{"id":"https://openalex.org/W7134814024","doi":"https://doi.org/10.48550/arxiv.2603.06604","title":"Know When You're Wrong: Aligning Confidence with Correctness for LLM Error Detection","display_name":"Know When You're Wrong: Aligning Confidence with Correctness for LLM Error Detection","publication_year":2026,"publication_date":"2026-02-18","ids":{"openalex":"https://openalex.org/W7134814024","doi":"https://doi.org/10.48550/arxiv.2603.06604"},"language":null,"primary_location":{"id":"pmh:doi:10.48550/arxiv.2603.06604","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"publisher-specific-oa","license_id":"https://openalex.org/licenses/publisher-specific-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},"type":"article","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":null,"any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5091013071","display_name":"Xie Xiaohu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiaohu, Xie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128669347","display_name":"Liu Xiaohu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiaohu, Liu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5128640056","display_name":"Yao Benjamin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Benjamin, Yao","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":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.28797257,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"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/T10028","display_name":"Topic Modeling","score":0.2867000102996826,"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/T10028","display_name":"Topic Modeling","score":0.2867000102996826,"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.14059999585151672,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.09769999980926514,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/overconfidence-effect","display_name":"Overconfidence effect","score":0.8220999836921692},{"id":"https://openalex.org/keywords/correctness","display_name":"Correctness","score":0.6682999730110168},{"id":"https://openalex.org/keywords/reliability","display_name":"Reliability (semiconductor)","score":0.5999000072479248},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.5647000074386597},{"id":"https://openalex.org/keywords/measure","display_name":"Measure (data warehouse)","score":0.5199000239372253},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.4887000024318695},{"id":"https://openalex.org/keywords/calibration","display_name":"Calibration","score":0.47690001130104065},{"id":"https://openalex.org/keywords/consistency","display_name":"Consistency (knowledge bases)","score":0.47269999980926514},{"id":"https://openalex.org/keywords/confidence-interval","display_name":"Confidence interval","score":0.44679999351501465}],"concepts":[{"id":"https://openalex.org/C51110983","wikidata":"https://www.wikidata.org/wiki/Q16503490","display_name":"Overconfidence effect","level":2,"score":0.8220999836921692},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6693999767303467},{"id":"https://openalex.org/C55439883","wikidata":"https://www.wikidata.org/wiki/Q360812","display_name":"Correctness","level":2,"score":0.6682999730110168},{"id":"https://openalex.org/C43214815","wikidata":"https://www.wikidata.org/wiki/Q7310987","display_name":"Reliability (semiconductor)","level":3,"score":0.5999000072479248},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.5647000074386597},{"id":"https://openalex.org/C2780009758","wikidata":"https://www.wikidata.org/wiki/Q6804172","display_name":"Measure (data warehouse)","level":2,"score":0.5199000239372253},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5019000172615051},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.4887000024318695},{"id":"https://openalex.org/C165838908","wikidata":"https://www.wikidata.org/wiki/Q736777","display_name":"Calibration","level":2,"score":0.47690001130104065},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.47269999980926514},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4699999988079071},{"id":"https://openalex.org/C44249647","wikidata":"https://www.wikidata.org/wiki/Q208498","display_name":"Confidence interval","level":2,"score":0.44679999351501465},{"id":"https://openalex.org/C2909755999","wikidata":"https://www.wikidata.org/wiki/Q4751126","display_name":"Low Confidence","level":2,"score":0.4341999888420105},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.387800008058548},{"id":"https://openalex.org/C103088060","wikidata":"https://www.wikidata.org/wiki/Q1062839","display_name":"Error detection and correction","level":2,"score":0.38339999318122864},{"id":"https://openalex.org/C183115368","wikidata":"https://www.wikidata.org/wiki/Q856577","display_name":"Weighting","level":2,"score":0.3822000026702881},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.365200012922287},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.36090001463890076},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.35499998927116394},{"id":"https://openalex.org/C13662513","wikidata":"https://www.wikidata.org/wiki/Q5160087","display_name":"Confidence distribution","level":3,"score":0.353300005197525},{"id":"https://openalex.org/C2776291640","wikidata":"https://www.wikidata.org/wiki/Q2912517","display_name":"Value (mathematics)","level":2,"score":0.3287000060081482},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.3057999908924103},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2930999994277954},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.2897999882698059},{"id":"https://openalex.org/C70136482","wikidata":"https://www.wikidata.org/wiki/Q13583781","display_name":"A-weighting","level":3,"score":0.2890999913215637},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.28349998593330383},{"id":"https://openalex.org/C153701036","wikidata":"https://www.wikidata.org/wiki/Q659974","display_name":"Trustworthiness","level":2,"score":0.2743000090122223},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.2709999978542328},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.26930001378059387},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.2603999972343445},{"id":"https://openalex.org/C6239989","wikidata":"https://www.wikidata.org/wiki/Q5419224","display_name":"Exact statistics","level":3,"score":0.25049999356269836}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:doi:10.48550/arxiv.2603.06604","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"publisher-specific-oa","license_id":"https://openalex.org/licenses/publisher-specific-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},{"id":"doi:10.48550/arxiv.2603.06604","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.06604","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":"pmh:doi:10.48550/arxiv.2603.06604","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"publisher-specific-oa","license_id":"https://openalex.org/licenses/publisher-specific-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},"sustainable_development_goals":[{"display_name":"Peace, Justice and strong institutions","score":0.8133288025856018,"id":"https://metadata.un.org/sdg/16"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"As":[0],"large":[1],"language":[2],"models":[3],"(LLMs)":[4],"are":[5],"increasingly":[6],"deployed":[7],"in":[8,140],"critical":[9],"decision-making":[10],"systems,":[11],"the":[12,187,201],"lack":[13],"of":[14,54,96,194,200],"reliable":[15,82],"methods":[16,119],"to":[17,136,154,162,197],"measure":[18],"their":[19],"uncertainty":[20],"presents":[21],"a":[22,28,73],"fundamental":[23],"trustworthiness":[24],"risk.":[25],"We":[26,65,173],"introduce":[27],"normalized":[29,74],"confidence":[30,75,83,112,138,171],"score":[31,76],"based":[32],"on":[33,164,206],"output":[34],"anchor":[35],"token":[36],"probabilities:":[37],"classification":[38],"labels":[39],"for":[40,47,85],"structured":[41],"tasks":[42,92],"and":[43,56,61,77,93,99,122,156,168],"self-evaluation":[44,78],"responses":[45],"(Yes/No)":[46],"open-ended":[48],"generation.":[49],"This":[50],"enables":[51],"direct":[52],"detection":[53,87],"errors":[55],"hallucinations":[57],"with":[58,134],"minimal":[59],"overhead":[60],"without":[62],"external":[63],"validation.":[64],"make":[66],"three":[67],"key":[68],"contributions.":[69],"First,":[70],"we":[71,130],"propose":[72,131],"framework":[79],"that":[80,106,146,182],"exposes":[81],"estimates":[84],"error":[86,159],"across":[88],"seven":[89],"diverse":[90],"benchmark":[91],"five":[94],"LLMs":[95],"varying":[97],"architectures":[98],"sizes.":[100],"Second,":[101],"our":[102],"theoretical":[103],"analysis":[104],"reveals":[105],"supervised":[107],"fine-tuning":[108],"(SFT)":[109],"yields":[110],"well-calibrated":[111],"through":[113,177],"maximum-likelihood":[114],"estimation,":[115],"whereas":[116],"reinforcement":[117],"learning":[118],"(PPO,":[120],"GRPO)":[121],"DPO":[123,169],"induce":[124],"overconfidence":[125],"via":[126],"reward":[127],"exploitation.":[128],"Third,":[129],"post-RL":[132],"SFT":[133,147],"self-distillation":[135],"restore":[137],"reliability":[139],"RL-trained":[141],"models.":[142],"Empirical":[143],"results":[144],"demonstrated":[145,174],"improved":[148],"average":[149],"confidence-correctness":[150],"AUROC":[151],"from":[152,160],"0.806":[153],"0.879":[155],"reduced":[157],"calibration":[158],"0.163":[161],"0.034":[163],"Qwen3-4B,":[165],"while":[166],"GRPO":[167],"degraded":[170],"reliability.":[172],"practical":[175],"value":[176],"adaptive":[178],"retrieval-augmented":[179],"generation":[180],"(RAG)":[181],"selectively":[183],"retrieves":[184],"context":[185],"when":[186],"model":[188],"lacks":[189],"confidence,":[190],"using":[191],"only":[192],"58\\%":[193],"retrieval":[195],"operations":[196],"recover":[198],"95\\%":[199],"maximum":[202],"achievable":[203],"accuracy":[204],"gain":[205],"TriviaQA":[207]},"counts_by_year":[],"updated_date":"2026-07-15T18:14:33.161393","created_date":"2026-03-11T00:00:00"}