{"paper":{"title":"What's Missing in Screen-to-Action? Towards a UI-in-the-Loop Paradigm for Multimodal GUI Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Treating GUI reasoning as a cyclic Screen-UI-Action process lets MLLMs explicitly learn element localization, semantics, and usage for more precise and interpretable decisions.","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Biao Yi, Huajun Chen, Songze Li, Tianqi Liu, Wen Zhang, Xiaoke Guo, Zhaoyan Gong, Zhiqiang Liu","submitted_at":"2026-04-08T12:12:09Z","abstract_excerpt":"Existing Graphical User Interface (GUI) reasoning tasks remain challenging, particularly in UI understanding. Current methods typically rely on direct screen-based decision-making, which lacks interpretability and overlooks a comprehensive understanding of UI elements, ultimately leading to task failure. To enhance the understanding and interaction with UIs, we propose an innovative GUI reasoning paradigm called UI-in-the-Loop (UILoop). Our approach treats the GUI reasoning task as a cyclic Screen-UI elements-Action process. By enabling Multimodal Large Language Models (MLLMs) to explicitly le"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"By enabling Multimodal Large Language Models (MLLMs) to explicitly learn the localization, semantic functions, and practical usage of key UI elements, UILoop achieves precise element discovery and performs interpretable reasoning.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That inserting an explicit UI-element learning step into a cyclic Screen-UI-Action process will produce both higher accuracy and better interpretability in MLLMs, without the added structure introducing new failure modes or requiring impractical amounts of supervision.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"UI-in-the-Loop makes multimodal models explicitly learn UI element locations, meanings, and uses in a cyclic screen-element-action loop, delivering better UI comprehension and GUI reasoning on a new 26K-sample benchmark.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Treating GUI reasoning as a cyclic Screen-UI-Action process lets MLLMs explicitly learn element localization, semantics, and usage for more precise and interpretable decisions.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"f61b6607d65f8abf67e5db36beb8920582745aafd00371cdb75d138cdfbbaf70"},"source":{"id":"2604.06995","kind":"arxiv","version":2},"verdict":{"id":"9d8eb0a4-40fe-44a5-8b4e-6af8b96256a5","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T17:40:58.055768Z","strongest_claim":"By enabling Multimodal Large Language Models (MLLMs) to explicitly learn the localization, semantic functions, and practical usage of key UI elements, UILoop achieves precise element discovery and performs interpretable reasoning.","one_line_summary":"UI-in-the-Loop makes multimodal models explicitly learn UI element locations, meanings, and uses in a cyclic screen-element-action loop, delivering better UI comprehension and GUI reasoning on a new 26K-sample benchmark.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That inserting an explicit UI-element learning step into a cyclic Screen-UI-Action process will produce both higher accuracy and better interpretability in MLLMs, without the added structure introducing new failure modes or requiring impractical amounts of supervision.","pith_extraction_headline":"Treating GUI reasoning as a cyclic Screen-UI-Action process lets MLLMs explicitly learn element localization, semantics, and usage for more precise and interpretable decisions."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.06995/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"}