{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:DHFWS5CXQYAEGZWGZMNPE2J2R3","short_pith_number":"pith:DHFWS5CX","schema_version":"1.0","canonical_sha256":"19cb69745786004366c6cb1af2693a8ef378ab11e3cdf3b415b4a6e271c6c3c1","source":{"kind":"arxiv","id":"2601.08118","version":3},"attestation_state":"computed","paper":{"title":"MirrorBench: A Benchmark to Evaluate Conversational User-Proxy Agents for Human-Likeness","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Anil Babu Ankisettipalli, Ashutosh Hathidara, Julien Yu, Sebastian Schreiber, Vaishali Senthil","submitted_at":"2026-01-13T01:16:13Z","abstract_excerpt":"Large language models (LLMs) are increasingly used as human simulators, both for evaluating conversational systems and for generating fine-tuning data. However, naive \"act-as-a-user\" prompting often yields verbose, unrealistic utterances, motivating principled evaluation of *user proxy agents*. We present **MirrorBench**, a reproducible and extensible benchmarking framework that evaluates user proxies solely on their ability to produce human-like user utterances across diverse conversational regimes, explicitly decoupled from downstream task success. **MirrorBench** combines three lexical-dive"},"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":"2601.08118","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2026-01-13T01:16:13Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"9014c3c057641df09df01b2a0929bd36dfc465a5ea972149b9abf6332759e8cd","abstract_canon_sha256":"f2359e9a587b2a53769b5eb306b64495be999cec481cb2fc55fed07378fcaf7e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:04:21.346452Z","signature_b64":"e7SFqp6aiHtbuEhFTwkZE8XgBFz5pkumgNx1WcERwCQJQLSslAzMkeE+ck2MRdJQ2nfij/z6LaqZR4h61821Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"19cb69745786004366c6cb1af2693a8ef378ab11e3cdf3b415b4a6e271c6c3c1","last_reissued_at":"2026-05-20T00:04:21.345567Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:04:21.345567Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MirrorBench: A Benchmark to Evaluate Conversational User-Proxy Agents for Human-Likeness","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Anil Babu Ankisettipalli, Ashutosh Hathidara, Julien Yu, Sebastian Schreiber, Vaishali Senthil","submitted_at":"2026-01-13T01:16:13Z","abstract_excerpt":"Large language models (LLMs) are increasingly used as human simulators, both for evaluating conversational systems and for generating fine-tuning data. However, naive \"act-as-a-user\" prompting often yields verbose, unrealistic utterances, motivating principled evaluation of *user proxy agents*. We present **MirrorBench**, a reproducible and extensible benchmarking framework that evaluates user proxies solely on their ability to produce human-like user utterances across diverse conversational regimes, explicitly decoupled from downstream task success. **MirrorBench** combines three lexical-dive"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2601.08118","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/2601.08118/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":"2601.08118","created_at":"2026-05-20T00:04:21.345729+00:00"},{"alias_kind":"arxiv_version","alias_value":"2601.08118v3","created_at":"2026-05-20T00:04:21.345729+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2601.08118","created_at":"2026-05-20T00:04:21.345729+00:00"},{"alias_kind":"pith_short_12","alias_value":"DHFWS5CXQYAE","created_at":"2026-05-20T00:04:21.345729+00:00"},{"alias_kind":"pith_short_16","alias_value":"DHFWS5CXQYAEGZWG","created_at":"2026-05-20T00:04:21.345729+00:00"},{"alias_kind":"pith_short_8","alias_value":"DHFWS5CX","created_at":"2026-05-20T00:04:21.345729+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":3,"sample":[{"citing_arxiv_id":"2605.20506","citing_title":"Reinforcing Human Behavior Simulation via Verbal Feedback","ref_index":9,"is_internal_anchor":true},{"citing_arxiv_id":"2605.18630","citing_title":"SCICONVBENCH: Benchmarking LLMs on Multi-Turn Clarification for Task Formulation in Computational Science","ref_index":31,"is_internal_anchor":true},{"citing_arxiv_id":"2604.02315","citing_title":"Beyond the Assistant Turn: User Turn Generation as a Probe of Interaction Awareness in Language Models","ref_index":7,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DHFWS5CXQYAEGZWGZMNPE2J2R3","json":"https://pith.science/pith/DHFWS5CXQYAEGZWGZMNPE2J2R3.json","graph_json":"https://pith.science/api/pith-number/DHFWS5CXQYAEGZWGZMNPE2J2R3/graph.json","events_json":"https://pith.science/api/pith-number/DHFWS5CXQYAEGZWGZMNPE2J2R3/events.json","paper":"https://pith.science/paper/DHFWS5CX"},"agent_actions":{"view_html":"https://pith.science/pith/DHFWS5CXQYAEGZWGZMNPE2J2R3","download_json":"https://pith.science/pith/DHFWS5CXQYAEGZWGZMNPE2J2R3.json","view_paper":"https://pith.science/paper/DHFWS5CX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2601.08118&json=true","fetch_graph":"https://pith.science/api/pith-number/DHFWS5CXQYAEGZWGZMNPE2J2R3/graph.json","fetch_events":"https://pith.science/api/pith-number/DHFWS5CXQYAEGZWGZMNPE2J2R3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DHFWS5CXQYAEGZWGZMNPE2J2R3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DHFWS5CXQYAEGZWGZMNPE2J2R3/action/storage_attestation","attest_author":"https://pith.science/pith/DHFWS5CXQYAEGZWGZMNPE2J2R3/action/author_attestation","sign_citation":"https://pith.science/pith/DHFWS5CXQYAEGZWGZMNPE2J2R3/action/citation_signature","submit_replication":"https://pith.science/pith/DHFWS5CXQYAEGZWGZMNPE2J2R3/action/replication_record"}},"created_at":"2026-05-20T00:04:21.345729+00:00","updated_at":"2026-05-20T00:04:21.345729+00:00"}