{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:VKJIMS4UYN5WMF4ELXGPTIWVJR","short_pith_number":"pith:VKJIMS4U","schema_version":"1.0","canonical_sha256":"aa92864b94c37b6617845dccf9a2d54c5c62d68acef005fcd16ac1d58ec73264","source":{"kind":"arxiv","id":"2212.08061","version":2},"attestation_state":"computed","paper":{"title":"On Second Thought, Let's Not Think Step by Step! Bias and Toxicity in Zero-Shot Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Diyi Yang, Hongxin Zhang, Michael Bernstein, Omar Shaikh, William Held","submitted_at":"2022-12-15T18:59:32Z","abstract_excerpt":"Generating a Chain of Thought (CoT) has been shown to consistently improve large language model (LLM) performance on a wide range of NLP tasks. However, prior work has mainly focused on logical reasoning tasks (e.g. arithmetic, commonsense QA); it remains unclear whether improvements hold for more diverse types of reasoning, especially in socially situated contexts. Concretely, we perform a controlled evaluation of zero-shot CoT across two socially sensitive domains: harmful questions and stereotype benchmarks. We find that zero-shot CoT reasoning in sensitive domains significantly increases a"},"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":"2212.08061","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-12-15T18:59:32Z","cross_cats_sorted":[],"title_canon_sha256":"233ed21030220b7b21e571b5bffae302fd43b24357cf2bc6dbee48940a32d2f6","abstract_canon_sha256":"b58a57d5dff22541c9fd0d83abeaa3f77b2b6588bab2bac4c2cba6af5b47e31e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:17:03.378061Z","signature_b64":"ITYw1ej3tVWHrnSXFIFfvPIKfgMfFb5yayJvUGDv8JlliMWIJ5ehJjftAXUoHKJ628gBEnDXAzKTeSA1us3vCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"aa92864b94c37b6617845dccf9a2d54c5c62d68acef005fcd16ac1d58ec73264","last_reissued_at":"2026-07-05T06:17:03.377660Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:17:03.377660Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"On Second Thought, Let's Not Think Step by Step! Bias and Toxicity in Zero-Shot Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Diyi Yang, Hongxin Zhang, Michael Bernstein, Omar Shaikh, William Held","submitted_at":"2022-12-15T18:59:32Z","abstract_excerpt":"Generating a Chain of Thought (CoT) has been shown to consistently improve large language model (LLM) performance on a wide range of NLP tasks. However, prior work has mainly focused on logical reasoning tasks (e.g. arithmetic, commonsense QA); it remains unclear whether improvements hold for more diverse types of reasoning, especially in socially situated contexts. Concretely, we perform a controlled evaluation of zero-shot CoT across two socially sensitive domains: harmful questions and stereotype benchmarks. We find that zero-shot CoT reasoning in sensitive domains significantly increases a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2212.08061","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/2212.08061/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":"2212.08061","created_at":"2026-07-05T06:17:03.377717+00:00"},{"alias_kind":"arxiv_version","alias_value":"2212.08061v2","created_at":"2026-07-05T06:17:03.377717+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2212.08061","created_at":"2026-07-05T06:17:03.377717+00:00"},{"alias_kind":"pith_short_12","alias_value":"VKJIMS4UYN5W","created_at":"2026-07-05T06:17:03.377717+00:00"},{"alias_kind":"pith_short_16","alias_value":"VKJIMS4UYN5WMF4E","created_at":"2026-07-05T06:17:03.377717+00:00"},{"alias_kind":"pith_short_8","alias_value":"VKJIMS4U","created_at":"2026-07-05T06:17:03.377717+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":10,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.12234","citing_title":"On The Effectiveness-Fluency Trade-Off In LLM Conditioning: A Systematic Study","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08068","citing_title":"DICE: Entropy-Regularized Equilibrium Selection for Stable Multi-Agent LLM Coordination","ref_index":153,"is_internal_anchor":false},{"citing_arxiv_id":"2503.02574","citing_title":"LLM-Safety Evaluations Lack Robustness","ref_index":50,"is_internal_anchor":false},{"citing_arxiv_id":"2503.19444","citing_title":"AI Failures in the Eyes of the Downstream Developer: A First Look at Concerns, Practices, and Challenges","ref_index":104,"is_internal_anchor":false},{"citing_arxiv_id":"2507.07871","citing_title":"Mitigating Watermark Forgery in Generative Models via Randomized Key Selection","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2408.00724","citing_title":"Inference Scaling Laws: An Empirical Analysis of Compute-Optimal Inference for Problem-Solving with Language Models","ref_index":252,"is_internal_anchor":false},{"citing_arxiv_id":"2402.10260","citing_title":"A StrongREJECT for Empty Jailbreaks","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2307.02483","citing_title":"Jailbroken: How Does LLM Safety Training Fail?","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10176","citing_title":"When Prompts Become Payloads: A Framework for Mitigating SQL Injection Attacks in Large Language Model-Driven Applications","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10639","citing_title":"Navigating the Sea of LLM Evaluation: Investigating Bias in Toxicity Benchmarks","ref_index":25,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VKJIMS4UYN5WMF4ELXGPTIWVJR","json":"https://pith.science/pith/VKJIMS4UYN5WMF4ELXGPTIWVJR.json","graph_json":"https://pith.science/api/pith-number/VKJIMS4UYN5WMF4ELXGPTIWVJR/graph.json","events_json":"https://pith.science/api/pith-number/VKJIMS4UYN5WMF4ELXGPTIWVJR/events.json","paper":"https://pith.science/paper/VKJIMS4U"},"agent_actions":{"view_html":"https://pith.science/pith/VKJIMS4UYN5WMF4ELXGPTIWVJR","download_json":"https://pith.science/pith/VKJIMS4UYN5WMF4ELXGPTIWVJR.json","view_paper":"https://pith.science/paper/VKJIMS4U","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2212.08061&json=true","fetch_graph":"https://pith.science/api/pith-number/VKJIMS4UYN5WMF4ELXGPTIWVJR/graph.json","fetch_events":"https://pith.science/api/pith-number/VKJIMS4UYN5WMF4ELXGPTIWVJR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VKJIMS4UYN5WMF4ELXGPTIWVJR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VKJIMS4UYN5WMF4ELXGPTIWVJR/action/storage_attestation","attest_author":"https://pith.science/pith/VKJIMS4UYN5WMF4ELXGPTIWVJR/action/author_attestation","sign_citation":"https://pith.science/pith/VKJIMS4UYN5WMF4ELXGPTIWVJR/action/citation_signature","submit_replication":"https://pith.science/pith/VKJIMS4UYN5WMF4ELXGPTIWVJR/action/replication_record"}},"created_at":"2026-07-05T06:17:03.377717+00:00","updated_at":"2026-07-05T06:17:03.377717+00:00"}