{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:PMC57UQEPSU2ELLXKFPR4L2MXD","short_pith_number":"pith:PMC57UQE","schema_version":"1.0","canonical_sha256":"7b05dfd2047ca9a22d77515f1e2f4cb8e67a68a51fd0fdf193e09b4fa03da28c","source":{"kind":"arxiv","id":"2504.09762","version":4},"attestation_state":"computed","paper":{"title":"Position: Stop Anthropomorphizing Intermediate Tokens as Reasoning/Thinking Traces!","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Durgesh Kalwar, Karthik Valmeekam, Kaya Stechly, Lucas Saldyt, Siddhant Bhambri, Soumya Rani Samineni, Subbarao Kambhampati, Upasana Biswas, Vardhan Palod","submitted_at":"2025-04-14T00:03:34Z","abstract_excerpt":"Intermediate token generation (ITG), where a model produces output before the solution, has become a standard method to improve the performance of language models on reasoning tasks. These intermediate tokens have been called \\say{reasoning traces} or even \\say{thinking traces} -- implicitly anthropomorphizing the traces, and implying that these traces resemble steps a human might take when solving a challenging problem, and as such can provide an interpretable window into the operation of the model's thinking process to the end user. In this position paper, we present evidence that this anthr"},"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":"2504.09762","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-04-14T00:03:34Z","cross_cats_sorted":[],"title_canon_sha256":"d91a0ae2d81775753d412e58a099bed9fff499a558fcbefe14c7b742b4748dd1","abstract_canon_sha256":"129c9d34afe59a05748551cf6ec4bc29a1e0f0e4a5f45a049ba04d7351215e81"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-11T01:09:12.229968Z","signature_b64":"aWl5NHX033hbSr8Y6VUc2sB2/4jgyNVDDuLisJpUHXVastLGK38DbX27iDIP04HdSYqYj8eUAy1BfihBSzsnCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7b05dfd2047ca9a22d77515f1e2f4cb8e67a68a51fd0fdf193e09b4fa03da28c","last_reissued_at":"2026-06-11T01:09:12.228925Z","signature_status":"signed_v1","first_computed_at":"2026-06-11T01:09:12.228925Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Position: Stop Anthropomorphizing Intermediate Tokens as Reasoning/Thinking Traces!","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Durgesh Kalwar, Karthik Valmeekam, Kaya Stechly, Lucas Saldyt, Siddhant Bhambri, Soumya Rani Samineni, Subbarao Kambhampati, Upasana Biswas, Vardhan Palod","submitted_at":"2025-04-14T00:03:34Z","abstract_excerpt":"Intermediate token generation (ITG), where a model produces output before the solution, has become a standard method to improve the performance of language models on reasoning tasks. These intermediate tokens have been called \\say{reasoning traces} or even \\say{thinking traces} -- implicitly anthropomorphizing the traces, and implying that these traces resemble steps a human might take when solving a challenging problem, and as such can provide an interpretable window into the operation of the model's thinking process to the end user. In this position paper, we present evidence that this anthr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.09762","kind":"arxiv","version":4},"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/2504.09762/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":"2504.09762","created_at":"2026-06-11T01:09:12.229073+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.09762v4","created_at":"2026-06-11T01:09:12.229073+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.09762","created_at":"2026-06-11T01:09:12.229073+00:00"},{"alias_kind":"pith_short_12","alias_value":"PMC57UQEPSU2","created_at":"2026-06-11T01:09:12.229073+00:00"},{"alias_kind":"pith_short_16","alias_value":"PMC57UQEPSU2ELLX","created_at":"2026-06-11T01:09:12.229073+00:00"},{"alias_kind":"pith_short_8","alias_value":"PMC57UQE","created_at":"2026-06-11T01:09:12.229073+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":5,"sample":[{"citing_arxiv_id":"2507.11198","citing_title":"Temperature and Persona Shape LLM Agent Consensus With Minimal Accuracy Gains in Qualitative Coding","ref_index":32,"is_internal_anchor":true},{"citing_arxiv_id":"2510.25426","citing_title":"Implicature in Interaction: Understanding Implicature Improves Alignment in Human-LLM Interaction","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2604.06192","citing_title":"The Stepwise Informativeness Assumption: Why are Entropy Dynamics and Reasoning Correlated in LLMs?","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"2605.13290","citing_title":"What properties of reasoning supervision are associated with improved downstream model quality?","ref_index":17,"is_internal_anchor":true},{"citing_arxiv_id":"2604.15726","citing_title":"LLM Reasoning Is Latent, Not the Chain of Thought","ref_index":43,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PMC57UQEPSU2ELLXKFPR4L2MXD","json":"https://pith.science/pith/PMC57UQEPSU2ELLXKFPR4L2MXD.json","graph_json":"https://pith.science/api/pith-number/PMC57UQEPSU2ELLXKFPR4L2MXD/graph.json","events_json":"https://pith.science/api/pith-number/PMC57UQEPSU2ELLXKFPR4L2MXD/events.json","paper":"https://pith.science/paper/PMC57UQE"},"agent_actions":{"view_html":"https://pith.science/pith/PMC57UQEPSU2ELLXKFPR4L2MXD","download_json":"https://pith.science/pith/PMC57UQEPSU2ELLXKFPR4L2MXD.json","view_paper":"https://pith.science/paper/PMC57UQE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.09762&json=true","fetch_graph":"https://pith.science/api/pith-number/PMC57UQEPSU2ELLXKFPR4L2MXD/graph.json","fetch_events":"https://pith.science/api/pith-number/PMC57UQEPSU2ELLXKFPR4L2MXD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PMC57UQEPSU2ELLXKFPR4L2MXD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PMC57UQEPSU2ELLXKFPR4L2MXD/action/storage_attestation","attest_author":"https://pith.science/pith/PMC57UQEPSU2ELLXKFPR4L2MXD/action/author_attestation","sign_citation":"https://pith.science/pith/PMC57UQEPSU2ELLXKFPR4L2MXD/action/citation_signature","submit_replication":"https://pith.science/pith/PMC57UQEPSU2ELLXKFPR4L2MXD/action/replication_record"}},"created_at":"2026-06-11T01:09:12.229073+00:00","updated_at":"2026-06-11T01:09:12.229073+00:00"}