{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:S7MNK6Q36M23NMWBE5IFZQB7EQ","short_pith_number":"pith:S7MNK6Q3","schema_version":"1.0","canonical_sha256":"97d8d57a1bf335b6b2c127505cc03f24322f2bcfd8080069ae2dadad43b2dbb5","source":{"kind":"arxiv","id":"2306.14308","version":1},"attestation_state":"computed","paper":{"title":"Let's Do a Thought Experiment: Using Counterfactuals to Improve Moral Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ahmad Beirami, Alex Beutel, Jilin Chen, Swaroop Mishra, Xiao Ma","submitted_at":"2023-06-25T18:40:43Z","abstract_excerpt":"Language models still struggle on moral reasoning, despite their impressive performance in many other tasks. In particular, the Moral Scenarios task in MMLU (Multi-task Language Understanding) is among the worst performing tasks for many language models, including GPT-3. In this work, we propose a new prompting framework, Thought Experiments, to teach language models to do better moral reasoning using counterfactuals. Experiment results show that our framework elicits counterfactual questions and answers from the model, which in turn helps improve the accuracy on Moral Scenarios task by 9-16% "},"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":"2306.14308","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-06-25T18:40:43Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"d89f45aaa9fcfc9dab9d4066b32b1ec3b8570172dc15ae3ecc517ddfd308ff36","abstract_canon_sha256":"edceaae26e6f8921d3c7ad840a51a034d0900e3d889115158154fef0de0f31fd"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:24:34.599006Z","signature_b64":"DPB4F2yFeOYdrEysWjOogtgIQG5tiQpSTRk8vZaPVY/qRmw/5OzBdfzQgTqkSk/ak18ptrHp+CaDfdFDEnJFCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"97d8d57a1bf335b6b2c127505cc03f24322f2bcfd8080069ae2dadad43b2dbb5","last_reissued_at":"2026-07-05T06:24:34.598459Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:24:34.598459Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Let's Do a Thought Experiment: Using Counterfactuals to Improve Moral Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ahmad Beirami, Alex Beutel, Jilin Chen, Swaroop Mishra, Xiao Ma","submitted_at":"2023-06-25T18:40:43Z","abstract_excerpt":"Language models still struggle on moral reasoning, despite their impressive performance in many other tasks. In particular, the Moral Scenarios task in MMLU (Multi-task Language Understanding) is among the worst performing tasks for many language models, including GPT-3. In this work, we propose a new prompting framework, Thought Experiments, to teach language models to do better moral reasoning using counterfactuals. Experiment results show that our framework elicits counterfactual questions and answers from the model, which in turn helps improve the accuracy on Moral Scenarios task by 9-16% "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2306.14308","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/2306.14308/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":"2306.14308","created_at":"2026-07-05T06:24:34.598520+00:00"},{"alias_kind":"arxiv_version","alias_value":"2306.14308v1","created_at":"2026-07-05T06:24:34.598520+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2306.14308","created_at":"2026-07-05T06:24:34.598520+00:00"},{"alias_kind":"pith_short_12","alias_value":"S7MNK6Q36M23","created_at":"2026-07-05T06:24:34.598520+00:00"},{"alias_kind":"pith_short_16","alias_value":"S7MNK6Q36M23NMWB","created_at":"2026-07-05T06:24:34.598520+00:00"},{"alias_kind":"pith_short_8","alias_value":"S7MNK6Q3","created_at":"2026-07-05T06:24:34.598520+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.12731","citing_title":"Normative Robustness as a Frontier for Non-Verifiable Reasoning in LLMs","ref_index":105,"is_internal_anchor":false},{"citing_arxiv_id":"2309.03409","citing_title":"Large Language Models as Optimizers","ref_index":20,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/S7MNK6Q36M23NMWBE5IFZQB7EQ","json":"https://pith.science/pith/S7MNK6Q36M23NMWBE5IFZQB7EQ.json","graph_json":"https://pith.science/api/pith-number/S7MNK6Q36M23NMWBE5IFZQB7EQ/graph.json","events_json":"https://pith.science/api/pith-number/S7MNK6Q36M23NMWBE5IFZQB7EQ/events.json","paper":"https://pith.science/paper/S7MNK6Q3"},"agent_actions":{"view_html":"https://pith.science/pith/S7MNK6Q36M23NMWBE5IFZQB7EQ","download_json":"https://pith.science/pith/S7MNK6Q36M23NMWBE5IFZQB7EQ.json","view_paper":"https://pith.science/paper/S7MNK6Q3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2306.14308&json=true","fetch_graph":"https://pith.science/api/pith-number/S7MNK6Q36M23NMWBE5IFZQB7EQ/graph.json","fetch_events":"https://pith.science/api/pith-number/S7MNK6Q36M23NMWBE5IFZQB7EQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/S7MNK6Q36M23NMWBE5IFZQB7EQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/S7MNK6Q36M23NMWBE5IFZQB7EQ/action/storage_attestation","attest_author":"https://pith.science/pith/S7MNK6Q36M23NMWBE5IFZQB7EQ/action/author_attestation","sign_citation":"https://pith.science/pith/S7MNK6Q36M23NMWBE5IFZQB7EQ/action/citation_signature","submit_replication":"https://pith.science/pith/S7MNK6Q36M23NMWBE5IFZQB7EQ/action/replication_record"}},"created_at":"2026-07-05T06:24:34.598520+00:00","updated_at":"2026-07-05T06:24:34.598520+00:00"}