,0.001987295,0.0009239861],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.002631844,0.001129692,0.1721755,0.001633779,0.0008890062,0.001977646,0.005663967,0.299105,0.04742057,0.02770426,0.0503367,0.3893321],"study_design_scores_gemma":[0.00009600273,0.0002539089,0.007121521,0.0001103818,0.00005382431,0.0006577207,0.0004770857,0.9470005,0.01302857,0.02505733,0.006075814,0.00006737278],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.4475481,0.001888984,0.5182476,0.004373233,0.0004999111,0.0006013654,0.001854204,0.01737118,0.007615388],"genre_scores_gemma":[0.9305428,0.0001342919,0.06483126,0.0009314984,0.0001442603,0.0001449548,0.001581902,0.0005308382,0.001158242],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.0130193,"threshold_uncertainty_score":0.06885344,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04444604225395427,"score_gpt":0.2662851425482298,"score_spread":0.2218391002942756,"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."}}