{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:RSMNFXCL3G3I6W25XLCC4Q4CD3","short_pith_number":"pith:RSMNFXCL","schema_version":"1.0","canonical_sha256":"8c98d2dc4bd9b68f5b5dbac42e43821ef2343fc523eb8d4f35e7c9330469cf14","source":{"kind":"arxiv","id":"2507.14406","version":1},"attestation_state":"computed","paper":{"title":"Fail Fast, or Ask: Mitigating the Deficiencies of Reasoning LLMs with Human-in-the-Loop Systems Engineering","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","cs.SY","eess.SY"],"primary_cat":"cs.AI","authors_text":"Matt Thomson, Michael J. Zellinger","submitted_at":"2025-07-18T23:25:26Z","abstract_excerpt":"State-of-the-art reasoning LLMs are powerful problem solvers, but they still occasionally make mistakes. However, adopting AI models in risk-sensitive domains often requires error rates near 0%. To address this gap, we propose collaboration between a reasoning model and a human expert who resolves queries the model cannot confidently answer. We find that quantifying the uncertainty of a reasoning model through the length of its reasoning trace yields an effective basis for deferral to a human, e.g., cutting the error rate of Qwen3 235B-A22B on difficult MATH problems from 3% to less than 1% wh"},"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":"2507.14406","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-07-18T23:25:26Z","cross_cats_sorted":["cs.LG","cs.SY","eess.SY"],"title_canon_sha256":"5bb076bafdcb47f86fe6866c91977b011e02fce87e98073cf30b664e13f1380e","abstract_canon_sha256":"e5ba3d728002088dc760a9b127274a3ded9d5ff5925cf98fa80be5433c4348f6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:40:05.723131Z","signature_b64":"1fFxg1E7F20npIacmN1Z5UofGULGp9er7ReK3Xxz0/u25DHWnuNsvirPIsfddAZejtZtgNulp9b2HejpB3etAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8c98d2dc4bd9b68f5b5dbac42e43821ef2343fc523eb8d4f35e7c9330469cf14","last_reissued_at":"2026-07-05T11:40:05.722638Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:40:05.722638Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Fail Fast, or Ask: Mitigating the Deficiencies of Reasoning LLMs with Human-in-the-Loop Systems Engineering","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","cs.SY","eess.SY"],"primary_cat":"cs.AI","authors_text":"Matt Thomson, Michael J. Zellinger","submitted_at":"2025-07-18T23:25:26Z","abstract_excerpt":"State-of-the-art reasoning LLMs are powerful problem solvers, but they still occasionally make mistakes. However, adopting AI models in risk-sensitive domains often requires error rates near 0%. To address this gap, we propose collaboration between a reasoning model and a human expert who resolves queries the model cannot confidently answer. We find that quantifying the uncertainty of a reasoning model through the length of its reasoning trace yields an effective basis for deferral to a human, e.g., cutting the error rate of Qwen3 235B-A22B on difficult MATH problems from 3% to less than 1% wh"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.14406","kind":"arxiv","version":1},"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/2507.14406/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":"2507.14406","created_at":"2026-07-05T11:40:05.722697+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.14406v1","created_at":"2026-07-05T11:40:05.722697+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.14406","created_at":"2026-07-05T11:40:05.722697+00:00"},{"alias_kind":"pith_short_12","alias_value":"RSMNFXCL3G3I","created_at":"2026-07-05T11:40:05.722697+00:00"},{"alias_kind":"pith_short_16","alias_value":"RSMNFXCL3G3I6W25","created_at":"2026-07-05T11:40:05.722697+00:00"},{"alias_kind":"pith_short_8","alias_value":"RSMNFXCL","created_at":"2026-07-05T11:40:05.722697+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RSMNFXCL3G3I6W25XLCC4Q4CD3","json":"https://pith.science/pith/RSMNFXCL3G3I6W25XLCC4Q4CD3.json","graph_json":"https://pith.science/api/pith-number/RSMNFXCL3G3I6W25XLCC4Q4CD3/graph.json","events_json":"https://pith.science/api/pith-number/RSMNFXCL3G3I6W25XLCC4Q4CD3/events.json","paper":"https://pith.science/paper/RSMNFXCL"},"agent_actions":{"view_html":"https://pith.science/pith/RSMNFXCL3G3I6W25XLCC4Q4CD3","download_json":"https://pith.science/pith/RSMNFXCL3G3I6W25XLCC4Q4CD3.json","view_paper":"https://pith.science/paper/RSMNFXCL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.14406&json=true","fetch_graph":"https://pith.science/api/pith-number/RSMNFXCL3G3I6W25XLCC4Q4CD3/graph.json","fetch_events":"https://pith.science/api/pith-number/RSMNFXCL3G3I6W25XLCC4Q4CD3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RSMNFXCL3G3I6W25XLCC4Q4CD3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RSMNFXCL3G3I6W25XLCC4Q4CD3/action/storage_attestation","attest_author":"https://pith.science/pith/RSMNFXCL3G3I6W25XLCC4Q4CD3/action/author_attestation","sign_citation":"https://pith.science/pith/RSMNFXCL3G3I6W25XLCC4Q4CD3/action/citation_signature","submit_replication":"https://pith.science/pith/RSMNFXCL3G3I6W25XLCC4Q4CD3/action/replication_record"}},"created_at":"2026-07-05T11:40:05.722697+00:00","updated_at":"2026-07-05T11:40:05.722697+00:00"}