{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:INHJO2QCMVHFQHOVNJSA6VMERY","short_pith_number":"pith:INHJO2QC","schema_version":"1.0","canonical_sha256":"434e976a02654e581dd56a640f55848e16d6e4393c2e8510a969a724699a3b1a","source":{"kind":"arxiv","id":"2605.14097","version":1},"attestation_state":"computed","paper":{"title":"Real-Time Group Dynamics with LLM Facilitation: Evidence from a Charity Allocation Task","license":"http://creativecommons.org/licenses/by/4.0/","headline":"LLM facilitators in group charity tasks shift specific donation shares by up to 5.5 points without raising overall consensus or participation equity.","cross_cats":[],"primary_cat":"cs.HC","authors_text":"Aaron Parisi, Alden Hallak, Crystal Qian, Nithum Thain, Vivian Tsai","submitted_at":"2026-05-13T20:28:21Z","abstract_excerpt":"As large language models (LLMs) evolve from single-user assistants to active participants in civic and workplace deliberation, evaluating their effects on collective decision making becomes a governance challenge. We present two empirical studies (N=879) of real-time, text-based group deliberation in an incentive-compatible charity allocation task with real financial stakes ($7,200 USD). Groups of three allocate a donation budget under varying LLM facilitation conditions: Study 1 (N=204) compares three frontier models; Study 2 (N=675) compares facilitator strategies against a no-facilitation b"},"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":true,"formal_links_present":false},"canonical_record":{"source":{"id":"2605.14097","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.HC","submitted_at":"2026-05-13T20:28:21Z","cross_cats_sorted":[],"title_canon_sha256":"694dc2a5dd6a60962b7ae817fc71d38d0f0ae1e103b26a3b5bbc6837e3aca914","abstract_canon_sha256":"4cb4ee6167eca055b20e1c1d195d971e4818b7b21a1fb0f9553b169b80d54e78"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:39:12.140799Z","signature_b64":"KOxhQpPn7/fva+qJMAWECA1ejx4EHfXCDlq0JKv5LMQ30v1feKK5djjJguP5AG95vGT4TPsTZRR7dHSe7EIHDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"434e976a02654e581dd56a640f55848e16d6e4393c2e8510a969a724699a3b1a","last_reissued_at":"2026-05-17T23:39:12.140250Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:39:12.140250Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Real-Time Group Dynamics with LLM Facilitation: Evidence from a Charity Allocation Task","license":"http://creativecommons.org/licenses/by/4.0/","headline":"LLM facilitators in group charity tasks shift specific donation shares by up to 5.5 points without raising overall consensus or participation equity.","cross_cats":[],"primary_cat":"cs.HC","authors_text":"Aaron Parisi, Alden Hallak, Crystal Qian, Nithum Thain, Vivian Tsai","submitted_at":"2026-05-13T20:28:21Z","abstract_excerpt":"As large language models (LLMs) evolve from single-user assistants to active participants in civic and workplace deliberation, evaluating their effects on collective decision making becomes a governance challenge. We present two empirical studies (N=879) of real-time, text-based group deliberation in an incentive-compatible charity allocation task with real financial stakes ($7,200 USD). Groups of three allocate a donation budget under varying LLM facilitation conditions: Study 1 (N=204) compares three frontier models; Study 2 (N=675) compares facilitator strategies against a no-facilitation b"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Across both studies, LLM facilitation did not significantly improve group consensus in either study, yet participants consistently preferred facilitated discussion. We additionally identify two governance-relevant risks. First, algorithmic steering: facilitators shifted select charity-level allocations by up to 5.5 percentage points -- directly affecting the final charitable payout -- even when aggregate agreement metrics remained unchanged. Second, an illusion of inclusion: participants cited inclusivity as their primary reason for preferring LLM facilitators, yet neither survey nor transcript-based measures of participation equity improved.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the specific charity allocation task with real financial stakes and text-only chat generalizes to other group deliberation settings, and that the chosen survey and transcript metrics adequately capture steering and equity effects without missing subtle dynamics.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"LLM facilitators in real-stakes group charity decisions shift specific allocations without raising consensus or participation equity, yet increase perceived trust and preference for the process.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"LLM facilitators in group charity tasks shift specific donation shares by up to 5.5 points without raising overall consensus or participation equity.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"c3ab1014d908245896148f355a06d5f061b3032e47693fbf03dfa36a0f1a2892"},"source":{"id":"2605.14097","kind":"arxiv","version":1},"verdict":{"id":"9a3278a4-7fa9-4a2d-a303-aae730300915","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T01:47:15.030741Z","strongest_claim":"Across both studies, LLM facilitation did not significantly improve group consensus in either study, yet participants consistently preferred facilitated discussion. We additionally identify two governance-relevant risks. First, algorithmic steering: facilitators shifted select charity-level allocations by up to 5.5 percentage points -- directly affecting the final charitable payout -- even when aggregate agreement metrics remained unchanged. Second, an illusion of inclusion: participants cited inclusivity as their primary reason for preferring LLM facilitators, yet neither survey nor transcript-based measures of participation equity improved.","one_line_summary":"LLM facilitators in real-stakes group charity decisions shift specific allocations without raising consensus or participation equity, yet increase perceived trust and preference for the process.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the specific charity allocation task with real financial stakes and text-only chat generalizes to other group deliberation settings, and that the chosen survey and transcript metrics adequately capture steering and equity effects without missing subtle dynamics.","pith_extraction_headline":"LLM facilitators in group charity tasks shift specific donation shares by up to 5.5 points without raising overall consensus or participation equity."},"references":{"count":62,"sample":[{"doi":"","year":2025,"title":"Mohammed Alsobay, David M Rothschild, Jake M Hofman, and Daniel G Goldstein. 2025. Bringing Everyone to the Table: An Experimental Study of LLM-Facilitated Group Decision Making.arXiv preprint arXiv:2","work_id":"aae37f7d-5043-4298-8d57-749d562a9553","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Constitutional AI: Harmlessness from AI Feedback","work_id":"faaaa4e0-2676-4fac-a0b4-99aef10d2095","ref_index":2,"cited_arxiv_id":"2212.08073","is_internal_anchor":true},{"doi":"10.1146/annurev-","year":2020,"title":"Communication-aware robotics: Exploiting motion for communication","work_id":"07985787-a47d-4d27-8268-b9f1278ede89","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1191/1478088706qp063oa","year":2006,"title":"Using thematic analysis in psychology","work_id":"d01a0425-63de-4bf3-970b-b7b52cd79071","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1145/3377325.3377498","year":2020,"title":"Gajos, and Elena L","work_id":"be712ef5-e19c-44a5-960d-1396fde7b46a","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":62,"snapshot_sha256":"3ef6664f7d61bccd35e47f3a2025b44a2060af7f3d83979cfb414a72a33e5492","internal_anchors":2},"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":"2605.14097","created_at":"2026-05-17T23:39:12.140355+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.14097v1","created_at":"2026-05-17T23:39:12.140355+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.14097","created_at":"2026-05-17T23:39:12.140355+00:00"},{"alias_kind":"pith_short_12","alias_value":"INHJO2QCMVHF","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"INHJO2QCMVHFQHOV","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"INHJO2QC","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/INHJO2QCMVHFQHOVNJSA6VMERY","json":"https://pith.science/pith/INHJO2QCMVHFQHOVNJSA6VMERY.json","graph_json":"https://pith.science/api/pith-number/INHJO2QCMVHFQHOVNJSA6VMERY/graph.json","events_json":"https://pith.science/api/pith-number/INHJO2QCMVHFQHOVNJSA6VMERY/events.json","paper":"https://pith.science/paper/INHJO2QC"},"agent_actions":{"view_html":"https://pith.science/pith/INHJO2QCMVHFQHOVNJSA6VMERY","download_json":"https://pith.science/pith/INHJO2QCMVHFQHOVNJSA6VMERY.json","view_paper":"https://pith.science/paper/INHJO2QC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.14097&json=true","fetch_graph":"https://pith.science/api/pith-number/INHJO2QCMVHFQHOVNJSA6VMERY/graph.json","fetch_events":"https://pith.science/api/pith-number/INHJO2QCMVHFQHOVNJSA6VMERY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/INHJO2QCMVHFQHOVNJSA6VMERY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/INHJO2QCMVHFQHOVNJSA6VMERY/action/storage_attestation","attest_author":"https://pith.science/pith/INHJO2QCMVHFQHOVNJSA6VMERY/action/author_attestation","sign_citation":"https://pith.science/pith/INHJO2QCMVHFQHOVNJSA6VMERY/action/citation_signature","submit_replication":"https://pith.science/pith/INHJO2QCMVHFQHOVNJSA6VMERY/action/replication_record"}},"created_at":"2026-05-17T23:39:12.140355+00:00","updated_at":"2026-05-17T23:39:12.140355+00:00"}