{"id":"W6891814438","doi":"10.48448/h82z-es28","title":"Clever Hans or Neural Theory of Mind? Stress Testing Social Reasoning in Large Language Models","year":2024,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of British Columbia","funders":"","keywords":"Heuristics; Adversarial system; Artificial neural network; Benchmark (surveying); Work (physics); Stress (linguistics)","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":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.002539742,0.0004617247,0.0005834897,0.001640028,0.0001465015,0.0001625254,0.001169489,0.0002951385,0.0008525585],"category_scores_gemma":[0.000889665,0.0003804793,0.00008850861,0.00317548,0.001229689,0.0003797227,0.0005699983,0.0007311695,0.0004526823],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000299942,"about_ca_system_score_gemma":0.0009270374,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0007488123,"about_ca_topic_score_gemma":0.005551905,"domain_scores_codex":[0.9960428,0.0001915076,0.0005388898,0.001024823,0.001179959,0.001022048],"domain_scores_gemma":[0.9985738,0.0002137805,0.0004488593,0.0005021573,0.000132229,0.0001291187],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001220064,0.005458456,0.009704775,0.006240922,0.0008657909,0.006137659,0.181907,0.02102358,0.09849388,0.243008,0.2193058,0.2066341],"study_design_scores_gemma":[0.00234248,0.0002427043,0.0005556523,0.006014669,0.0002839366,0.00007574658,0.0257578,0.950389,0.001546336,0.007702323,0.002729693,0.002359673],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.06366718,0.002780631,0.003191729,0.0001405059,0.001777379,0.00181528,0.006798802,0.00138864,0.9184399],"genre_scores_gemma":[0.7916327,0.000003412402,0.0140753,0.00008832421,0.0008214123,0.00002516081,0.00008595265,0.001232294,0.1920354],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.9293654,"threshold_uncertainty_score":0.9998647,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05831946151645435,"score_gpt":0.3393438508082843,"score_spread":0.28102438929183,"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."}}