{"id":"W4400233908","doi":"10.1109/noms59830.2024.10575429","title":"Intent Assurance using LLMs guided by Intent Drift","year":2024,"lang":"en","type":"article","venue":"","topic":"Safety Systems Engineering in Autonomy","field":"Engineering","cited_by":18,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science","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":[0.0001569776,0.000228968,0.0002099895,0.0001187342,0.00002115351,0.000100453,0.0001870007,0.00008733175,0.0001553609],"category_scores_gemma":[0.00002504514,0.0002159016,0.00009627903,0.0002572106,0.00001762243,0.0001661077,0.00003914414,0.0002029868,0.0002909807],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002733201,"about_ca_system_score_gemma":0.00002223893,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0000843905,"about_ca_topic_score_gemma":0.000004413056,"domain_scores_codex":[0.9988897,0.000009849478,0.0003758237,0.0002362274,0.000164447,0.0003239226],"domain_scores_gemma":[0.9995176,0.0000607457,0.00001115577,0.0002895455,0.0000322807,0.00008871067],"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":[0.000002063283,0.00002033154,0.0001922742,0.001105555,0.0004191323,0.00005642638,0.0004762277,0.5738202,0.185987,0.004131037,0.2279442,0.005845587],"study_design_scores_gemma":[0.00008421546,0.00000677814,0.00005870565,0.0002992376,0.00001295713,0.00005211662,0.00003923747,0.7675838,0.01100584,0.00001471121,0.22058,0.0002623981],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03516388,0.006251455,0.9191148,0.0002131101,0.007425839,0.0003719997,0.0000204855,0.005021385,0.02641701],"genre_scores_gemma":[0.990627,0.00004191935,0.00668646,0.00004384067,0.0001718985,0.00003143805,0.00000808126,0.0001076692,0.002281724],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9554631,"threshold_uncertainty_score":0.8804212,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01541141093161201,"score_gpt":0.2292902315967318,"score_spread":0.2138788206651198,"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."}}