{"id":"W4410830098","doi":"10.1039/d5dd00090d","title":"Evaluating the performance and robustness of LLMs in materials science Q&A and property predictions","year":2025,"lang":"en","type":"article","venue":"Digital Discovery","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":7,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Robustness (evolution); Property (philosophy); Economics; Biology; Genetics; Epistemology","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":[],"consensus_categories":[],"category_scores_codex":[0.001712488,0.0001176305,0.0001849795,0.0001239755,0.0002694209,0.001022179,0.000376201,0.00002773681,0.00002427203],"category_scores_gemma":[0.0007902435,0.00006482578,0.00001114026,0.0004471195,0.001303002,0.002125539,0.0004832537,0.00006469931,0.000003874532],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003101052,"about_ca_system_score_gemma":0.0001843164,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00008001455,"about_ca_topic_score_gemma":0.000009397002,"domain_scores_codex":[0.9986361,0.00006837245,0.0003303239,0.0003716191,0.0003376912,0.0002559313],"domain_scores_gemma":[0.9993773,0.0001247865,0.000116706,0.0002852608,0.00006546065,0.00003052014],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","study_design_scores_codex":[0.00004592917,0.00003318772,0.01424967,0.0001496171,0.000001460503,2.471235e-7,0.0002520769,0.01120047,0.9720058,0.000661796,0.00001921851,0.001380473],"study_design_scores_gemma":[0.0008616018,0.0003588363,0.5924766,0.001056565,0.00002800468,0.00004226761,0.0005154489,0.07697054,0.3261802,0.0008259906,0.0002076833,0.0004763053],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9971999,0.00004283527,0.0005634119,0.000158127,0.0004620317,0.0002680803,0.00004284996,0.00002469637,0.001238046],"genre_scores_gemma":[0.9984176,0.00001508299,0.0005691135,0.00002617076,0.0000216205,0.00002874153,0.000002591867,0.000005326105,0.0009137593],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6458257,"threshold_uncertainty_score":0.9856898,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02812550467023981,"score_gpt":0.3167299339162702,"score_spread":0.2886044292460304,"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."}}