{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:TO5UXKDGBWQFAREJ5RDBOUKK5O","short_pith_number":"pith:TO5UXKDG","schema_version":"1.0","canonical_sha256":"9bbb4ba8660da0504489ec4617514aebb6ad1244fb49d6c6dc4d98b552ec60c9","source":{"kind":"arxiv","id":"2504.07114","version":2},"attestation_state":"computed","paper":{"title":"ChatBench: From Static Benchmarks to Human-AI Evaluation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CY","cs.HC"],"primary_cat":"cs.CL","authors_text":"Ashton Anderson, Jake M. Hofman, Serina Chang","submitted_at":"2025-03-22T01:21:40Z","abstract_excerpt":"With the rapid adoption of LLM-based chatbots, there is a pressing need to evaluate what humans and LLMs can achieve together. However, standard benchmarks, such as MMLU, measure LLM capabilities in isolation (i.e., \"AI-alone\"). Here, we design and conduct a user study to convert MMLU questions into user-AI conversations, by seeding the user with the question and having them carry out a conversation with the LLM to answer their question. We release ChatBench, a new dataset with AI-alone, user-alone, and user-AI data for 396 questions and two LLMs, including 144K answers and 7,336 user-AI conve"},"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":"2504.07114","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-03-22T01:21:40Z","cross_cats_sorted":["cs.AI","cs.CY","cs.HC"],"title_canon_sha256":"f518d71aa23614ea898987f042dab1f50773f63e8c3b9fc20580729dbbcd4181","abstract_canon_sha256":"df6c957c914e28f9ff83b8e08d45cb4573ee6a39eff0137153ce177c9754c981"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:52:19.795968Z","signature_b64":"ZCqhchKRP2tSpocHz2MbgP0rb49W7XdEeGbhHBapSzTd357kEvsLoXMzf/kYPiecf2rqu/yxhFxyJRQDcLupCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9bbb4ba8660da0504489ec4617514aebb6ad1244fb49d6c6dc4d98b552ec60c9","last_reissued_at":"2026-07-05T11:52:19.795445Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:52:19.795445Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ChatBench: From Static Benchmarks to Human-AI Evaluation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CY","cs.HC"],"primary_cat":"cs.CL","authors_text":"Ashton Anderson, Jake M. Hofman, Serina Chang","submitted_at":"2025-03-22T01:21:40Z","abstract_excerpt":"With the rapid adoption of LLM-based chatbots, there is a pressing need to evaluate what humans and LLMs can achieve together. However, standard benchmarks, such as MMLU, measure LLM capabilities in isolation (i.e., \"AI-alone\"). Here, we design and conduct a user study to convert MMLU questions into user-AI conversations, by seeding the user with the question and having them carry out a conversation with the LLM to answer their question. We release ChatBench, a new dataset with AI-alone, user-alone, and user-AI data for 396 questions and two LLMs, including 144K answers and 7,336 user-AI conve"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.07114","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/2504.07114/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":"2504.07114","created_at":"2026-07-05T11:52:19.795506+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.07114v2","created_at":"2026-07-05T11:52:19.795506+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.07114","created_at":"2026-07-05T11:52:19.795506+00:00"},{"alias_kind":"pith_short_12","alias_value":"TO5UXKDGBWQF","created_at":"2026-07-05T11:52:19.795506+00:00"},{"alias_kind":"pith_short_16","alias_value":"TO5UXKDGBWQFAREJ","created_at":"2026-07-05T11:52:19.795506+00:00"},{"alias_kind":"pith_short_8","alias_value":"TO5UXKDG","created_at":"2026-07-05T11:52:19.795506+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2510.18731","citing_title":"Mitigating Lost in Multi-turn Conversation via Curriculum RL with Verifiable Accuracy and Abstention Rewards","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2505.06120","citing_title":"LLMs Get Lost In Multi-Turn Conversation","ref_index":7,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TO5UXKDGBWQFAREJ5RDBOUKK5O","json":"https://pith.science/pith/TO5UXKDGBWQFAREJ5RDBOUKK5O.json","graph_json":"https://pith.science/api/pith-number/TO5UXKDGBWQFAREJ5RDBOUKK5O/graph.json","events_json":"https://pith.science/api/pith-number/TO5UXKDGBWQFAREJ5RDBOUKK5O/events.json","paper":"https://pith.science/paper/TO5UXKDG"},"agent_actions":{"view_html":"https://pith.science/pith/TO5UXKDGBWQFAREJ5RDBOUKK5O","download_json":"https://pith.science/pith/TO5UXKDGBWQFAREJ5RDBOUKK5O.json","view_paper":"https://pith.science/paper/TO5UXKDG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.07114&json=true","fetch_graph":"https://pith.science/api/pith-number/TO5UXKDGBWQFAREJ5RDBOUKK5O/graph.json","fetch_events":"https://pith.science/api/pith-number/TO5UXKDGBWQFAREJ5RDBOUKK5O/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TO5UXKDGBWQFAREJ5RDBOUKK5O/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TO5UXKDGBWQFAREJ5RDBOUKK5O/action/storage_attestation","attest_author":"https://pith.science/pith/TO5UXKDGBWQFAREJ5RDBOUKK5O/action/author_attestation","sign_citation":"https://pith.science/pith/TO5UXKDGBWQFAREJ5RDBOUKK5O/action/citation_signature","submit_replication":"https://pith.science/pith/TO5UXKDGBWQFAREJ5RDBOUKK5O/action/replication_record"}},"created_at":"2026-07-05T11:52:19.795506+00:00","updated_at":"2026-07-05T11:52:19.795506+00:00"}