{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ZIEIYDADG363SGDZTH5X2XACHO","short_pith_number":"pith:ZIEIYDAD","schema_version":"1.0","canonical_sha256":"ca088c0c0336fdb9187999fb7d5c023b9a7926fc7d9e4f5e5d10003957f507e6","source":{"kind":"arxiv","id":"2509.09658","version":2},"attestation_state":"computed","paper":{"title":"Measuring Epistemic Humility in Multimodal Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bingkui Tong, Jiaer Xia, Kaiyang Zhou, Sifeng Shang","submitted_at":"2025-09-11T17:54:00Z","abstract_excerpt":"Hallucinations in multimodal large language models (MLLMs) -- where the model generates content inconsistent with the input image -- pose significant risks in real-world applications, from misinformation in visual question answering to unsafe errors in decision-making. Existing benchmarks primarily test recognition accuracy, i.e., evaluating whether models can select the correct answer among distractors. This overlooks another important capability for trustworthy AI: recognizing when none of the provided options is supported by the image and abstaining from committing to a false choice, a humi"},"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":"2509.09658","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-09-11T17:54:00Z","cross_cats_sorted":[],"title_canon_sha256":"a74ec5648640f4964281a4e4f30643f0180a109a7343062b6f2a0cb4d766f9ff","abstract_canon_sha256":"8b523736f1db80e987ba98e797d22e5075f328d9b305aeea54164d17f4a6cefb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T02:03:54.663657Z","signature_b64":"Hev2EnZYZ3lt8RbIusSylr12Ap8G2Kk+nvSF3UofLTyUVQLSnPXgiqJwOk0Ydixim5eSYWDdFnde9uVhmJ2dDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ca088c0c0336fdb9187999fb7d5c023b9a7926fc7d9e4f5e5d10003957f507e6","last_reissued_at":"2026-05-26T02:03:54.662476Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T02:03:54.662476Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Measuring Epistemic Humility in Multimodal Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bingkui Tong, Jiaer Xia, Kaiyang Zhou, Sifeng Shang","submitted_at":"2025-09-11T17:54:00Z","abstract_excerpt":"Hallucinations in multimodal large language models (MLLMs) -- where the model generates content inconsistent with the input image -- pose significant risks in real-world applications, from misinformation in visual question answering to unsafe errors in decision-making. Existing benchmarks primarily test recognition accuracy, i.e., evaluating whether models can select the correct answer among distractors. This overlooks another important capability for trustworthy AI: recognizing when none of the provided options is supported by the image and abstaining from committing to a false choice, a humi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.09658","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/2509.09658/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":"2509.09658","created_at":"2026-05-26T02:03:54.662626+00:00"},{"alias_kind":"arxiv_version","alias_value":"2509.09658v2","created_at":"2026-05-26T02:03:54.662626+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.09658","created_at":"2026-05-26T02:03:54.662626+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZIEIYDADG363","created_at":"2026-05-26T02:03:54.662626+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZIEIYDADG363SGDZ","created_at":"2026-05-26T02:03:54.662626+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZIEIYDAD","created_at":"2026-05-26T02:03:54.662626+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2604.17843","citing_title":"Learning from AVA: Early Lessons from a Curated and Trustworthy Generative AI for Policy and Development Research","ref_index":106,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZIEIYDADG363SGDZTH5X2XACHO","json":"https://pith.science/pith/ZIEIYDADG363SGDZTH5X2XACHO.json","graph_json":"https://pith.science/api/pith-number/ZIEIYDADG363SGDZTH5X2XACHO/graph.json","events_json":"https://pith.science/api/pith-number/ZIEIYDADG363SGDZTH5X2XACHO/events.json","paper":"https://pith.science/paper/ZIEIYDAD"},"agent_actions":{"view_html":"https://pith.science/pith/ZIEIYDADG363SGDZTH5X2XACHO","download_json":"https://pith.science/pith/ZIEIYDADG363SGDZTH5X2XACHO.json","view_paper":"https://pith.science/paper/ZIEIYDAD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2509.09658&json=true","fetch_graph":"https://pith.science/api/pith-number/ZIEIYDADG363SGDZTH5X2XACHO/graph.json","fetch_events":"https://pith.science/api/pith-number/ZIEIYDADG363SGDZTH5X2XACHO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZIEIYDADG363SGDZTH5X2XACHO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZIEIYDADG363SGDZTH5X2XACHO/action/storage_attestation","attest_author":"https://pith.science/pith/ZIEIYDADG363SGDZTH5X2XACHO/action/author_attestation","sign_citation":"https://pith.science/pith/ZIEIYDADG363SGDZTH5X2XACHO/action/citation_signature","submit_replication":"https://pith.science/pith/ZIEIYDADG363SGDZTH5X2XACHO/action/replication_record"}},"created_at":"2026-05-26T02:03:54.662626+00:00","updated_at":"2026-05-26T02:03:54.662626+00:00"}