{"id":"W4283170666","doi":"10.1145/3531146.3533088","title":"Taxonomy of Risks posed by Language Models","year":2022,"lang":"en","type":"article","venue":"2022 ACM Conference on Fairness, Accountability, and Transparency","topic":"Hate Speech and Cyberbullying Detection","field":"Computer Science","cited_by":604,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Misinformation; Taxonomy (biology); Harm; Risk analysis (engineering); Futures studies; Computer science; Risk management; Data science; Management science; Knowledge management; Engineering ethics; Psychology; Business; Artificial intelligence; Computer security; Social psychology; Engineering","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":[0.0005194382,0.0002561519,0.0003589221,0.0001351476,0.000365616,0.0001199701,0.001403964,0.00007670643,0.0006060687],"category_scores_gemma":[0.0000354442,0.000260722,0.0001016003,0.000449184,0.0001334169,0.0005841546,0.0003229642,0.0004050639,0.000008658008],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006476713,"about_ca_system_score_gemma":0.0001639535,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002286603,"about_ca_topic_score_gemma":0.000251633,"domain_scores_codex":[0.9977123,0.000177794,0.0004243418,0.0007364692,0.0005938177,0.0003553242],"domain_scores_gemma":[0.9983899,0.0001400711,0.0001597186,0.001092971,0.0001069748,0.0001103367],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.001163763,0.006132419,0.04087622,0.001412772,0.0004209699,0.0001067134,0.05048202,0.002806311,0.06201307,0.2694945,0.005704102,0.5593871],"study_design_scores_gemma":[0.01898149,0.009576393,0.04135215,0.0003107178,0.000411216,0.000235059,0.04504932,0.2746852,0.06852086,0.4795615,0.05262062,0.008695458],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9055712,0.000993384,0.08024574,0.0008480967,0.0008172418,0.0009134602,0.0004819879,0.0003006416,0.009828256],"genre_scores_gemma":[0.9982824,0.0002349882,0.0009071092,0.0001027932,0.00002505478,0.0002736638,0.0000408157,0.00001328604,0.0001199434],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.5506917,"threshold_uncertainty_score":0.9999845,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07387969715411008,"score_gpt":0.2846232755073295,"score_spread":0.2107435783532194,"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."}}