{"id":"W4407793840","doi":"10.1098/rsos.241313","title":"Text understanding in GPT-4 versus humans","year":2025,"lang":"en","type":"article","venue":"Royal Society Open Science","topic":"Text Readability and Simplification","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"ca_institutions":"Mila - Quebec Artificial Intelligence Institute; McGill University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Biology; Computer science; Computational biology","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":[0.005385288,0.0006893727,0.000698682,0.001616033,0.0004749312,0.003476393,0.001099543,0.001691026,0.006093708],"category_scores_gemma":[0.03726127,0.000294748,0.0004630651,0.0008614115,0.001676512,0.005117282,0.0026762,0.001611208,0.00184182],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007947576,"about_ca_system_score_gemma":0.0005882694,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002694008,"about_ca_topic_score_gemma":0.00194021,"domain_scores_codex":[0.9949474,0.002091741,0.0003284239,0.001297643,0.0009652602,0.0003695014],"domain_scores_gemma":[0.9780926,0.0136388,0.001977883,0.00358865,0.001765575,0.0009365562],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.004752717,0.001417508,0.2111197,0.001651367,0.0008683586,0.002158288,0.08244564,0.05119912,0.1293547,0.02701234,0.01966091,0.4683594],"study_design_scores_gemma":[0.0005815469,0.005882988,0.3516258,0.0005341557,0.0004616082,0.003412441,0.03188328,0.3258476,0.09314667,0.1085788,0.07733084,0.000714196],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9657004,0.0004023433,0.01352227,0.0005414544,0.00006101629,0.0001021719,0.0006463961,0.0009970721,0.01802685],"genre_scores_gemma":[0.9859578,0.0001173834,0.009542277,0.0002400358,0.00002135478,0.00006793308,0.001097677,0.0001088836,0.002846548],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.006093708,"threshold_uncertainty_score":0.02848053,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06907462797592029,"score_gpt":0.3414232356844297,"score_spread":0.2723486077085094,"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."}}