{"id":"W4417114405","doi":"10.48550/arxiv.2506.01372","title":"AI Scientists Fail Without Strong Implementation Capability","year":2025,"lang":"en","type":"preprint","venue":"ArXiv.org","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Institute for Catastrophic Loss Reduction","keywords":"Executor; Bottleneck; Workflow; Bridge (graph theory); Position paper; Domain (mathematical analysis); Applications of artificial intelligence","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.002995401,0.0004931664,0.0006086795,0.0002127992,0.0004903385,0.0006579185,0.001676622,0.0002515473,0.004299432],"category_scores_gemma":[0.0004351164,0.0004744481,0.0001486527,0.0003082235,0.0005033614,0.0004224576,0.002443693,0.0006535016,0.0007538929],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003647901,"about_ca_system_score_gemma":0.0007441994,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00289032,"about_ca_topic_score_gemma":0.0006241959,"domain_scores_codex":[0.9951446,0.0006218308,0.0008701766,0.001720728,0.0008506574,0.0007920302],"domain_scores_gemma":[0.9972646,0.0001072668,0.0005940677,0.00154241,0.0003156844,0.0001760076],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00002361117,0.00005072945,0.7947879,0.0004030108,0.00001181251,0.000005278689,0.0007606471,0.003151271,0.1988957,0.0004376029,0.001141853,0.0003305722],"study_design_scores_gemma":[0.0007084006,0.00009029921,0.8055754,0.0003426679,0.0001122937,0.000007142557,0.0003586795,0.003010946,0.1831503,0.00187661,0.003694157,0.001073105],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9852155,0.00004777626,0.005201445,0.001077006,0.005856975,0.0008512816,0.0002897865,0.0003838709,0.001076301],"genre_scores_gemma":[0.9928046,0.000006390578,0.004764202,0.0004998095,0.0002834762,0.0001555644,0.000123451,0.00002644172,0.00133603],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01574534,"threshold_uncertainty_score":0.9997707,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03447634958603177,"score_gpt":0.3691485353916548,"score_spread":0.334672185805623,"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."}}