{"id":"W4387389810","doi":"10.48550/arxiv.2310.02457","title":"The Empty Signifier Problem: Towards Clearer Paradigms for Operationalising \"Alignment\" in Large Language Models","year":2023,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Topic Modeling","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Economic and Social Research Council; European Commission; University of Oxford; York University","keywords":"Parallels; Transparency (behavior); Through-the-lens metering; Politics; Vocabulary; Epistemology; Sociology; Computer science; Cognitive science; Linguistics; Lens (geology); Political science; Psychology; Economics; Engineering; Law; Philosophy","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.08592337,0.001955634,0.002789349,0.007316244,0.00491072,0.01889615,0.005076678,0.005256362,0.005164854],"category_scores_gemma":[0.2961133,0.001506251,0.002224106,0.009174901,0.03664698,0.06164201,0.02124844,0.01742469,0.001081315],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004251844,"about_ca_system_score_gemma":0.005901491,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003025432,"about_ca_topic_score_gemma":0.002292905,"domain_scores_codex":[0.901719,0.0758118,0.005056969,0.009178881,0.007247352,0.0009859562],"domain_scores_gemma":[0.726062,0.2103182,0.01583601,0.03749309,0.007736514,0.002554273],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00005658278,0.000019156,0.001816356,0.0002092881,0.0000543882,0.00005474163,0.007484775,0.002066411,0.0003387209,0.9692068,0.0007013294,0.01799141],"study_design_scores_gemma":[0.00001416623,0.00002325671,0.0002951045,0.0001532325,0.00002085561,0.00006860159,0.001271359,0.01194208,0.0003311186,0.9797251,0.0061203,0.00003484141],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.009662534,0.0007499312,0.9762325,0.007568828,0.0001631544,0.0001177997,0.0001694556,0.0002409282,0.005094847],"genre_scores_gemma":[0.4232033,0.0009724947,0.5697346,0.001907427,0.0005845274,0.00114267,0.0007173746,0.0006090201,0.001128541],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.08592337,"threshold_uncertainty_score":0.4544117,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1224754965076678,"score_gpt":0.2319155566154391,"score_spread":0.1094400601077713,"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."}}