{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ZNKQMDFEUEYMJHCXDWT3AX4QIN","short_pith_number":"pith:ZNKQMDFE","schema_version":"1.0","canonical_sha256":"cb55060ca4a130c49c571da7b05f90434c82a3d12f7424024524915b0c4fdac6","source":{"kind":"arxiv","id":"2504.15236","version":1},"attestation_state":"computed","paper":{"title":"Values in the Wild: Discovering and Analyzing Values in Real-World Language Model Interactions","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CY","cs.LG"],"primary_cat":"cs.CL","authors_text":"Alex Tamkin, Arushi Somani, Deep Ganguli, Esin Durmus, Jerry Hong, Kunal Handa, Michael Stern, Miles McCain, Saffron Huang, Xiuruo Zhang","submitted_at":"2025-04-21T17:13:16Z","abstract_excerpt":"AI assistants can impart value judgments that shape people's decisions and worldviews, yet little is known empirically about what values these systems rely on in practice. To address this, we develop a bottom-up, privacy-preserving method to extract the values (normative considerations stated or demonstrated in model responses) that Claude 3 and 3.5 models exhibit in hundreds of thousands of real-world interactions. We empirically discover and taxonomize 3,307 AI values and study how they vary by context. We find that Claude expresses many practical and epistemic values, and typically supports"},"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":"2504.15236","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-21T17:13:16Z","cross_cats_sorted":["cs.AI","cs.CY","cs.LG"],"title_canon_sha256":"0130d527bba35fa5bb619d0b40127b5c8a54949eb2d424129e562bf11aec08bd","abstract_canon_sha256":"6e9efe8ca3bee27ac7d19c7ec8f666b5d2c58f7aa9f362910eff18abdd5eca1e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:51:58.604849Z","signature_b64":"3Y3uRNP2mHLRsp6atMKNxtbixCE2aNk85zt98iOI9Lyx6CV0vyZfkottodGBOm/gCk5pfXdScLoaOu0Ubm9MDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cb55060ca4a130c49c571da7b05f90434c82a3d12f7424024524915b0c4fdac6","last_reissued_at":"2026-07-05T10:51:58.604360Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:51:58.604360Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Values in the Wild: Discovering and Analyzing Values in Real-World Language Model Interactions","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CY","cs.LG"],"primary_cat":"cs.CL","authors_text":"Alex Tamkin, Arushi Somani, Deep Ganguli, Esin Durmus, Jerry Hong, Kunal Handa, Michael Stern, Miles McCain, Saffron Huang, Xiuruo Zhang","submitted_at":"2025-04-21T17:13:16Z","abstract_excerpt":"AI assistants can impart value judgments that shape people's decisions and worldviews, yet little is known empirically about what values these systems rely on in practice. To address this, we develop a bottom-up, privacy-preserving method to extract the values (normative considerations stated or demonstrated in model responses) that Claude 3 and 3.5 models exhibit in hundreds of thousands of real-world interactions. We empirically discover and taxonomize 3,307 AI values and study how they vary by context. We find that Claude expresses many practical and epistemic values, and typically supports"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.15236","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/2504.15236/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":"2504.15236","created_at":"2026-07-05T10:51:58.604416+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.15236v1","created_at":"2026-07-05T10:51:58.604416+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.15236","created_at":"2026-07-05T10:51:58.604416+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZNKQMDFEUEYM","created_at":"2026-07-05T10:51:58.604416+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZNKQMDFEUEYMJHCX","created_at":"2026-07-05T10:51:58.604416+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZNKQMDFE","created_at":"2026-07-05T10:51:58.604416+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":11,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2607.07916","citing_title":"Persona Cartography: Charting Language Model Personality Traits in Weight Space","ref_index":30,"is_internal_anchor":true},{"citing_arxiv_id":"2607.07605","citing_title":"User identity conditions moral wrongness ratings in non-reasoning large language models","ref_index":29,"is_internal_anchor":true},{"citing_arxiv_id":"2606.09809","citing_title":"Evaluation Cards: An Interpretive Layer for AI Evaluation Reporting","ref_index":54,"is_internal_anchor":false},{"citing_arxiv_id":"2605.13339","citing_title":"Probing Persona-Dependent Preferences in Language Models","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2506.00079","citing_title":"Who Gets the Kidney? Human-AI Alignment, Indecision, and Moral Values","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10310","citing_title":"Positive Alignment: Artificial Intelligence for Human Flourishing","ref_index":86,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10310","citing_title":"Positive Alignment: Artificial Intelligence for Human Flourishing","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00497","citing_title":"\"What Are You Really Trying to Do?\": Co-Creating Life Goals from Everyday Computer Use","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07105","citing_title":"Theoretical Limits of Language Model Alignment","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2605.03855","citing_title":"Evaluating Generative Models as Interactive Emergent Representations of Human-Like Collaborative Behavior","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.03855","citing_title":"Evaluating Generative Models as Interactive Emergent Representations of Human-Like Collaborative Behavior","ref_index":11,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZNKQMDFEUEYMJHCXDWT3AX4QIN","json":"https://pith.science/pith/ZNKQMDFEUEYMJHCXDWT3AX4QIN.json","graph_json":"https://pith.science/api/pith-number/ZNKQMDFEUEYMJHCXDWT3AX4QIN/graph.json","events_json":"https://pith.science/api/pith-number/ZNKQMDFEUEYMJHCXDWT3AX4QIN/events.json","paper":"https://pith.science/paper/ZNKQMDFE"},"agent_actions":{"view_html":"https://pith.science/pith/ZNKQMDFEUEYMJHCXDWT3AX4QIN","download_json":"https://pith.science/pith/ZNKQMDFEUEYMJHCXDWT3AX4QIN.json","view_paper":"https://pith.science/paper/ZNKQMDFE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.15236&json=true","fetch_graph":"https://pith.science/api/pith-number/ZNKQMDFEUEYMJHCXDWT3AX4QIN/graph.json","fetch_events":"https://pith.science/api/pith-number/ZNKQMDFEUEYMJHCXDWT3AX4QIN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZNKQMDFEUEYMJHCXDWT3AX4QIN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZNKQMDFEUEYMJHCXDWT3AX4QIN/action/storage_attestation","attest_author":"https://pith.science/pith/ZNKQMDFEUEYMJHCXDWT3AX4QIN/action/author_attestation","sign_citation":"https://pith.science/pith/ZNKQMDFEUEYMJHCXDWT3AX4QIN/action/citation_signature","submit_replication":"https://pith.science/pith/ZNKQMDFEUEYMJHCXDWT3AX4QIN/action/replication_record"}},"created_at":"2026-07-05T10:51:58.604416+00:00","updated_at":"2026-07-05T10:51:58.604416+00:00"}