{"id":"W4413417613","doi":"10.70777/si.v2i5.15569","title":"Superintelligent Agents Pose Catastrophic Risks: Can Scientist AI Offer a Safer Path?","year":2025,"lang":"en","type":"article","venue":"SuperIntelligence - Robotics - Safety & Alignment","topic":"Space Science and Extraterrestrial Life","field":"Physics and Astronomy","cited_by":11,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University; Mila - Quebec Artificial Intelligence Institute","funders":"","keywords":"SAFER; Path (computing); Computer science; Artificial intelligence; Risk analysis (engineering); Computer security; Medicine","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","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0006294249,0.0006749246,0.0006342855,0.0002823609,0.0007349091,0.000423936,0.001164023,0.0001296149,0.002451066],"category_scores_gemma":[0.00003068806,0.0006322396,0.0004060161,0.001017204,0.0005106073,0.00031999,0.0005097779,0.0004625345,0.000595393],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003780815,"about_ca_system_score_gemma":0.0005758604,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002638886,"about_ca_topic_score_gemma":0.0001050039,"domain_scores_codex":[0.9952983,0.0001588989,0.001073751,0.001221065,0.0008840425,0.001363899],"domain_scores_gemma":[0.9978737,0.00008779901,0.0001392066,0.001188024,0.0001954025,0.0005158925],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000756804,0.003396174,0.2272124,0.0002181705,0.001350044,0.0001633831,0.00677656,0.3982562,0.01180731,0.2602153,0.019957,0.06989064],"study_design_scores_gemma":[0.006101593,0.002168883,0.05550541,0.002585726,0.001663148,0.00005859929,0.04256151,0.07147599,0.2930809,0.06554618,0.4496382,0.009613909],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2351901,0.0006071493,0.7084872,0.01368698,0.01493147,0.003896209,0.001330609,0.0003707793,0.02149957],"genre_scores_gemma":[0.9927804,0.0001369422,0.001537198,0.0008080727,0.0004376436,0.00008408337,0.0002014807,0.00004725656,0.003966924],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7575903,"threshold_uncertainty_score":0.9996129,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02974086538017891,"score_gpt":0.3159395806666005,"score_spread":0.2861987152864216,"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."}}