{"id":"https://openalex.org/W7165642620","doi":"https://doi.org/10.48550/arxiv.2606.21409","title":"Don't Blindly Trust It: How Unreliable Feedback Breaks Tool-Using LLM Agents","display_name":"Don't Blindly Trust It: How Unreliable Feedback Breaks Tool-Using LLM Agents","publication_year":2026,"publication_date":"2026-06-19","ids":{"openalex":"https://openalex.org/W7165642620","doi":"https://doi.org/10.48550/arxiv.2606.21409"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.21409","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.21409","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2606.21409","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139145150","display_name":"Chubin Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Chubin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139201428","display_name":"Zhenglin Wan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wan, Zhenglin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139217196","display_name":"Xingrui Yu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yu, Xingrui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139190781","display_name":"Pengfei Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhou, Pengfei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5081255558","display_name":"Wangbo Zhao","orcid":"https://orcid.org/0000-0001-9545-7991"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Wangbo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139182899","display_name":"Jingxuan Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Jingxuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122234540","display_name":"Yaxin Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhou, Yaxin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139155174","display_name":"Ivor Tsang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tsang, Ivor","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.2992999851703644,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.2992999851703644,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.08860000222921371,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.04450000077486038,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.6025000214576721},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.447299987077713},{"id":"https://openalex.org/keywords/value","display_name":"Value (mathematics)","score":0.41609999537467957},{"id":"https://openalex.org/keywords/action","display_name":"Action (physics)","score":0.39329999685287476},{"id":"https://openalex.org/keywords/simple","display_name":"Simple (philosophy)","score":0.3305000066757202}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6098999977111816},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.6025000214576721},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.447299987077713},{"id":"https://openalex.org/C2776291640","wikidata":"https://www.wikidata.org/wiki/Q2912517","display_name":"Value (mathematics)","level":2,"score":0.41609999537467957},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.39329999685287476},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.39100000262260437},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3314000070095062},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.3305000066757202},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.2818000018596649},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.2700999975204468},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.26669999957084656}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.21409","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.21409","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2606.21409","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.21409","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Tool-augmented":[0],"agents":[1,76],"are":[2,172],"typically":[3],"evaluated":[4],"by":[5],"their":[6],"gains":[7,14,161],"under":[8,110,116],"reliable":[9],"external":[10],"feedback.":[11],"Yet":[12],"these":[13],"leave":[15],"open":[16],"a":[17,39,73],"key":[18],"counterfactual:":[19],"when":[20,126,150],"feedback":[21,71],"is":[22,154],"unreliable,":[23],"would":[24],"the":[25,45,56,86,132,151],"agent":[26,46],"be":[27],"better":[28],"off":[29],"receiving":[30],"no":[31,103],"task":[32],"evidence?":[33],"We":[34],"study":[35],"this":[36],"question":[37,64],"with":[38,96,102],"controlled":[40],"matched-loop":[41],"comparison":[42],"that":[43,77,159,167],"fixes":[44],"loop,":[47],"prompt,":[48],"action":[49],"space,":[50],"and":[51,66,99,120,166],"decoding,":[52],"while":[53],"varying":[54],"only":[55,149],"returned":[57],"observation:":[58],"faithful,":[59],"misleading,":[60],"or":[61],"absent.":[62],"Across":[63],"answering":[65],"fact":[67],"verification,":[68],"persistent":[69],"misleading":[70],"produces":[72],"value":[74],"inversion:":[75],"benefit":[78],"from":[79],"clean":[80,97,118,128],"tools":[81],"can":[82,130,162],"perform":[83],"worse":[84],"than":[85],"matched":[87,168],"no-feedback":[88,169],"fallback.":[89],"On":[90],"HotpotQA,":[91],"Qwen2.5-7B":[92],"reaches":[93],"44.8":[94],"F1":[95,101,109],"retrieval":[98,119],"22.3":[100],"feedback,":[104],"but":[105,124],"drops":[106],"to":[107],"4.7":[108],"shuffled":[111],"retrieval.":[112],"The":[113],"inversion":[114],"persists":[115],"stronger":[117],"locally":[121],"plausible":[122],"distractors,":[123],"weakens":[125],"later":[127],"evidence":[129,147],"repair":[131],"trajectory.":[133],"Early":[134],"trajectory":[135],"signals":[136],"predict":[137],"many":[138],"failures,":[139],"yet":[140],"simple":[141],"repairs":[142],"remain":[143],"fallback-limited:":[144],"rejecting":[145],"bad":[146],"helps":[148],"exposed":[152],"fallback":[153,170],"reliable.":[155],"These":[156],"results":[157],"show":[158],"clean-tool":[160],"overstate":[163],"tool":[164],"value,":[165],"controls":[171],"necessary":[173],"for":[174],"evaluating":[175],"tool-augmented":[176],"agents.":[177]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-24T00:00:00"}