@inproceedings{wang-etal-2026-ai,
title = "From {AI} Reviewers to Evidence Assistants: Quantifying the Human-{AI} Responsibility Boundary in Peer Review",
author = "Wang, Zhouyang and
Xie, Qiujie and
Zhu, Minjun and
Li, Shichen and
Huang, Shulin and
Cui, Han and
Ding, Yiran and
Lu, Panzhong and
Liu, Zhenhao and
Shen, Fuchen and
Pan, Junshu and
Yin, Dalv and
Sun, Ke and
Ning, Zhiyuan and
Weng, Yixuan and
Li, Peifeng and
Zhang, Yue",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.1296/",
pages = "29494--29519",
ISBN = "979-8-89176-459-0",
abstract = "The rapid growth of AI conference submissions is putting new pressure on peer review. AI reviewer systems are increasingly proposed as support, but prior work leaves unresolved what responsibility their outputs should carry when they can surface useful critiques yet remain risky as independent judgments. We frame this as a responsibility-boundary problem. Using 600 ICLR 2026 submissions, 2,231 human reviews, and 3,600 AI reviews, we operationalize this boundary through usable feedback, rating calibration, panel breadth, and grounded synthesis. AI can prepare critiques and organize evidence, while scoring, independent panel judgment, high-level synthesis, and final responsibility remain human-led. A GLM-5.2 validation on 100 NeurIPS 2022 papers with pre-ChatGPT reviews reproduces the workflow and calibration patterns. Motivated by this boundary, we develop Review Copilot, a workflow in which AI suggestions are inspected, edited, or rejected by human reviewers. In an initial controlled reviewer-in-the-loop study, Human+AI reviews improve actionability, evidence support, and professionalism relative to standalone baselines, while scores and recommendations remain human-authored. A user study of 275 reviewers across 411 papers further supports its practical usefulness. Together, these results support a review paradigm in which AI expands evidence-grounded critique while humans retain judgment, synthesis, and accountability."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wang-etal-2026-ai">
<titleInfo>
<title>From AI Reviewers to Evidence Assistants: Quantifying the Human-AI Responsibility Boundary in Peer Review</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zhouyang</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qiujie</namePart>
<namePart type="family">Xie</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Minjun</namePart>
<namePart type="family">Zhu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shichen</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shulin</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Han</namePart>
<namePart type="family">Cui</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yiran</namePart>
<namePart type="family">Ding</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Panzhong</namePart>
<namePart type="family">Lu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhenhao</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fuchen</namePart>
<namePart type="family">Shen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Junshu</namePart>
<namePart type="family">Pan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dalv</namePart>
<namePart type="family">Yin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ke</namePart>
<namePart type="family">Sun</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhiyuan</namePart>
<namePart type="family">Ning</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yixuan</namePart>
<namePart type="family">Weng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Peifeng</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Zhang</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 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</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">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>The rapid growth of AI conference submissions is putting new pressure on peer review. AI reviewer systems are increasingly proposed as support, but prior work leaves unresolved what responsibility their outputs should carry when they can surface useful critiques yet remain risky as independent judgments. We frame this as a responsibility-boundary problem. Using 600 ICLR 2026 submissions, 2,231 human reviews, and 3,600 AI reviews, we operationalize this boundary through usable feedback, rating calibration, panel breadth, and grounded synthesis. AI can prepare critiques and organize evidence, while scoring, independent panel judgment, high-level synthesis, and final responsibility remain human-led. A GLM-5.2 validation on 100 NeurIPS 2022 papers with pre-ChatGPT reviews reproduces the workflow and calibration patterns. Motivated by this boundary, we develop Review Copilot, a workflow in which AI suggestions are inspected, edited, or rejected by human reviewers. In an initial controlled reviewer-in-the-loop study, Human+AI reviews improve actionability, evidence support, and professionalism relative to standalone baselines, while scores and recommendations remain human-authored. A user study of 275 reviewers across 411 papers further supports its practical usefulness. Together, these results support a review paradigm in which AI expands evidence-grounded critique while humans retain judgment, synthesis, and accountability.</abstract>
<identifier type="citekey">wang-etal-2026-ai</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1296/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>29494</start>
<end>29519</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T From AI Reviewers to Evidence Assistants: Quantifying the Human-AI Responsibility Boundary in Peer Review
%A Wang, Zhouyang
%A Xie, Qiujie
%A Zhu, Minjun
%A Li, Shichen
%A Huang, Shulin
%A Cui, Han
%A Ding, Yiran
%A Lu, Panzhong
%A Liu, Zhenhao
%A Shen, Fuchen
%A Pan, Junshu
%A Yin, Dalv
%A Sun, Ke
%A Ning, Zhiyuan
%A Weng, Yixuan
%A Li, Peifeng
%A Zhang, Yue
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F wang-etal-2026-ai
%X The rapid growth of AI conference submissions is putting new pressure on peer review. AI reviewer systems are increasingly proposed as support, but prior work leaves unresolved what responsibility their outputs should carry when they can surface useful critiques yet remain risky as independent judgments. We frame this as a responsibility-boundary problem. Using 600 ICLR 2026 submissions, 2,231 human reviews, and 3,600 AI reviews, we operationalize this boundary through usable feedback, rating calibration, panel breadth, and grounded synthesis. AI can prepare critiques and organize evidence, while scoring, independent panel judgment, high-level synthesis, and final responsibility remain human-led. A GLM-5.2 validation on 100 NeurIPS 2022 papers with pre-ChatGPT reviews reproduces the workflow and calibration patterns. Motivated by this boundary, we develop Review Copilot, a workflow in which AI suggestions are inspected, edited, or rejected by human reviewers. In an initial controlled reviewer-in-the-loop study, Human+AI reviews improve actionability, evidence support, and professionalism relative to standalone baselines, while scores and recommendations remain human-authored. A user study of 275 reviewers across 411 papers further supports its practical usefulness. Together, these results support a review paradigm in which AI expands evidence-grounded critique while humans retain judgment, synthesis, and accountability.
%U https://aclanthology.org/2026.emnlp-main.1296/
%P 29494-29519
Markdown (Informal)
[From AI Reviewers to Evidence Assistants: Quantifying the Human-AI Responsibility Boundary in Peer Review](https://aclanthology.org/2026.emnlp-main.1296/) (Wang et al., EMNLP 2026)
ACL
- Zhouyang Wang, Qiujie Xie, Minjun Zhu, Shichen Li, Shulin Huang, Han Cui, Yiran Ding, Panzhong Lu, Zhenhao Liu, Fuchen Shen, Junshu Pan, Dalv Yin, Ke Sun, Zhiyuan Ning, Yixuan Weng, Peifeng Li, and Yue Zhang. 2026. From AI Reviewers to Evidence Assistants: Quantifying the Human-AI Responsibility Boundary in Peer Review. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 29494–29519, Budapest, Hungary. Association for Computational Linguistics.