{"id":"W4392619070","doi":"10.1145/3643991.3644926","title":"Whodunit: Classifying Code as Human Authored or GPT-4 Generated - A case study on CodeChef problems","year":2024,"lang":"en","type":"preprint","venue":"","topic":"Software Engineering Research","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Classifier (UML); Computer science; Machine learning; Artificial intelligence; Genetic programming; Stylometry; Natural language processing; Code (set theory); Source code; Receiver operating characteristic; Programming language","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002733799,0.0008728382,0.0004850421,0.004089516,0.000982833,0.001762672,0.001218055,0.002119207,0.001994217],"category_scores_gemma":[0.02664211,0.000234661,0.0006686692,0.002499035,0.001266381,0.002538435,0.00141919,0.001781857,0.001810396],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001606873,"about_ca_system_score_gemma":0.001112249,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005247887,"about_ca_topic_score_gemma":0.01075604,"domain_scores_codex":[0.9960056,0.0009884038,0.0003739695,0.0009171686,0.001454208,0.0002606078],"domain_scores_gemma":[0.9776585,0.01337551,0.001613571,0.003524978,0.00294722,0.0008801972],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.001649994,0.001508477,0.198552,0.002120849,0.0001921502,0.003351633,0.004586858,0.035323,0.01348964,0.01063442,0.1690776,0.5595133],"study_design_scores_gemma":[0.0003229494,0.0009945531,0.1308233,0.0006205547,0.00009709062,0.00638376,0.003956012,0.6023185,0.04754247,0.02081859,0.1858826,0.0002396324],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9224656,0.002049015,0.03379813,0.00175169,0.0004676016,0.0005110744,0.01383465,0.01078321,0.01433909],"genre_scores_gemma":[0.8743815,0.0005677409,0.07536564,0.0006246192,0.0001474797,0.0003688863,0.03723767,0.001513509,0.009793051],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.005247887,"threshold_uncertainty_score":0.01445788,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1743010673399469,"score_gpt":0.4017712111245808,"score_spread":0.2274701437846338,"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."}}