AI & Scientific Integrity

While large language models (LLMs) have been widely studied in scholarly workflows, e.g., for citation recommendation and literature summarization and more, their role in supporting research integrity remains underexplored. In this talk, we share our experience building and deploying two real-world systems that audit peer reviews and verify their factual grounding at scale. We evaluate the capabilities and limitations of LLMs in two key tasks: (1) assessing review quality along dimensions like specificity and tone, and (2) verifying whether reviewer claims are supported by the submitted paper. Using expert-annotated benchmarks, we compare static metrics, ML baselines, zero-shot LLMs, and fine-tuned models to assess alignment with human judgment. The talk will highlight methodological choices, deployment lessons, and empirical insights into where LLMs succeed and where hybrid approaches with interpretable ML and retrieval perform more reliably. We conclude with reflections on what infrastructure is needed to make use LLMs as a robust foundation for research integrity at scale.

Peer review is central to scientific decision-making, yet it is rarely evaluated or audited at scale. Growing submission volumes and the increasing use of large language models (LLMs) in drafting reviews have introduced new challenges for transparency, accountability, and quality control. While many communities have focused on detecting AI-generated text, authorship alone does not address the core issues: Are reviews constructive? Are reviewer claims grounded in the submitted paper? How should we reason about hybrid human–AI collaboration? In this industry talk, we present Reviewerly’s retrieval-centered infrastructure for auditing peer review at scale. We describe three deployed systems: Peeriscope, which evaluates review quality across multiple dimensions; Peerispect, which uses retrieval-augmented generation (RAG) to verify whether reviewer claims are supported by the manuscript; and PeerPrism, which studies human–AI collaboration and disentangles idea origin from text origin. We will share architectural design decisions, lessons learned from real-world deployment, and practical trade-offs between interpretability and model complexity. The talk will include short live demonstrations (https://app.reviewer.ly/app/peerispect & https://app.reviewer.ly/app/peeriscope) of our systems to illustrate how retrieval and LLM-based pipelines can be used to support editors and stakeholders in high-volume evaluation settings.

Peer-review evaluation is increasingly being automated with LLM-as-a-judge metrics, but this creates a measurement risk. A review may receive a high score because it is fluent, organized, and polished, rather than because it provides a strong evaluation of the paper. This risk is especially important in AI-assisted reviewing, where reviewers may use LLMs to improve clarity or presentation while preserving the underlying judgments. We propose a statistical framework for testing whether peer-review evaluation metrics capture substantive review quality beyond surface-level linguistic form. The framework compares original human reviews with faithful LLM rewrites that preserve the same evaluative content while changing wording and presentation. Using 4,044 meaning-preserving rewrites derived from 674 human reviews from ICLR and NeurIPS, we evaluate 29 content-oriented peer-review evaluation metrics drawn from four prior works through complementary tests of surface sensitivity and robustness. Although these metrics are intended to capture review properties beyond surface-level, writing-dependent characteristics, we find that sensitivity to rewriting is widespread. Twenty-three metrics assign significantly different scores to reviews whose evaluative content is preserved, while only seven satisfy our robustness criterion. The patterns are consistent across two LLM judge models, suggesting that the issue is not specific to a single judge. These findings show that many peer-review evaluation metrics partially conflate review quality with linguistic presentation, and indicate that robustness to meaning-preserving rewriting should be validated before such metrics are used to compare human-written, AI-assisted, and AI-generated reviews. All evaluation results, prompts, and implementation code are publicly available on our GitHub repository at urlhttps://anonymous.4open.science/r/judging_reviews_by_cover-BB10/

Peer review plays a central role in scholarly publishing, yet verifying whether reviewer claims are supported by manuscript evidence remains a largely manual and time-consuming process. We present Peerify, a pipeline for manuscript-grounded verification of peer-review claims. Given a manuscript and a review comment, Peerify pipeline decomposes reviews into atomic claims, retrieves relevant manuscript evidence, and determines whether each claim is supported by the paper. To support the development and evaluation of the pipeline, we construct a benchmark of 800 claims derived from authentic peer-review interactions collected from NeurIPS 2024 and ICLR 2024. We evaluate state-of-the-art language models and retrieval strategies within Peerify pipeline. Our results demonstrate the importance of retrieval-centered verification and claim decomposition, while highlighting the challenges posed by ambiguous and interpretive reviewer claims.

Abstract The increasing scale and variability of peer review in scholarly venues has created an urgent need for systematic, interpretable, and extensible tools to assess review quality. We present PeeriScope, a modular platform that integrates structured features, rubric-guided large language model assessments, and supervised prediction to evaluate peer review quality along multiple dimensions. Designed for openness and integration, PeeriScope provides both a public interface and a documented API, supporting practical deployment and research extensibility. The demonstration illustrates its use for reviewer self-assessment, editorial triage, and large-scale auditing, and it enables continued development of quality evaluation methods within scientific peer review. PeeriScope is live and publicly accessible at https://app.reviewer.ly/app/peeriscope

Peer review is central to scientific publishing, yet reviewers frequently include claims that are subjective, rhetorical, or misaligned with the submitted work. Assessing whether review statements are factual and verifiable is crucial for fairness and accountability. At the scale of modern conferences and journals, manually inspecting the grounding of such claims is infeasible. We present Peerispect, an interactive system that operationalizes claim-level verification in peer reviews by extracting check-worthy claims from peer reviews, retrieving relevant evidence from the manuscript, and verifying the claims through natural language inference. Results are presented through a visual interface that highlights evidence directly in the paper, enabling rapid inspection and interpretation. Peerispect is designed as a modular Information Retrieval (IR) pipeline, supporting alternative retrievers, rerankers, and verifiers, and is intended for use by reviewers, authors, and program committees. We demonstrate Peerispect through a live, publicly available demo (https://app.reviewer.ly/app/peerispect) and API services (https://github.com/Reviewerly-Inc/Peerispect), accompanied by a video tutorial (https://bit.ly/3LMobm8).

Large Language Models (LLMs) are increasingly used in scientific peer review, assisting with drafting, rewriting, expansion, and refinement. However, existing peer-review LLM detection methods largely treat authorship as a binary problem—human vs. AI—without accounting for the hybrid nature of modern review workflows. In practice, evaluative ideas and surface realization may originate from different sources, creating a spectrum of human-AI collaboration. In this work, we introduce PeerPrism, a large-scale benchmark of 20,690 peer reviews explicitly designed to disentangle idea provenance from text provenance. We construct controlled generation regimes spanning fully human, fully synthetic, and multiple hybrid transformations. This design enables systematic evaluation of whether detectors identify the origin of the surface text or the origin of the evaluative reasoning. We benchmark state-of-the-art LLM text detection methods on PeerPrism. While several methods achieve high accuracy on the standard binary task (human vs. fully synthetic), their predictions diverge sharply under hybrid regimes. In particular, when ideas originate from humans but the surface text is AI-generated, detectors frequently disagree and produce contradictory classifications. Accompanied by stylometric and semantic analyses, our results show that current detection methods conflate surface realization with intellectual contribution. Overall, we demonstrate that LLM detection in peer review cannot be reduced to a binary attribution problem. Instead, authorship must be modeled as a multidimensional construct spanning semantic reasoning and stylistic realization. PeerPrism is the first benchmark evaluating human-AI collaboration in these settings. We release all code, data, prompts, and evaluation scripts to facilitate reproducible research at https://github.com/Reviewerly-Inc/PeerPrism.

Background. Predicting the future influence of scientific papers remains a longstanding challenge in bibliometrics and information retrieval. Traditional regression and embedding-based methods estimate citation counts from textual features but fail to capture the inherently relational and temporal nature of scholarly impact. Objective. This paper proposes a preference-aligned framework for forecasting and generating scholarly influence from paper abstracts and temporal cues. Rather than predicting absolute citation values, we model the relative likelihood that one paper will accrue more citations than another within a shared temporal time frame. Methodology. Impact-DPO integrates temporally informed prompting with direct preference optimization, enabling LLMs to learn comparative influence patterns without explicit graph message passing. We formalize citation forecasting as pairwise preference learning on temporal text-attributed graphs, using publication year as a minimal temporal signal. Experiments were conducted on two large-scale domains, i.e., Computer Science and Physics, spanning three temporal splits (2018–2020). Results. We show that Impact-DPO achieves the highest pairwise accuracy across all splits, reaching up to 83%, outperforming SPECTER2+SVR by 12 pp and zero-shot prompting by more than 15 pp. Preference alignment yields an average improvement of roughly +30 pp over binary-classification baselines overall; importantly, supplementary same-backbone Qwen 2.5–7B BC runs remain near chance (about 50–58% across splits), showing that the gain is not explained by backbone capacity alone. A lower regularization parameter (β = 0.1) consistently produces optimal results, consistent with a low-gap preference regime in citation data. Generative evaluations further reveal that model-generated text aligns more closely with highly cited papers (KS=0.0744, p=0.0065), demonstrating emergent generative alignment with influential scholarly language. Code and data availability. All code, data-preprocessing scripts, and evaluation notebooks are available at impact-dpo repository (https://github.com/parhamouni/impact-dpo).

Large language models (LLMs) are rapidly moving from experimental prototypes into operational peer review workflows, where they are used to polish, rewrite, expand, synthesize, screen, and in some cases generate reviews. This shift creates an urgent challenge for the organizations that depend on expert assessment, including publishers, conferences, funding agencies, research institutions, and enterprise research teams. Rather than asking only whether AI can improve peer review, this talk argues that we must first confront a more fundamental evaluation question. Can we reliably measure what makes a review useful, fair, accurate, and decision-relevant? The talk examines this question through the lens of AI-assisted peer review and shows that commonly used signals, including recommendation agreement, similarity to reference reviews, writing fluency, and single score LLM-as-a-judge evaluation, often fail to capture substantive review quality. Drawing on real world deployment experience at Reviewerly, we present case studies in which evaluation metrics reward surface-level improvements such as polish, organization, and LLM like wording while overlooking the evidentiary content, specificity, correctness, and editorial usefulness of the review. We then discuss practical lessons for building reliable evaluation systems in real peer review workflows, including stress testing review metrics, separating surface sensitivity from practical robustness, comparing judge models under realistic perturbations, and designing layered evaluation pipelines that are interpretable, evidence-grounded, and aligned with editorial needs. Although peer review is the central case study, the broader message applies to many industry and organizational settings where AI is being introduced into expert judgment workflows. Responsible automation requires more than stronger generation models. It requires evaluation methods that measure the qualities we actually want AI systems to improve.

Peer review is foundational to academic publishing, yet the quality of reviews remains difficult to assess at scale due to subjectivity, inconsistency, and the lack of standardized evaluation mechanisms. This talk presents our experience developing and deploying a scalable framework for assessing review quality in operational settings. We combine two complementary approaches: interpretable machine learning models built on quantifiable review- and reviewer-level features, and the application of large language models (LLMs), including Qwen, Phi, and GPT-4o, in zero- and few-shot configurations for textual quality evaluation. We also explore the fine-tuning of LLMs on expert-annotated datasets to examine their upper-bound capabilities. To benchmark these methods, we constructed a dataset of over 700 paper–review pairs labeled by domain experts across multiple quality dimensions. Our findings demonstrate that transparent, feature-based models consistently outperform LLMs in reliability and generalization, particularly when evaluating conceptual depth and argumentative structure. The talk will highlight key engineering choices, deployment challenges, and broader implications for integrating automated review evaluation into scholarly workflows.

No abstract available for this paper.

The quality of peer review plays a critical role in scientific publishing, yet remains poorly understood and challenging to evaluate at scale. In this work, we introduce RottenReviews, a benchmark designed to facilitate systematic assessment of review quality. RottenReviews comprises over 15,000 submissions from four distinct academic venues enriched with over 9,000 reviewer scholarly profiles and paper metadata. We define and compute a diverse set of quantifiable review-dependent and reviewer-dependent metrics, and compare them against structured assessments from large language models (LLMs) and expert human annotations. Our humanannotated subset includes over 700 paper–review pairs labeled across 13 explainable and conceptual dimensions of review quality. Our empirical findings reveal that LLMs, both zero-shot and finetuned, exhibit limited alignment with human expert evaluations of peer review quality. Surprisingly, simple interpretable models trained on quantifiable features outperform fine-tuned LLMs in predicting overall review quality. We publicly release all data, code, and models at https://www.github.com/sadjadeb/RottenReviews to support further research in this area.

The peer review process is a fundamental aspect of academic publishing, ensuring the quality and credibility of scholarly work. In this talk, we will explore the critical challenges associated specifically with the assignment of reviewers to submitted papers. We will introduce Reviewerly, our innovative solution designed to enhance the efficiency and effectiveness of reviewer assignments by leveraging data from diverse sources, including OpenAlex, PubMed, and DBLP. By modeling the reviewer assignment problem as an information retrieval task, we focus on retrieving a pool of relevant and diverse reviewers for each paper. We will highlight the challenges we faced and showcase the benefits of this approach in addressing the reviewer assignment problem.