{"id":"W3156545714","doi":"10.1177/10597123221095880","title":"What’s a good prediction? Challenges in evaluating an agent’s knowledge","year":2022,"lang":"en","type":"article","venue":"Adaptive Behavior","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada; Alberta Innovates; University of Alberta; Canada Research Chairs; DeepMind; Alberta Machine Intelligence Institute; Canadian Institute for Advanced Research","keywords":"Computer science; Artificial intelligence; Knowledge management; Cognitive science; Machine learning; Psychology","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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00100485,0.000164892,0.0001646602,0.0002176783,0.0002969275,0.0001454133,0.0009173243,0.00004261714,0.0001099286],"category_scores_gemma":[0.0000283433,0.0001885825,0.00005531312,0.0003962307,0.00003216326,0.001425309,0.0007776383,0.0004144853,0.00004413104],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003293278,"about_ca_system_score_gemma":0.0001436574,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0000214042,"about_ca_topic_score_gemma":0.00003884419,"domain_scores_codex":[0.9977512,0.0004685158,0.0003222412,0.0005492193,0.0005592231,0.000349639],"domain_scores_gemma":[0.9989809,0.00008672394,0.0001336449,0.0006186867,0.00008771716,0.00009230536],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00003899776,0.00162498,0.01481255,0.00001985779,0.00003886432,0.0001618272,0.04502352,0.5101568,0.001003715,0.04992398,0.0001744164,0.3770205],"study_design_scores_gemma":[0.001107199,0.003082147,0.1765205,0.00004997241,0.00004223873,0.00004732022,0.009659877,0.8057382,0.0001005069,0.0001933353,0.003027622,0.0004310596],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6806553,0.00869062,0.2719027,0.001606501,0.01507471,0.00442031,0.00002797382,0.00173754,0.01588433],"genre_scores_gemma":[0.9908423,0.00009975864,0.006791405,0.00004763075,0.00006395802,0.0006733878,0.00001070616,0.00002210943,0.001448711],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3765894,"threshold_uncertainty_score":0.7690172,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1911370875658001,"score_gpt":0.3612687422696316,"score_spread":0.1701316547038315,"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."}}