{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:5AUE5C3CZ3EBOR2MVGYVW57KOA","short_pith_number":"pith:5AUE5C3C","schema_version":"1.0","canonical_sha256":"e8284e8b62cec817474ca9b15b77ea7009f22b650b1bbc7d666f4bf9ca0688b0","source":{"kind":"arxiv","id":"2107.00061","version":2},"attestation_state":"computed","paper":{"title":"All That's 'Human' Is Not Gold: Evaluating Human Evaluation of Generated Text","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Elizabeth Clark, Nikita Haduong, Noah A. Smith, Sofia Serrano, Suchin Gururangan, Tal August","submitted_at":"2021-06-30T19:00:25Z","abstract_excerpt":"Human evaluations are typically considered the gold standard in natural language generation, but as models' fluency improves, how well can evaluators detect and judge machine-generated text? We run a study assessing non-experts' ability to distinguish between human- and machine-authored text (GPT2 and GPT3) in three domains (stories, news articles, and recipes). We find that, without training, evaluators distinguished between GPT3- and human-authored text at random chance level. We explore three approaches for quickly training evaluators to better identify GPT3-authored text (detailed instruct"},"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":"2107.00061","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2021-06-30T19:00:25Z","cross_cats_sorted":[],"title_canon_sha256":"54c254f22496843352f8e983bcfa3f43ee7a88e37122e15c48bb29d9f31edad9","abstract_canon_sha256":"303634eef574df5d52601015edcc42369b342419300cddb5c33321f37ba00006"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:55:56.456583Z","signature_b64":"ow6o3p1OPROI4QB2ulgiJO4fjsuDJEl1d2XyvtHe363W6jXaaBmoOxTUQPz0garHTHhIlXRJtJFVxIJGRkvtDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e8284e8b62cec817474ca9b15b77ea7009f22b650b1bbc7d666f4bf9ca0688b0","last_reissued_at":"2026-07-05T02:55:56.456148Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:55:56.456148Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"All That's 'Human' Is Not Gold: Evaluating Human Evaluation of Generated Text","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Elizabeth Clark, Nikita Haduong, Noah A. Smith, Sofia Serrano, Suchin Gururangan, Tal August","submitted_at":"2021-06-30T19:00:25Z","abstract_excerpt":"Human evaluations are typically considered the gold standard in natural language generation, but as models' fluency improves, how well can evaluators detect and judge machine-generated text? We run a study assessing non-experts' ability to distinguish between human- and machine-authored text (GPT2 and GPT3) in three domains (stories, news articles, and recipes). We find that, without training, evaluators distinguished between GPT3- and human-authored text at random chance level. We explore three approaches for quickly training evaluators to better identify GPT3-authored text (detailed instruct"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2107.00061","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/2107.00061/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":"2107.00061","created_at":"2026-07-05T02:55:56.456203+00:00"},{"alias_kind":"arxiv_version","alias_value":"2107.00061v2","created_at":"2026-07-05T02:55:56.456203+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2107.00061","created_at":"2026-07-05T02:55:56.456203+00:00"},{"alias_kind":"pith_short_12","alias_value":"5AUE5C3CZ3EB","created_at":"2026-07-05T02:55:56.456203+00:00"},{"alias_kind":"pith_short_16","alias_value":"5AUE5C3CZ3EBOR2M","created_at":"2026-07-05T02:55:56.456203+00:00"},{"alias_kind":"pith_short_8","alias_value":"5AUE5C3C","created_at":"2026-07-05T02:55:56.456203+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2511.12468","citing_title":"Detecting LLM-Assisted Academic Dishonesty using Keystroke Dynamics","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16304","citing_title":"Results-Actionability Gap: Understanding How Practitioners Evaluate LLM Products in the Wild","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2604.11312","citing_title":"Network Effects and Agreement Drift in LLM Debates","ref_index":5,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5AUE5C3CZ3EBOR2MVGYVW57KOA","json":"https://pith.science/pith/5AUE5C3CZ3EBOR2MVGYVW57KOA.json","graph_json":"https://pith.science/api/pith-number/5AUE5C3CZ3EBOR2MVGYVW57KOA/graph.json","events_json":"https://pith.science/api/pith-number/5AUE5C3CZ3EBOR2MVGYVW57KOA/events.json","paper":"https://pith.science/paper/5AUE5C3C"},"agent_actions":{"view_html":"https://pith.science/pith/5AUE5C3CZ3EBOR2MVGYVW57KOA","download_json":"https://pith.science/pith/5AUE5C3CZ3EBOR2MVGYVW57KOA.json","view_paper":"https://pith.science/paper/5AUE5C3C","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2107.00061&json=true","fetch_graph":"https://pith.science/api/pith-number/5AUE5C3CZ3EBOR2MVGYVW57KOA/graph.json","fetch_events":"https://pith.science/api/pith-number/5AUE5C3CZ3EBOR2MVGYVW57KOA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5AUE5C3CZ3EBOR2MVGYVW57KOA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5AUE5C3CZ3EBOR2MVGYVW57KOA/action/storage_attestation","attest_author":"https://pith.science/pith/5AUE5C3CZ3EBOR2MVGYVW57KOA/action/author_attestation","sign_citation":"https://pith.science/pith/5AUE5C3CZ3EBOR2MVGYVW57KOA/action/citation_signature","submit_replication":"https://pith.science/pith/5AUE5C3CZ3EBOR2MVGYVW57KOA/action/replication_record"}},"created_at":"2026-07-05T02:55:56.456203+00:00","updated_at":"2026-07-05T02:55:56.456203+00:00"}