{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:M6V4PYRXINLWGR2PACM5A2JLHD","short_pith_number":"pith:M6V4PYRX","schema_version":"1.0","canonical_sha256":"67abc7e237435763474f0099d0692b38cb744efa56369a9a0057d4f909e5e474","source":{"kind":"arxiv","id":"2305.00050","version":3},"attestation_state":"computed","paper":{"title":"Causal Reasoning and Large Language Models: Opening a New Frontier for Causality","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.CY","cs.HC","cs.LG","stat.ME"],"primary_cat":"cs.AI","authors_text":"Amit Sharma, Chenhao Tan, Emre K{\\i}c{\\i}man, Robert Ness","submitted_at":"2023-04-28T19:00:43Z","abstract_excerpt":"The causal capabilities of large language models (LLMs) are a matter of significant debate, with critical implications for the use of LLMs in societally impactful domains such as medicine, science, law, and policy. We conduct a \"behavorial\" study of LLMs to benchmark their capability in generating causal arguments. Across a wide range of tasks, we find that LLMs can generate text corresponding to correct causal arguments with high probability, surpassing the best-performing existing methods. Algorithms based on GPT-3.5 and 4 outperform existing algorithms on a pairwise causal discovery task (9"},"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":"2305.00050","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2023-04-28T19:00:43Z","cross_cats_sorted":["cs.CL","cs.CY","cs.HC","cs.LG","stat.ME"],"title_canon_sha256":"794cf13c31ee1373029f347dbedde92f144fc15f43cc57d79a188c864bb70a04","abstract_canon_sha256":"1e77225ae1d3d53e664449939a9f255f62305736979039d3210abcba19887bd5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:56:55.600243Z","signature_b64":"WE1PEHc4bV8p+i39Xzg4xZsjM4sDXAd+xYHq5tdoqf1w0Dz0toat24jAgpA8xuluMRdAX0+mCxakE6exQOf7Dw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"67abc7e237435763474f0099d0692b38cb744efa56369a9a0057d4f909e5e474","last_reissued_at":"2026-07-05T08:56:55.599787Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:56:55.599787Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Causal Reasoning and Large Language Models: Opening a New Frontier for Causality","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.CY","cs.HC","cs.LG","stat.ME"],"primary_cat":"cs.AI","authors_text":"Amit Sharma, Chenhao Tan, Emre K{\\i}c{\\i}man, Robert Ness","submitted_at":"2023-04-28T19:00:43Z","abstract_excerpt":"The causal capabilities of large language models (LLMs) are a matter of significant debate, with critical implications for the use of LLMs in societally impactful domains such as medicine, science, law, and policy. We conduct a \"behavorial\" study of LLMs to benchmark their capability in generating causal arguments. Across a wide range of tasks, we find that LLMs can generate text corresponding to correct causal arguments with high probability, surpassing the best-performing existing methods. Algorithms based on GPT-3.5 and 4 outperform existing algorithms on a pairwise causal discovery task (9"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.00050","kind":"arxiv","version":3},"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/2305.00050/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":"2305.00050","created_at":"2026-07-05T08:56:55.599842+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.00050v3","created_at":"2026-07-05T08:56:55.599842+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.00050","created_at":"2026-07-05T08:56:55.599842+00:00"},{"alias_kind":"pith_short_12","alias_value":"M6V4PYRXINLW","created_at":"2026-07-05T08:56:55.599842+00:00"},{"alias_kind":"pith_short_16","alias_value":"M6V4PYRXINLWGR2P","created_at":"2026-07-05T08:56:55.599842+00:00"},{"alias_kind":"pith_short_8","alias_value":"M6V4PYRX","created_at":"2026-07-05T08:56:55.599842+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":23,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.24370","citing_title":"When Helpfulness Overrides Causal Caution: Context-Dependent Suppression and Recovery in LLMs","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2606.10934","citing_title":"WorldKernel: A World Model is the Coupling Kernel of Admissible Possible Worlds","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2606.04915","citing_title":"Caliper: Probing Lexical Anchors versus Causal Structure in LLMs","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27593","citing_title":"Odyssey: Constructing Verifiable Local Truth-Preserving Foundation Models","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27378","citing_title":"Formalizing Latent Thoughts: Four Axioms of Thought Representation in LLMs","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30363","citing_title":"Enhancing Regime Shift Detection Using Unstructured Data: A Study on the Treasury Market","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2606.20600","citing_title":"The New Associationism: Lessons from Deep Learning","ref_index":101,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27022","citing_title":"ORCA: An End-to-End Interactive Copilot for Optimized Root Cause Analysis","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2406.10521","citing_title":"MALLM-GAN: Multi-Agent Large Language Model as Generative Adversarial Network for Synthesizing Tabular Data","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2502.11008","citing_title":"CounterBench: Evaluating and Improving Counterfactual Reasoning in Large Language Models","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2503.19786","citing_title":"Gemma 3 Technical Report","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21928","citing_title":"CausalGuard: Conformal Inference under Graph Uncertainty","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17528","citing_title":"CasualSynth: Generating Structurally Sound Synthetic Data","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2506.16234","citing_title":"Sequential Causal Discovery with Noisy Language Model Priors","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2601.08258","citing_title":"Diagnosing and Mitigating Sycophancy and Skepticism in LLM Causal Judgment","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12835","citing_title":"PROMETHEUS: Automating Deep Causal Research Integrating Text, Data and Models","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2604.03387","citing_title":"Hume's Representational Conditions for Causal Judgment: What Bayesian Formalization Abstracted Away","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11348","citing_title":"Large Language Models for Causal Relations Extraction in Social Media: A Validation Framework for Disaster Intelligence","ref_index":61,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09168","citing_title":"CIVeX: Causal Intervention Verification for Language Agents","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2604.23054","citing_title":"DeepImagine: Learning Biomedical Reasoning via Successive Counterfactual Imagining","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01669","citing_title":"PRCD-MAP: Learning How Much to Trust Imperfect Priors in Causal Discovery","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10511","citing_title":"Thinking Fast, Thinking Wrong: Intuitiveness Modulates LLM Counterfactual Reasoning in Policy Evaluation","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.03045","citing_title":"TCD-Arena: Assessing Robustness of Time Series Causal Discovery Methods Against Assumption Violations","ref_index":266,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/M6V4PYRXINLWGR2PACM5A2JLHD","json":"https://pith.science/pith/M6V4PYRXINLWGR2PACM5A2JLHD.json","graph_json":"https://pith.science/api/pith-number/M6V4PYRXINLWGR2PACM5A2JLHD/graph.json","events_json":"https://pith.science/api/pith-number/M6V4PYRXINLWGR2PACM5A2JLHD/events.json","paper":"https://pith.science/paper/M6V4PYRX"},"agent_actions":{"view_html":"https://pith.science/pith/M6V4PYRXINLWGR2PACM5A2JLHD","download_json":"https://pith.science/pith/M6V4PYRXINLWGR2PACM5A2JLHD.json","view_paper":"https://pith.science/paper/M6V4PYRX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.00050&json=true","fetch_graph":"https://pith.science/api/pith-number/M6V4PYRXINLWGR2PACM5A2JLHD/graph.json","fetch_events":"https://pith.science/api/pith-number/M6V4PYRXINLWGR2PACM5A2JLHD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/M6V4PYRXINLWGR2PACM5A2JLHD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/M6V4PYRXINLWGR2PACM5A2JLHD/action/storage_attestation","attest_author":"https://pith.science/pith/M6V4PYRXINLWGR2PACM5A2JLHD/action/author_attestation","sign_citation":"https://pith.science/pith/M6V4PYRXINLWGR2PACM5A2JLHD/action/citation_signature","submit_replication":"https://pith.science/pith/M6V4PYRXINLWGR2PACM5A2JLHD/action/replication_record"}},"created_at":"2026-07-05T08:56:55.599842+00:00","updated_at":"2026-07-05T08:56:55.599842+00:00"}