{"id":"https://openalex.org/W4412945016","doi":"https://doi.org/10.18653/v1/2025.acl-long.1016","title":"Navigating Rifts in Human-LLM Grounding: Study and Benchmark","display_name":"Navigating Rifts in Human-LLM Grounding: Study and Benchmark","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412945016","doi":"https://doi.org/10.18653/v1/2025.acl-long.1016"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.acl-long.1016","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.1016","pdf_url":"https://aclanthology.org/2025.acl-long.1016.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.acl-long.1016.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5044691513","display_name":"Omar Shaikh","orcid":"https://orcid.org/0000-0003-1393-8041"},"institutions":[{"id":"https://openalex.org/I1290206253","display_name":"Microsoft (United States)","ror":"https://ror.org/00d0nc645","country_code":"US","type":"company","lineage":["https://openalex.org/I1290206253"]},{"id":"https://openalex.org/I4210164937","display_name":"Microsoft Research (United Kingdom)","ror":"https://ror.org/05k87vq12","country_code":"GB","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}],"countries":["GB","US"],"is_corresponding":false,"raw_author_name":"Omar Shaikh","raw_affiliation_strings":["Stanford University ,  Microsoft Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Stanford University ,  Microsoft Research","institution_ids":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5004148354","display_name":"Hussein Mozannar","orcid":"https://orcid.org/0009-0001-9660-9878"},"institutions":[{"id":"https://openalex.org/I1290206253","display_name":"Microsoft (United States)","ror":"https://ror.org/00d0nc645","country_code":"US","type":"company","lineage":["https://openalex.org/I1290206253"]},{"id":"https://openalex.org/I4210164937","display_name":"Microsoft Research (United Kingdom)","ror":"https://ror.org/05k87vq12","country_code":"GB","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}],"countries":["GB","US"],"is_corresponding":false,"raw_author_name":"Hussein Mozannar","raw_affiliation_strings":["Stanford University ,  Microsoft Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Stanford University ,  Microsoft Research","institution_ids":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5089760175","display_name":"Gagan Bansal","orcid":"https://orcid.org/0000-0002-8607-7011"},"institutions":[{"id":"https://openalex.org/I1290206253","display_name":"Microsoft (United States)","ror":"https://ror.org/00d0nc645","country_code":"US","type":"company","lineage":["https://openalex.org/I1290206253"]},{"id":"https://openalex.org/I4210164937","display_name":"Microsoft Research (United Kingdom)","ror":"https://ror.org/05k87vq12","country_code":"GB","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}],"countries":["GB","US"],"is_corresponding":false,"raw_author_name":"Gagan Bansal","raw_affiliation_strings":["Stanford University ,  Microsoft Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Stanford University ,  Microsoft Research","institution_ids":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5071293344","display_name":"Adam Fourney","orcid":"https://orcid.org/0000-0002-4986-7794"},"institutions":[{"id":"https://openalex.org/I1290206253","display_name":"Microsoft (United States)","ror":"https://ror.org/00d0nc645","country_code":"US","type":"company","lineage":["https://openalex.org/I1290206253"]},{"id":"https://openalex.org/I4210164937","display_name":"Microsoft Research (United Kingdom)","ror":"https://ror.org/05k87vq12","country_code":"GB","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}],"countries":["GB","US"],"is_corresponding":false,"raw_author_name":"Adam Fourney","raw_affiliation_strings":["Stanford University ,  Microsoft Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Stanford University ,  Microsoft Research","institution_ids":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5043228682","display_name":"Eric Horvitz","orcid":"https://orcid.org/0000-0002-8823-0614"},"institutions":[{"id":"https://openalex.org/I1290206253","display_name":"Microsoft (United States)","ror":"https://ror.org/00d0nc645","country_code":"US","type":"company","lineage":["https://openalex.org/I1290206253"]},{"id":"https://openalex.org/I4210164937","display_name":"Microsoft Research (United Kingdom)","ror":"https://ror.org/05k87vq12","country_code":"GB","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}],"countries":["GB","US"],"is_corresponding":false,"raw_author_name":"Eric Horvitz","raw_affiliation_strings":["Stanford University ,  Microsoft Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Stanford University ,  Microsoft Research","institution_ids":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":3,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"20832","last_page":"20847"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9154000282287598,"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":0.9154000282287598,"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/benchmark","display_name":"Benchmark (surveying)","score":0.6319796442985535},{"id":"https://openalex.org/keywords/ground","display_name":"Ground","score":0.567969799041748},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.475104421377182},{"id":"https://openalex.org/keywords/geology","display_name":"Geology","score":0.35295045375823975},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.2908996343612671},{"id":"https://openalex.org/keywords/electrical-engineering","display_name":"Electrical engineering","score":0.2124234139919281},{"id":"https://openalex.org/keywords/geodesy","display_name":"Geodesy","score":0.15626108646392822}],"concepts":[{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.6319796442985535},{"id":"https://openalex.org/C168993435","wikidata":"https://www.wikidata.org/wiki/Q6501125","display_name":"Ground","level":2,"score":0.567969799041748},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.475104421377182},{"id":"https://openalex.org/C127313418","wikidata":"https://www.wikidata.org/wiki/Q1069","display_name":"Geology","level":0,"score":0.35295045375823975},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.2908996343612671},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.2124234139919281},{"id":"https://openalex.org/C13280743","wikidata":"https://www.wikidata.org/wiki/Q131089","display_name":"Geodesy","level":1,"score":0.15626108646392822}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.acl-long.1016","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.1016","pdf_url":"https://aclanthology.org/2025.acl-long.1016.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.acl-long.1016","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.1016","pdf_url":"https://aclanthology.org/2025.acl-long.1016.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/13","display_name":"Climate action","score":0.6399999856948853}],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412945016.pdf","grobid_xml":"https://content.openalex.org/works/W4412945016.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2378211422","https://openalex.org/W4321353415","https://openalex.org/W2745001401","https://openalex.org/W2130974462","https://openalex.org/W2028665553","https://openalex.org/W2086519370","https://openalex.org/W4246352526"],"abstract_inverted_index":{"Language":[0],"models":[1,70,143],"excel":[2],"at":[3,170,179],"following":[4],"instructions":[5],"but":[6],"often":[7],"struggle":[8],"with":[9],"the":[10,149,191],"collaborative":[11],"aspects":[12],"of":[13,65,190],"conversation":[14,25],"that":[15,108,140,196],"humans":[16],"naturally":[17],"employ.This":[18],"limitation":[19],"in":[20,40,47,81,198],"groundingthe":[21],"process":[22],"by":[23],"which":[24],"participants":[26,197],"establish":[27],"mutual":[28],"understanding-can":[29],"lead":[30],"to":[31,37,71,92,100,136,151,185,187],"outcomes":[32],"ranging":[33],"from":[34,53,125],"frustrated":[35],"users":[36],"serious":[38],"consequences":[39],"high-stakes":[41],"scenarios.To":[42],"systematically":[43],"study":[44],"grounding":[45,66,75,110,172,184],"challenges":[46],"human-LLM":[48,84],"interactions,":[49],"we":[50,106,119,154,164],"analyze":[51],"logs":[52],"three":[54,88],"human-assistant":[55],"datasets:":[56],"WildChat,":[57],"MultiWOZ,":[58],"and":[59,68,73,83,95,156,194],"Bing":[60],"Chat.We":[61],"develop":[62,165],"a":[63,122,166,199],"taxonomy":[64],"acts":[67],"build":[69],"annotate":[72],"forecast":[74],"behavior.Our":[76],"findings":[77],"reveal":[78],"significant":[79],"differences":[80],"humanhuman":[82],"grounding:":[85],"LLMs":[86,134,158],"were":[87],"times":[89,97],"less":[90,98],"likely":[91,99],"initiate":[93,137],"clarification":[94],"sixteen":[96],"provide":[101],"follow-up":[102],"requests":[103],"than":[104],"humans.Additionally,":[105],"find":[107],"early":[109],"failures":[111],"predict":[112],"later":[113],"interaction":[114,129],"breakdowns.Building":[115],"on":[116,146],"these":[117],"insights,":[118],"introduce":[120],"RIFTS,":[121,147],"benchmark":[123],"derived":[124],"publicly":[126],"available":[127],"LLM":[128],"data":[130],"containing":[131],"situations":[132],"where":[133],"fail":[135],"grounding.We":[138],"note":[139],"current":[141],"frontier":[142],"perform":[144],"poorly":[145],"highlighting":[148],"need":[150],"reconsider":[152],"how":[153],"train":[155],"prompt":[157],"for":[159],"human":[160],"interaction.To":[161],"this":[162],"end,":[163],"preliminary":[167],"intervention":[168],"aimed":[169],"mitigating":[171],"failures.*":[173],"Research":[174],"performed":[175],"during":[176],"an":[177],"internship":[178],"Microsoft.":[180],"1":[181],"We":[182],"use":[183],"refer":[186],"Clark's":[188],"formulation":[189],"language,":[192],"gestures,":[193],"signaling":[195],"conver-":[200]},"counts_by_year":[{"year":2026,"cited_by_count":3}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}