{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2R4NXBFLTQXJBDGNTKOURDSRRG","short_pith_number":"pith:2R4NXBFL","schema_version":"1.0","canonical_sha256":"d478db84ab9c2e908ccd9a9d488e51898a95a30379c45d7effeab1baff92f1a7","source":{"kind":"arxiv","id":"2502.12743","version":2},"attestation_state":"computed","paper":{"title":"\"I know myself better, but not really greatly\": How Well Can LLMs Detect and Explain LLM-Generated Texts?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Jiazhou Ji, Jie Guo, Ruizhe Li, Shujun Li, Weidong Qiu, Xiaoyu Jiang, Xinru Lu, Yang Xu, Zheng Huang","submitted_at":"2025-02-18T11:00:28Z","abstract_excerpt":"Distinguishing between human- and LLM-generated texts is crucial given the risks associated with misuse of LLMs. This paper investigates detection and explanation capabilities of current LLMs across two settings: binary (human vs. LLM-generated) and ternary classification (including an ``undecided'' class). We evaluate 6 close- and open-source LLMs of varying sizes and find that self-detection (LLMs identifying their own outputs) consistently outperforms cross-detection (identifying outputs from other LLMs), though both remain suboptimal. Introducing a ternary classification framework improves"},"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":"2502.12743","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-02-18T11:00:28Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"03e237077e14fa47eb73107a3e773f2bc05ef02e0b68241b622bbed6f76fd3bb","abstract_canon_sha256":"0790288f255ea2cc080966daf89a166af2b03a6a6b2c4d1f75c6ee87a5011d26"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:26:26.257159Z","signature_b64":"DeT195JjmHEovKfNfYJeE2NWQOGvk65soROYrQMf75N57TaY6TyJl//S/JAfFYFBl8A5heWWZtCQeNMoZkdoAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d478db84ab9c2e908ccd9a9d488e51898a95a30379c45d7effeab1baff92f1a7","last_reissued_at":"2026-07-05T11:26:26.256668Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:26:26.256668Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"\"I know myself better, but not really greatly\": How Well Can LLMs Detect and Explain LLM-Generated Texts?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Jiazhou Ji, Jie Guo, Ruizhe Li, Shujun Li, Weidong Qiu, Xiaoyu Jiang, Xinru Lu, Yang Xu, Zheng Huang","submitted_at":"2025-02-18T11:00:28Z","abstract_excerpt":"Distinguishing between human- and LLM-generated texts is crucial given the risks associated with misuse of LLMs. This paper investigates detection and explanation capabilities of current LLMs across two settings: binary (human vs. LLM-generated) and ternary classification (including an ``undecided'' class). We evaluate 6 close- and open-source LLMs of varying sizes and find that self-detection (LLMs identifying their own outputs) consistently outperforms cross-detection (identifying outputs from other LLMs), though both remain suboptimal. Introducing a ternary classification framework improves"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.12743","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/2502.12743/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":"2502.12743","created_at":"2026-07-05T11:26:26.256727+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.12743v2","created_at":"2026-07-05T11:26:26.256727+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.12743","created_at":"2026-07-05T11:26:26.256727+00:00"},{"alias_kind":"pith_short_12","alias_value":"2R4NXBFLTQXJ","created_at":"2026-07-05T11:26:26.256727+00:00"},{"alias_kind":"pith_short_16","alias_value":"2R4NXBFLTQXJBDGN","created_at":"2026-07-05T11:26:26.256727+00:00"},{"alias_kind":"pith_short_8","alias_value":"2R4NXBFL","created_at":"2026-07-05T11:26:26.256727+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2506.15683","citing_title":"PhantomHunter: Detecting Unseen Privately-Tuned LLM-Generated Text via Family-Aware Learning","ref_index":19,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2R4NXBFLTQXJBDGNTKOURDSRRG","json":"https://pith.science/pith/2R4NXBFLTQXJBDGNTKOURDSRRG.json","graph_json":"https://pith.science/api/pith-number/2R4NXBFLTQXJBDGNTKOURDSRRG/graph.json","events_json":"https://pith.science/api/pith-number/2R4NXBFLTQXJBDGNTKOURDSRRG/events.json","paper":"https://pith.science/paper/2R4NXBFL"},"agent_actions":{"view_html":"https://pith.science/pith/2R4NXBFLTQXJBDGNTKOURDSRRG","download_json":"https://pith.science/pith/2R4NXBFLTQXJBDGNTKOURDSRRG.json","view_paper":"https://pith.science/paper/2R4NXBFL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.12743&json=true","fetch_graph":"https://pith.science/api/pith-number/2R4NXBFLTQXJBDGNTKOURDSRRG/graph.json","fetch_events":"https://pith.science/api/pith-number/2R4NXBFLTQXJBDGNTKOURDSRRG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2R4NXBFLTQXJBDGNTKOURDSRRG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2R4NXBFLTQXJBDGNTKOURDSRRG/action/storage_attestation","attest_author":"https://pith.science/pith/2R4NXBFLTQXJBDGNTKOURDSRRG/action/author_attestation","sign_citation":"https://pith.science/pith/2R4NXBFLTQXJBDGNTKOURDSRRG/action/citation_signature","submit_replication":"https://pith.science/pith/2R4NXBFLTQXJBDGNTKOURDSRRG/action/replication_record"}},"created_at":"2026-07-05T11:26:26.256727+00:00","updated_at":"2026-07-05T11:26:26.256727+00:00"}