{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:6U74O6SBC6ADS2UNLUSCNCL3VW","short_pith_number":"pith:6U74O6SB","schema_version":"1.0","canonical_sha256":"f53fc77a411780396a8d5d2426897badbd029b2230b7ef56c835cbbd71e63491","source":{"kind":"arxiv","id":"2505.12349","version":1},"attestation_state":"computed","paper":{"title":"Wisdom from Diversity: Bias Mitigation Through Hybrid Human-LLM Crowds","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CY","cs.HC","cs.LG"],"primary_cat":"cs.CL","authors_text":"Axel Abels, Tom Lenaerts","submitted_at":"2025-05-18T10:29:24Z","abstract_excerpt":"Despite their performance, large language models (LLMs) can inadvertently perpetuate biases found in the data they are trained on. By analyzing LLM responses to bias-eliciting headlines, we find that these models often mirror human biases. To address this, we explore crowd-based strategies for mitigating bias through response aggregation. We first demonstrate that simply averaging responses from multiple LLMs, intended to leverage the \"wisdom of the crowd\", can exacerbate existing biases due to the limited diversity within LLM crowds. In contrast, we show that locally weighted aggregation meth"},"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":"2505.12349","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-05-18T10:29:24Z","cross_cats_sorted":["cs.AI","cs.CY","cs.HC","cs.LG"],"title_canon_sha256":"3997d09ff112bb96e911be7c50efb0990b0b3622b13f581ed2cae31fa699141a","abstract_canon_sha256":"4bed7148f6e63cc72f5d5dcc2f19dfbffe5ae0cf843199fdddf3e5f172e4aea1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:05:02.799968Z","signature_b64":"gbnW956Q67/gmqMUgNk+727wdnKha73tbgdEJk4qWL/+DjODoSlmSV/+y7PbFMe21WSa5m6XDXRhze3vFawdAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f53fc77a411780396a8d5d2426897badbd029b2230b7ef56c835cbbd71e63491","last_reissued_at":"2026-07-05T11:05:02.799435Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:05:02.799435Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Wisdom from Diversity: Bias Mitigation Through Hybrid Human-LLM Crowds","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CY","cs.HC","cs.LG"],"primary_cat":"cs.CL","authors_text":"Axel Abels, Tom Lenaerts","submitted_at":"2025-05-18T10:29:24Z","abstract_excerpt":"Despite their performance, large language models (LLMs) can inadvertently perpetuate biases found in the data they are trained on. By analyzing LLM responses to bias-eliciting headlines, we find that these models often mirror human biases. To address this, we explore crowd-based strategies for mitigating bias through response aggregation. We first demonstrate that simply averaging responses from multiple LLMs, intended to leverage the \"wisdom of the crowd\", can exacerbate existing biases due to the limited diversity within LLM crowds. In contrast, we show that locally weighted aggregation meth"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.12349","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/2505.12349/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":"2505.12349","created_at":"2026-07-05T11:05:02.799496+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.12349v1","created_at":"2026-07-05T11:05:02.799496+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.12349","created_at":"2026-07-05T11:05:02.799496+00:00"},{"alias_kind":"pith_short_12","alias_value":"6U74O6SBC6AD","created_at":"2026-07-05T11:05:02.799496+00:00"},{"alias_kind":"pith_short_16","alias_value":"6U74O6SBC6ADS2UN","created_at":"2026-07-05T11:05:02.799496+00:00"},{"alias_kind":"pith_short_8","alias_value":"6U74O6SB","created_at":"2026-07-05T11:05:02.799496+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2608.01704","citing_title":"Floor, Ceiling, and the Fusion Gap: How Much of Crowd Reading Attention Can Machines Predict?","ref_index":6,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6U74O6SBC6ADS2UNLUSCNCL3VW","json":"https://pith.science/pith/6U74O6SBC6ADS2UNLUSCNCL3VW.json","graph_json":"https://pith.science/api/pith-number/6U74O6SBC6ADS2UNLUSCNCL3VW/graph.json","events_json":"https://pith.science/api/pith-number/6U74O6SBC6ADS2UNLUSCNCL3VW/events.json","paper":"https://pith.science/paper/6U74O6SB"},"agent_actions":{"view_html":"https://pith.science/pith/6U74O6SBC6ADS2UNLUSCNCL3VW","download_json":"https://pith.science/pith/6U74O6SBC6ADS2UNLUSCNCL3VW.json","view_paper":"https://pith.science/paper/6U74O6SB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.12349&json=true","fetch_graph":"https://pith.science/api/pith-number/6U74O6SBC6ADS2UNLUSCNCL3VW/graph.json","fetch_events":"https://pith.science/api/pith-number/6U74O6SBC6ADS2UNLUSCNCL3VW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6U74O6SBC6ADS2UNLUSCNCL3VW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6U74O6SBC6ADS2UNLUSCNCL3VW/action/storage_attestation","attest_author":"https://pith.science/pith/6U74O6SBC6ADS2UNLUSCNCL3VW/action/author_attestation","sign_citation":"https://pith.science/pith/6U74O6SBC6ADS2UNLUSCNCL3VW/action/citation_signature","submit_replication":"https://pith.science/pith/6U74O6SBC6ADS2UNLUSCNCL3VW/action/replication_record"}},"created_at":"2026-07-05T11:05:02.799496+00:00","updated_at":"2026-07-05T11:05:02.799496+00:00"}