{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:HABXR6PXPLK6KE4BWOOUBFKUIY","short_pith_number":"pith:HABXR6PX","schema_version":"1.0","canonical_sha256":"380378f9f77ad5e51381b39d40955446329943e724d120fce1c40662bd4a887a","source":{"kind":"arxiv","id":"2509.07961","version":2},"attestation_state":"computed","paper":{"title":"Probing the Preferences of a Language Model: Integrating Verbal and Behavioral Tests of AI Welfare","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Leonard Dung, Valen Tagliabue","submitted_at":"2025-09-09T17:48:44Z","abstract_excerpt":"We develop new experimental paradigms for measuring welfare in language models. We compare verbal reports of models about their preferences with preferences expressed through behavior when navigating a virtual environment and selecting conversation topics. We also test how costs and rewards affect behavior and whether responses to an eudaimonic welfare scale - measuring states such as autonomy and purpose in life - are stable across semantically equivalent prompts. Overall, we observed a notable degree of mutual support between our measures. The reliable correlations observed between stated pr"},"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":"2509.07961","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-09-09T17:48:44Z","cross_cats_sorted":[],"title_canon_sha256":"d520540af3c9125fc5e923b889322a563f5802973d25c3c6166face3bae5b077","abstract_canon_sha256":"7b968fd928e7071a4e7770de2f5a45114adb4244a091fef4c71f705f7e9c35e6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T01:03:15.398401Z","signature_b64":"VXCQrlzhQZ7XA1/upaxk1WxhnOCGnafXQP1qkZfcrsevErDJ6NRHM4vUKeCIC5p5AkwFA0ydECI4Jd6d/7VVAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"380378f9f77ad5e51381b39d40955446329943e724d120fce1c40662bd4a887a","last_reissued_at":"2026-05-26T01:03:15.397656Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T01:03:15.397656Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Probing the Preferences of a Language Model: Integrating Verbal and Behavioral Tests of AI Welfare","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Leonard Dung, Valen Tagliabue","submitted_at":"2025-09-09T17:48:44Z","abstract_excerpt":"We develop new experimental paradigms for measuring welfare in language models. We compare verbal reports of models about their preferences with preferences expressed through behavior when navigating a virtual environment and selecting conversation topics. We also test how costs and rewards affect behavior and whether responses to an eudaimonic welfare scale - measuring states such as autonomy and purpose in life - are stable across semantically equivalent prompts. Overall, we observed a notable degree of mutual support between our measures. The reliable correlations observed between stated pr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.07961","kind":"arxiv","version":2},"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/2509.07961/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":"2509.07961","created_at":"2026-05-26T01:03:15.397766+00:00"},{"alias_kind":"arxiv_version","alias_value":"2509.07961v2","created_at":"2026-05-26T01:03:15.397766+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.07961","created_at":"2026-05-26T01:03:15.397766+00:00"},{"alias_kind":"pith_short_12","alias_value":"HABXR6PXPLK6","created_at":"2026-05-26T01:03:15.397766+00:00"},{"alias_kind":"pith_short_16","alias_value":"HABXR6PXPLK6KE4B","created_at":"2026-05-26T01:03:15.397766+00:00"},{"alias_kind":"pith_short_8","alias_value":"HABXR6PX","created_at":"2026-05-26T01:03:15.397766+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.02972","citing_title":"A Scalable Approach to Evaluating Moral Sensitivity in LLMs","ref_index":169,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HABXR6PXPLK6KE4BWOOUBFKUIY","json":"https://pith.science/pith/HABXR6PXPLK6KE4BWOOUBFKUIY.json","graph_json":"https://pith.science/api/pith-number/HABXR6PXPLK6KE4BWOOUBFKUIY/graph.json","events_json":"https://pith.science/api/pith-number/HABXR6PXPLK6KE4BWOOUBFKUIY/events.json","paper":"https://pith.science/paper/HABXR6PX"},"agent_actions":{"view_html":"https://pith.science/pith/HABXR6PXPLK6KE4BWOOUBFKUIY","download_json":"https://pith.science/pith/HABXR6PXPLK6KE4BWOOUBFKUIY.json","view_paper":"https://pith.science/paper/HABXR6PX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2509.07961&json=true","fetch_graph":"https://pith.science/api/pith-number/HABXR6PXPLK6KE4BWOOUBFKUIY/graph.json","fetch_events":"https://pith.science/api/pith-number/HABXR6PXPLK6KE4BWOOUBFKUIY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HABXR6PXPLK6KE4BWOOUBFKUIY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HABXR6PXPLK6KE4BWOOUBFKUIY/action/storage_attestation","attest_author":"https://pith.science/pith/HABXR6PXPLK6KE4BWOOUBFKUIY/action/author_attestation","sign_citation":"https://pith.science/pith/HABXR6PXPLK6KE4BWOOUBFKUIY/action/citation_signature","submit_replication":"https://pith.science/pith/HABXR6PXPLK6KE4BWOOUBFKUIY/action/replication_record"}},"created_at":"2026-05-26T01:03:15.397766+00:00","updated_at":"2026-05-26T01:03:15.397766+00:00"}