{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:VLM26PUJVHUFXI77MW3NRVTFW7","short_pith_number":"pith:VLM26PUJ","schema_version":"1.0","canonical_sha256":"aad9af3e89a9e85ba3ff65b6d8d665b7d120ba970095babfb74826957aa7d055","source":{"kind":"arxiv","id":"2508.03712","version":1},"attestation_state":"computed","paper":{"title":"How Deep Is Representational Bias in LLMs? The Cases of Caste and Religion","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Aditya Vashistha, Agrima Seth, Kalika Bali, Kentaro Toyama, Monojit Choudhary, Sunayana Sitaram","submitted_at":"2025-07-22T17:28:37Z","abstract_excerpt":"Representational bias in large language models (LLMs) has predominantly been measured through single-response interactions and has focused on Global North-centric identities like race and gender. We expand on that research by conducting a systematic audit of GPT-4 Turbo to reveal how deeply encoded representational biases are and how they extend to less-explored dimensions of identity. We prompt GPT-4 Turbo to generate over 7,200 stories about significant life events (such as weddings) in India, using prompts designed to encourage diversity to varying extents. Comparing the diversity of religi"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2508.03712","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-07-22T17:28:37Z","cross_cats_sorted":[],"title_canon_sha256":"6bed0d2f82f22f66c049f9e08fb48813d49a63e11612aad7cdca777c9fe29d95","abstract_canon_sha256":"40cfad8ca77420325aa963691907504cae8471e4507bc4f00f5ad4987cfd1772"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:49:13.424976Z","signature_b64":"KDVUJy6OGX+Owe8qf2A78s46LP028MiUkJx+1WwJsS/GJBI5J6tpgnNaXegf3/nEQNqzLZEipfNYXjMdU2cmAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"aad9af3e89a9e85ba3ff65b6d8d665b7d120ba970095babfb74826957aa7d055","last_reissued_at":"2026-07-05T11:49:13.424526Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:49:13.424526Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"How Deep Is Representational Bias in LLMs? The Cases of Caste and Religion","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Aditya Vashistha, Agrima Seth, Kalika Bali, Kentaro Toyama, Monojit Choudhary, Sunayana Sitaram","submitted_at":"2025-07-22T17:28:37Z","abstract_excerpt":"Representational bias in large language models (LLMs) has predominantly been measured through single-response interactions and has focused on Global North-centric identities like race and gender. We expand on that research by conducting a systematic audit of GPT-4 Turbo to reveal how deeply encoded representational biases are and how they extend to less-explored dimensions of identity. We prompt GPT-4 Turbo to generate over 7,200 stories about significant life events (such as weddings) in India, using prompts designed to encourage diversity to varying extents. Comparing the diversity of religi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.03712","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2508.03712/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2508.03712","created_at":"2026-07-05T11:49:13.424583+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.03712v1","created_at":"2026-07-05T11:49:13.424583+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.03712","created_at":"2026-07-05T11:49:13.424583+00:00"},{"alias_kind":"pith_short_12","alias_value":"VLM26PUJVHUF","created_at":"2026-07-05T11:49:13.424583+00:00"},{"alias_kind":"pith_short_16","alias_value":"VLM26PUJVHUFXI77","created_at":"2026-07-05T11:49:13.424583+00:00"},{"alias_kind":"pith_short_8","alias_value":"VLM26PUJ","created_at":"2026-07-05T11:49:13.424583+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.07915","citing_title":"Validating LLMs in social science: Epistemic threats and emerging norms","ref_index":60,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VLM26PUJVHUFXI77MW3NRVTFW7","json":"https://pith.science/pith/VLM26PUJVHUFXI77MW3NRVTFW7.json","graph_json":"https://pith.science/api/pith-number/VLM26PUJVHUFXI77MW3NRVTFW7/graph.json","events_json":"https://pith.science/api/pith-number/VLM26PUJVHUFXI77MW3NRVTFW7/events.json","paper":"https://pith.science/paper/VLM26PUJ"},"agent_actions":{"view_html":"https://pith.science/pith/VLM26PUJVHUFXI77MW3NRVTFW7","download_json":"https://pith.science/pith/VLM26PUJVHUFXI77MW3NRVTFW7.json","view_paper":"https://pith.science/paper/VLM26PUJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.03712&json=true","fetch_graph":"https://pith.science/api/pith-number/VLM26PUJVHUFXI77MW3NRVTFW7/graph.json","fetch_events":"https://pith.science/api/pith-number/VLM26PUJVHUFXI77MW3NRVTFW7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VLM26PUJVHUFXI77MW3NRVTFW7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VLM26PUJVHUFXI77MW3NRVTFW7/action/storage_attestation","attest_author":"https://pith.science/pith/VLM26PUJVHUFXI77MW3NRVTFW7/action/author_attestation","sign_citation":"https://pith.science/pith/VLM26PUJVHUFXI77MW3NRVTFW7/action/citation_signature","submit_replication":"https://pith.science/pith/VLM26PUJVHUFXI77MW3NRVTFW7/action/replication_record"}},"created_at":"2026-07-05T11:49:13.424583+00:00","updated_at":"2026-07-05T11:49:13.424583+00:00"}