{"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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008583246,0.0002375348,0.0002366749,0.0001549033,0.0002857898,0.0003055321,0.001629542,0.0002053552,0.000006078632],"category_scores_gemma":[0.00003407965,0.0002217548,0.0001566966,0.000333344,0.0000390569,0.0004406591,0.001516008,0.0003827971,0.00002025586],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002548271,"about_ca_system_score_gemma":0.0002611456,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003964491,"about_ca_topic_score_gemma":0.000410701,"domain_scores_codex":[0.9979998,0.0001307522,0.0002699658,0.0009754117,0.0001454666,0.0004786329],"domain_scores_gemma":[0.9985098,0.0001659302,0.0001339304,0.00100621,0.0001000852,0.00008403498],"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":[0.00001016868,0.00002751596,0.0001816591,0.00003421823,0.00002910298,0.00004830855,0.001449965,0.5948639,0.000008785352,0.4025695,0.0001617219,0.0006151369],"study_design_scores_gemma":[0.0003752858,0.00001402716,0.00009516965,0.00006165465,0.00001249954,7.541558e-7,0.0002804926,0.8414149,0.00003459935,0.1571333,0.0003463794,0.0002309835],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.06609872,0.0000647732,0.9307037,0.0005687338,0.0004266797,0.0007831227,0.00004109655,0.0001813496,0.001131798],"genre_scores_gemma":[0.9894665,0.00007392022,0.006772029,0.0001316351,0.00008785234,0.00001391866,0.00003445833,0.00002567094,0.00339399],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9239317,"threshold_uncertainty_score":0.9042898,"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."}}