{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:GWX5LQSWIQNRJJU6TGMAMCE2DK","short_pith_number":"pith:GWX5LQSW","schema_version":"1.0","canonical_sha256":"35afd5c256441b14a69e999806089a1aa987c1e6078a1c93d530b9b956376e7e","source":{"kind":"arxiv","id":"2508.05469","version":4},"attestation_state":"computed","paper":{"title":"Let's Measure Information Step-by-Step: AI-Based Evaluation Beyond Vibes","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Treating AI evaluation as mutual information estimation via prompting lets total variation distance resist adversarial attacks without ground truth.","cross_cats":["cs.IT","math.IT"],"primary_cat":"cs.LG","authors_text":"Sanmi Koyejo, Zachary Robertson","submitted_at":"2025-08-07T15:11:43Z","abstract_excerpt":"We evaluate artificial intelligence (AI) systems without ground truth by exploiting a link between strategic gaming and information loss. Building on established information theory, we analyze which mechanisms resist adversarial manipulation. This motivates mutual evaluation, where the overseer is treated as a strategic player estimating mutual information by prompting, making truthful agent reporting an optimal strategy. We show that certain f-divergences, such as total variation distance (TVD), maintain polynomial guarantees under attack, building on an established exponential barrier for es"},"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":true},"canonical_record":{"source":{"id":"2508.05469","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-08-07T15:11:43Z","cross_cats_sorted":["cs.IT","math.IT"],"title_canon_sha256":"f63bbda0467da21521db117d5cf6bd28ba7585450ee8b5740d2afb5112ae82fc","abstract_canon_sha256":"e70c7b7aeceee79bb18742f113955ba86826c00c591beba8aae5dbd05dc83b69"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-23T03:13:46.603753Z","signature_b64":"R4p9a6bKfbjz/yJzr1eItee7/D4hP6QvIb4DLiE+mg7/9ZJqd2LzcN5xwLWvObtOlNW6DGFhWR2zQkixTsp9CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"35afd5c256441b14a69e999806089a1aa987c1e6078a1c93d530b9b956376e7e","last_reissued_at":"2026-06-23T03:13:46.603272Z","signature_status":"signed_v1","first_computed_at":"2026-06-23T03:13:46.603272Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Let's Measure Information Step-by-Step: AI-Based Evaluation Beyond Vibes","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Treating AI evaluation as mutual information estimation via prompting lets total variation distance resist adversarial attacks without ground truth.","cross_cats":["cs.IT","math.IT"],"primary_cat":"cs.LG","authors_text":"Sanmi Koyejo, Zachary Robertson","submitted_at":"2025-08-07T15:11:43Z","abstract_excerpt":"We evaluate artificial intelligence (AI) systems without ground truth by exploiting a link between strategic gaming and information loss. Building on established information theory, we analyze which mechanisms resist adversarial manipulation. This motivates mutual evaluation, where the overseer is treated as a strategic player estimating mutual information by prompting, making truthful agent reporting an optimal strategy. We show that certain f-divergences, such as total variation distance (TVD), maintain polynomial guarantees under attack, building on an established exponential barrier for es"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Certain f-divergences such as total variation distance maintain polynomial guarantees under attack for mutual information estimation in AI evaluation, with TVD-MI retaining AUC 0.70-0.77 under adversarial attacks while other approaches decay toward chance.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The assumption that treating the overseer as a strategic player estimating mutual information by prompting makes truthful agent reporting an optimal strategy, which underpins the claim that the method resists adversarial manipulation without ground truth.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"The paper introduces mutual evaluation with f-divergences like total variation distance to robustly estimate information relationships in AI systems without ground truth, showing better resistance to adversarial attacks than other methods.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Treating AI evaluation as mutual information estimation via prompting lets total variation distance resist adversarial attacks without ground truth.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"3414d5498b7a321e96edf0fb29c1ca4cf0aa453f744a83c80d3ac4383065e287"},"source":{"id":"2508.05469","kind":"arxiv","version":4},"verdict":{"id":"760b1f5f-804d-4630-85e0-eab79172a198","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-18T23:41:16.558764Z","strongest_claim":"Certain f-divergences such as total variation distance maintain polynomial guarantees under attack for mutual information estimation in AI evaluation, with TVD-MI retaining AUC 0.70-0.77 under adversarial attacks while other approaches decay toward chance.","one_line_summary":"The paper introduces mutual evaluation with f-divergences like total variation distance to robustly estimate information relationships in AI systems without ground truth, showing better resistance to adversarial attacks than other methods.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The assumption that treating the overseer as a strategic player estimating mutual information by prompting makes truthful agent reporting an optimal strategy, which underpins the claim that the method resists adversarial manipulation without ground truth.","pith_extraction_headline":"Treating AI evaluation as mutual information estimation via prompting lets total variation distance resist adversarial attacks without ground truth."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2508.05469/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":2,"snapshot_sha256":"b6d8228ae0fdd687baafdead3d1d16faf3261e616b7d1c1e1d9114fda09251b3"},"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":"2508.05469","created_at":"2026-06-23T03:13:46.603329+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.05469v4","created_at":"2026-06-23T03:13:46.603329+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.05469","created_at":"2026-06-23T03:13:46.603329+00:00"},{"alias_kind":"pith_short_12","alias_value":"GWX5LQSWIQNR","created_at":"2026-06-23T03:13:46.603329+00:00"},{"alias_kind":"pith_short_16","alias_value":"GWX5LQSWIQNRJJU6","created_at":"2026-06-23T03:13:46.603329+00:00"},{"alias_kind":"pith_short_8","alias_value":"GWX5LQSW","created_at":"2026-06-23T03:13:46.603329+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2608.01423","citing_title":"Scoring Rules! Statistical and Strategic Alignment for Text Evaluation Metrics","ref_index":13,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GWX5LQSWIQNRJJU6TGMAMCE2DK","json":"https://pith.science/pith/GWX5LQSWIQNRJJU6TGMAMCE2DK.json","graph_json":"https://pith.science/api/pith-number/GWX5LQSWIQNRJJU6TGMAMCE2DK/graph.json","events_json":"https://pith.science/api/pith-number/GWX5LQSWIQNRJJU6TGMAMCE2DK/events.json","paper":"https://pith.science/paper/GWX5LQSW"},"agent_actions":{"view_html":"https://pith.science/pith/GWX5LQSWIQNRJJU6TGMAMCE2DK","download_json":"https://pith.science/pith/GWX5LQSWIQNRJJU6TGMAMCE2DK.json","view_paper":"https://pith.science/paper/GWX5LQSW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.05469&json=true","fetch_graph":"https://pith.science/api/pith-number/GWX5LQSWIQNRJJU6TGMAMCE2DK/graph.json","fetch_events":"https://pith.science/api/pith-number/GWX5LQSWIQNRJJU6TGMAMCE2DK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GWX5LQSWIQNRJJU6TGMAMCE2DK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GWX5LQSWIQNRJJU6TGMAMCE2DK/action/storage_attestation","attest_author":"https://pith.science/pith/GWX5LQSWIQNRJJU6TGMAMCE2DK/action/author_attestation","sign_citation":"https://pith.science/pith/GWX5LQSWIQNRJJU6TGMAMCE2DK/action/citation_signature","submit_replication":"https://pith.science/pith/GWX5LQSWIQNRJJU6TGMAMCE2DK/action/replication_record"}},"created_at":"2026-06-23T03:13:46.603329+00:00","updated_at":"2026-06-23T03:13:46.603329+00:00"}