{"paper":{"title":"Interactive Benchmarks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Interactive benchmarks using budgeted multi-turn interaction with objective feedback assess AI reasoning more robustly than fixed tests or preference judgments.","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Baoqing Yue, Brian Fan, Hufei Yang, Jichen Feng, Mengdi Wang, Qian Sun, Yifan Zhang, Yutong Han, Zihan Zhu","submitted_at":"2026-03-05T02:18:26Z","abstract_excerpt":"Existing reasoning evaluation paradigms suffer from different limitations: fixed benchmarks are increasingly saturated and vulnerable to contamination, while preference-based evaluations rely on subjective judgments. We argue that a core aspect of intelligence is the ability to decide what information to acquire and how to use it effectively. We propose Interactive Benchmarks, a unified evaluation paradigm that assesses a model's reasoning ability through budgeted multi-turn interaction. We evaluate models under this framework in two settings: Interactive Proofs, where models interact with a j"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our results show that interactive benchmarks provide a more robust assessment of this dimension of model intelligence, revealing substantial room for improvement in interactive scenarios.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That budgeted multi-turn interaction with objective feedback accurately isolates and measures core reasoning ability without introducing new biases from the interaction protocol or judge design.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Interactive Benchmarks assess AI reasoning via budgeted multi-turn interactions in proof and game settings, offering a more robust alternative to saturated fixed benchmarks and subjective preferences.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Interactive benchmarks using budgeted multi-turn interaction with objective feedback assess AI reasoning more robustly than fixed tests or preference judgments.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"d53f749ba466aa07ccf604acdfc951e2c92db283ffd8b5bcb5d49614a28c86bf"},"source":{"id":"2603.04737","kind":"arxiv","version":4},"verdict":{"id":"2bbbcc01-c884-4d57-96b4-f154a9b89bb5","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T17:08:18.972350Z","strongest_claim":"Our results show that interactive benchmarks provide a more robust assessment of this dimension of model intelligence, revealing substantial room for improvement in interactive scenarios.","one_line_summary":"Interactive Benchmarks assess AI reasoning via budgeted multi-turn interactions in proof and game settings, offering a more robust alternative to saturated fixed benchmarks and subjective preferences.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That budgeted multi-turn interaction with objective feedback accurately isolates and measures core reasoning ability without introducing new biases from the interaction protocol or judge design.","pith_extraction_headline":"Interactive benchmarks using budgeted multi-turn interaction with objective feedback assess AI reasoning more robustly than fixed tests or preference judgments."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2603.04737/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"}