@inproceedings{nguyen-etal-2026-feedback,
title = "Feedback Blindness: Diagnosing Silent Failures in Closed-Loop Agentic Data-Generation Pipelines",
author = "Nguyen, Loc and
Long, Nguyen Khanh and
Nam, Vu Gia and
Nhat, Nguyen Tran Minh",
editor = "Kamalloo, Ehsan and
Gontier, Nicolas and
Rozanov, Nikolai and
Hashemzadeh, Maryam and
Arora, Akhil",
booktitle = "Proceedings of the Second Workshop for {RE}search on Agent Language Models ({REALM} 2026)",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.realm-1.25/",
pages = "374--381",
ISBN = "979-8-89176-485-9",
abstract = "Closed-loop synthetic-data pipelines combine planning, LLM generation, verification, and deduplication; accepted outputs steer later generation. We study a production pipeline for Vietnamese contact-center compliance data and identify feedback blindness: an internal feedback mechanism stops serving its intended function while every stage completes normally, leaving only downstream metric degradation as evidence. We diagnose four instances: cumulative-count eviction empties a recency window; batch-level deduplication misses candidates scored before any batch member is committed; static boundary exemplars repeatedly anchor generation; and a fixed-size anti-repeat reference set loses informativeness as the dataset grows. We describe architecture-level fixes, mechanism-specific health signals, and a controlled ablation and scaling evaluation. The central lesson is that aggregate duplicate and acceptance rates are insufficient for diagnosis; pipelines should also monitor the feedback mechanisms that produce those outcomes."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="nguyen-etal-2026-feedback">
<titleInfo>
<title>Feedback Blindness: Diagnosing Silent Failures in Closed-Loop Agentic Data-Generation Pipelines</title>
</titleInfo>
<name type="personal">
<namePart type="given">Loc</namePart>
<namePart type="family">Nguyen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nguyen</namePart>
<namePart type="given">Khanh</namePart>
<namePart type="family">Long</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Vu</namePart>
<namePart type="given">Gia</namePart>
<namePart type="family">Nam</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nguyen</namePart>
<namePart type="given">Tran</namePart>
<namePart type="given">Minh</namePart>
<namePart type="family">Nhat</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the Second Workshop for REsearch on Agent Language Models (REALM 2026)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ehsan</namePart>
<namePart type="family">Kamalloo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nicolas</namePart>
<namePart type="family">Gontier</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nikolai</namePart>
<namePart type="family">Rozanov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Maryam</namePart>
<namePart type="family">Hashemzadeh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Akhil</namePart>
<namePart type="family">Arora</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-485-9</identifier>
</relatedItem>
<abstract>Closed-loop synthetic-data pipelines combine planning, LLM generation, verification, and deduplication; accepted outputs steer later generation. We study a production pipeline for Vietnamese contact-center compliance data and identify feedback blindness: an internal feedback mechanism stops serving its intended function while every stage completes normally, leaving only downstream metric degradation as evidence. We diagnose four instances: cumulative-count eviction empties a recency window; batch-level deduplication misses candidates scored before any batch member is committed; static boundary exemplars repeatedly anchor generation; and a fixed-size anti-repeat reference set loses informativeness as the dataset grows. We describe architecture-level fixes, mechanism-specific health signals, and a controlled ablation and scaling evaluation. The central lesson is that aggregate duplicate and acceptance rates are insufficient for diagnosis; pipelines should also monitor the feedback mechanisms that produce those outcomes.</abstract>
<identifier type="citekey">nguyen-etal-2026-feedback</identifier>
<location>
<url>https://aclanthology.org/2026.realm-1.25/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>374</start>
<end>381</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Feedback Blindness: Diagnosing Silent Failures in Closed-Loop Agentic Data-Generation Pipelines
%A Nguyen, Loc
%A Long, Nguyen Khanh
%A Nam, Vu Gia
%A Nhat, Nguyen Tran Minh
%Y Kamalloo, Ehsan
%Y Gontier, Nicolas
%Y Rozanov, Nikolai
%Y Hashemzadeh, Maryam
%Y Arora, Akhil
%S Proceedings of the Second Workshop for REsearch on Agent Language Models (REALM 2026)
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-485-9
%F nguyen-etal-2026-feedback
%X Closed-loop synthetic-data pipelines combine planning, LLM generation, verification, and deduplication; accepted outputs steer later generation. We study a production pipeline for Vietnamese contact-center compliance data and identify feedback blindness: an internal feedback mechanism stops serving its intended function while every stage completes normally, leaving only downstream metric degradation as evidence. We diagnose four instances: cumulative-count eviction empties a recency window; batch-level deduplication misses candidates scored before any batch member is committed; static boundary exemplars repeatedly anchor generation; and a fixed-size anti-repeat reference set loses informativeness as the dataset grows. We describe architecture-level fixes, mechanism-specific health signals, and a controlled ablation and scaling evaluation. The central lesson is that aggregate duplicate and acceptance rates are insufficient for diagnosis; pipelines should also monitor the feedback mechanisms that produce those outcomes.
%U https://aclanthology.org/2026.realm-1.25/
%P 374-381
Markdown (Informal)
[Feedback Blindness: Diagnosing Silent Failures in Closed-Loop Agentic Data-Generation Pipelines](https://aclanthology.org/2026.realm-1.25/) (Nguyen et al., REALM 2026)
ACL