{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:DKTY3RAWZD5S3DZ5TR5IIRB6K3","short_pith_number":"pith:DKTY3RAW","schema_version":"1.0","canonical_sha256":"1aa78dc416c8fb2d8f3d9c7a84443e56de6a96b9e942b664ae996139f259072d","source":{"kind":"arxiv","id":"2012.14983","version":2},"attestation_state":"computed","paper":{"title":"Reducing conversational agents' overconfidence through linguistic calibration","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Arthur Szlam, Emily Dinan, Sabrina J. Mielke, Y-Lan Boureau","submitted_at":"2020-12-30T00:12:36Z","abstract_excerpt":"While improving neural dialogue agents' factual accuracy is the object of much research, another important aspect of communication, less studied in the setting of neural dialogue, is transparency about ignorance. In this work, we analyze to what extent state-of-the-art chit-chat models are linguistically calibrated in the sense that their verbalized expression of doubt (or confidence) matches the likelihood that the model's responses are factually incorrect (or correct). We find that these models are poorly calibrated, yet we show that likelihood of correctness can accurately be predicted. By "},"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":"2012.14983","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-12-30T00:12:36Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"ff5e109306c0fa7afefa55708ed6d5ce6b42dab8409d38beb892d998c33370d5","abstract_canon_sha256":"dcb10d40f8e4b4690f465af1919aee305bd598b78c4a9d2518b99dadb5814639"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:34:44.650086Z","signature_b64":"ahFv9J5AoV7sIe43JCcVmnPQjL8DxyU9ON+pKPOXWaWS30AiS67/og+vXM0tSnDCN/uD6Y02RtuNDXiZgmLUCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1aa78dc416c8fb2d8f3d9c7a84443e56de6a96b9e942b664ae996139f259072d","last_reissued_at":"2026-07-05T04:34:44.649597Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:34:44.649597Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Reducing conversational agents' overconfidence through linguistic calibration","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Arthur Szlam, Emily Dinan, Sabrina J. Mielke, Y-Lan Boureau","submitted_at":"2020-12-30T00:12:36Z","abstract_excerpt":"While improving neural dialogue agents' factual accuracy is the object of much research, another important aspect of communication, less studied in the setting of neural dialogue, is transparency about ignorance. In this work, we analyze to what extent state-of-the-art chit-chat models are linguistically calibrated in the sense that their verbalized expression of doubt (or confidence) matches the likelihood that the model's responses are factually incorrect (or correct). We find that these models are poorly calibrated, yet we show that likelihood of correctness can accurately be predicted. By "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2012.14983","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/2012.14983/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":"2012.14983","created_at":"2026-07-05T04:34:44.649656+00:00"},{"alias_kind":"arxiv_version","alias_value":"2012.14983v2","created_at":"2026-07-05T04:34:44.649656+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2012.14983","created_at":"2026-07-05T04:34:44.649656+00:00"},{"alias_kind":"pith_short_12","alias_value":"DKTY3RAWZD5S","created_at":"2026-07-05T04:34:44.649656+00:00"},{"alias_kind":"pith_short_16","alias_value":"DKTY3RAWZD5S3DZ5","created_at":"2026-07-05T04:34:44.649656+00:00"},{"alias_kind":"pith_short_8","alias_value":"DKTY3RAW","created_at":"2026-07-05T04:34:44.649656+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2302.09664","citing_title":"Semantic Uncertainty: Linguistic Invariances for Uncertainty Estimation in Natural Language Generation","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2207.05221","citing_title":"Language Models (Mostly) Know What They Know","ref_index":289,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DKTY3RAWZD5S3DZ5TR5IIRB6K3","json":"https://pith.science/pith/DKTY3RAWZD5S3DZ5TR5IIRB6K3.json","graph_json":"https://pith.science/api/pith-number/DKTY3RAWZD5S3DZ5TR5IIRB6K3/graph.json","events_json":"https://pith.science/api/pith-number/DKTY3RAWZD5S3DZ5TR5IIRB6K3/events.json","paper":"https://pith.science/paper/DKTY3RAW"},"agent_actions":{"view_html":"https://pith.science/pith/DKTY3RAWZD5S3DZ5TR5IIRB6K3","download_json":"https://pith.science/pith/DKTY3RAWZD5S3DZ5TR5IIRB6K3.json","view_paper":"https://pith.science/paper/DKTY3RAW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2012.14983&json=true","fetch_graph":"https://pith.science/api/pith-number/DKTY3RAWZD5S3DZ5TR5IIRB6K3/graph.json","fetch_events":"https://pith.science/api/pith-number/DKTY3RAWZD5S3DZ5TR5IIRB6K3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DKTY3RAWZD5S3DZ5TR5IIRB6K3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DKTY3RAWZD5S3DZ5TR5IIRB6K3/action/storage_attestation","attest_author":"https://pith.science/pith/DKTY3RAWZD5S3DZ5TR5IIRB6K3/action/author_attestation","sign_citation":"https://pith.science/pith/DKTY3RAWZD5S3DZ5TR5IIRB6K3/action/citation_signature","submit_replication":"https://pith.science/pith/DKTY3RAWZD5S3DZ5TR5IIRB6K3/action/replication_record"}},"created_at":"2026-07-05T04:34:44.649656+00:00","updated_at":"2026-07-05T04:34:44.649656+00:00"}