{"id":"https://openalex.org/W7165669400","doi":"https://doi.org/10.48550/arxiv.2606.20898","title":"The Token Tax of Epistemic Accuracy: Comparing RAG and Long-Context Architectures for Document-Grounded Generative AI Applications","display_name":"The Token Tax of Epistemic Accuracy: Comparing RAG and Long-Context Architectures for Document-Grounded Generative AI Applications","publication_year":2026,"publication_date":"2026-06-18","ids":{"openalex":"https://openalex.org/W7165669400","doi":"https://doi.org/10.48550/arxiv.2606.20898"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.20898","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.20898","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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.2606.20898","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5076886608","display_name":"Austin Hamilton","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hamilton, Austin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139216310","display_name":"Ryan Singh","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Singh, Ryan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5001331990","display_name":"Michael Wise","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wise, Michael","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139207383","display_name":"Ibrahim Yousif","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yousif, Ibrahim","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139185284","display_name":"Arthur Carvalho","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Carvalho, Arthur","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139129357","display_name":"Zhe Shan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shan, Zhe","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139161807","display_name":"Mohammad Mayyas","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mayyas, Mohammad","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139163989","display_name":"Lora A. Cavuoto","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cavuoto, Lora A.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5024863972","display_name":"Fadel M. Megahed","orcid":"https://orcid.org/0000-0003-2194-5110"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Megahed, Fadel M.","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/T10028","display_name":"Topic Modeling","score":0.3864000141620636,"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":0.3864000141620636,"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/T11636","display_name":"Artificial Intelligence in Healthcare and Education","score":0.13899999856948853,"subfield":{"id":"https://openalex.org/subfields/2718","display_name":"Health Informatics"},"field":{"id":"https://openalex.org/fields/27","display_name":"Medicine"},"domain":{"id":"https://openalex.org/domains/4","display_name":"Health Sciences"}},{"id":"https://openalex.org/T10883","display_name":"Ethics and Social Impacts of AI","score":0.08169999718666077,"subfield":{"id":"https://openalex.org/subfields/3311","display_name":"Safety Research"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.7962999939918518},{"id":"https://openalex.org/keywords/correctness","display_name":"Correctness","score":0.7847999930381775},{"id":"https://openalex.org/keywords/framing","display_name":"Framing (construction)","score":0.5896000266075134},{"id":"https://openalex.org/keywords/generative-grammar","display_name":"Generative grammar","score":0.571399986743927},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.4481000006198883},{"id":"https://openalex.org/keywords/generative-model","display_name":"Generative model","score":0.3188000023365021}],"concepts":[{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.7962999939918518},{"id":"https://openalex.org/C55439883","wikidata":"https://www.wikidata.org/wiki/Q360812","display_name":"Correctness","level":2,"score":0.7847999930381775},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7254999876022339},{"id":"https://openalex.org/C169087156","wikidata":"https://www.wikidata.org/wiki/Q2131593","display_name":"Framing (construction)","level":2,"score":0.5896000266075134},{"id":"https://openalex.org/C39890363","wikidata":"https://www.wikidata.org/wiki/Q36108","display_name":"Generative grammar","level":2,"score":0.571399986743927},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.4481000006198883},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3995000123977661},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3930000066757202},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.3188000023365021},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.29499998688697815},{"id":"https://openalex.org/C133462117","wikidata":"https://www.wikidata.org/wiki/Q4929239","display_name":"Data collection","level":2,"score":0.2874000072479248},{"id":"https://openalex.org/C204434341","wikidata":"https://www.wikidata.org/wiki/Q357789","display_name":"Adjudication","level":2,"score":0.2786000072956085},{"id":"https://openalex.org/C118930307","wikidata":"https://www.wikidata.org/wiki/Q600590","display_name":"Tuple","level":2,"score":0.27790001034736633}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.20898","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.20898","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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.2606.20898","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.20898","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Document-grounded":[0],"assistants":[1],"built":[2],"on":[3,19,67,81],"large":[4],"language":[5,139],"models":[6],"are":[7],"increasingly":[8],"used":[9],"in":[10,56,106,122],"high-stakes,":[11],"knowledge-intensive":[12],"work.":[13],"Their":[14],"usefulness,":[15],"however,":[16],"may":[17],"depend":[18],"how":[20],"evidence":[21],"is":[22],"allocated":[23],"before":[24],"generation.":[25],"We":[26,58,71,86,114,166,178],"investigate":[27],"such":[28],"a":[29,42,99,103,119],"claim":[30],"by":[31],"comparing":[32],"two":[33,62,137],"grounding":[34],"architectures:":[35],"(a)":[36],"retrieval-augmented":[37],"generation":[38],"(RAG)":[39],"that":[40,79,88],"retrieves":[41],"few":[43],"relevant":[44],"passages,":[45],"and":[46,141],"(b)":[47],"long-context":[48],"prompting,":[49],"which":[50],"loads":[51],"the":[52,83,149,162,171,181],"whole":[53],"document":[54],"collection":[55],"context.":[57],"view":[59],"these":[60],"as":[61,170],"regimes":[63],"of":[64,174,183],"\"epistemic":[65,73],"access\"":[66],"an":[68,127],"accuracy--cost":[69],"frontier.":[70],"use":[72],"accuracy\"":[74],"to":[75,109],"capture":[76],"model":[77],"correctness":[78,151],"depends":[80],"having":[82],"right":[84],"evidence.":[85],"posit":[87],"broader":[89,175],"access":[90],"(via":[91],"long":[92],"context)":[93],"can":[94],"increase":[95,105],"it,":[96],"but":[97,158],"with":[98,118],"\"token":[100],"tax\"":[101],"(i.e.,":[102],"substantial":[104],"cost":[107],"due":[108],"larger":[110],"input":[111],"token":[112,164,172],"consumption).":[113],"probe":[115],"this":[116,168],"framing":[117],"case":[120],"study":[121],"manufacturing":[123],"safety":[124],"training.":[125],"Using":[126],"expert-validated":[128],"benchmark,":[129],"we":[130],"evaluate":[131],"972":[132],"answers":[133],"across":[134],"three":[135,142],"machines,":[136],"small":[138],"models,":[140],"retrieval/in-context":[143],"prompting":[144,147],"approaches.":[145],"Long-context":[146],"achieved":[148],"highest":[150],"(73.1%":[152],"vs.":[153],"65.4%":[154],"for":[155,186],"semantic":[156],"RAG),":[157],"at":[159],"26":[160],"times":[161],"per-query":[163],"cost.":[165],"interpret":[167],"gap":[169],"tax":[173],"evidentiary":[176],"access.":[177],"carefully":[179],"discuss":[180],"implications":[182],"our":[184],"findings":[185],"resource-constrained":[187],"organizations.":[188]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-06-24T00:00:00"}