Large Language Model (LLM) agents, which integrate planning, memory, reflection, and tool-use modules, have shown promise in solving complex, multi-step tasks. Yet their sophisticated architectures amplify vulnerability to cascading failures, where a single root-cause error propagates through subsequent decisions, leading to task failure. Current systems lack a framework that can comprehensively understand agent error in a modular and systemic way, and therefore fail to detect these errors accordingly. We address this gap with three contributions. First, we introduce the AgentErrorTaxonomy, a modular classification of failure modes spanning memory, reflection, planning, action, and system-level operations. Second, we construct AgentErrorBench, the first dataset of systematically annotated failure trajectories from ALFWorld, GAIA, and WebShop, grounding error analysis in real-world agent rollouts. Third, we propose AgentDebug, a debugging framework that isolates root-cause failures and provides corrective feedback, enabling agents to recover and iteratively improve. Experiments on AgentErrorBench show that AgentDebug achieves 24% higher all-correct accuracy and 17% higher step accuracy compared to the strongest baseline. Beyond detection, the targeted feedback generated by AgentDebug enables LLM agents to iteratively recover from failures, yielding up to 26% relative improvements in task success across ALFWorld, GAIA, and WebShop. These results establish principled debugging as a pathway to more reliable and adaptive LLM agents. The code and data will be available at https://github.com/ulab-uiuc/AgentDebug

Retractions play a vital role in maintaining scientific integrity, yet systematic studies of retractions in computer science and other STEM fields remain scarce. We present WithdrarXiv, the first large-scale dataset of withdrawn papers from arXiv, containing over 14,000 papers and their associated retraction comments spanning the repository's entire history through September 2024. Through careful analysis of author comments, we develop a comprehensive taxonomy of retraction reasons, identifying 10 distinct categories ranging from critical errors to policy violations. We demonstrate a simple yet highly accurate zero-shot automatic categorization of retraction reasons, achieving a weighted average F1-score of 0.96. Additionally, we release WithdrarXiv-SciFy, an enriched version including scripts for parsed full-text PDFs, specifically designed to enable research in scientific feasibility studies, claim verification, and automated theorem proving. These findings provide valuable insights for improving scientific quality control and automated verification systems. Finally, and most importantly, we discuss ethical issues and take a number of steps to implement responsible data release while fostering open science in this area.

With the advent of Neural Machine Translation (NMT) systems, the MT output has reached unprecedented accuracy levels which resulted in the ubiquity of MT tools on almost all online platforms with multilingual content. However, NMT systems, like other state-of-the-art AI generative systems, are prone to errors that are deemed machine hallucinations. The problem with NMT hallucinations is that they are remarkably \textit{fluent} hallucinations. Since they are trained to produce grammatically correct utterances, NMT systems are capable of producing mistranslations that are too fluent to be recognised by both users of the MT tool, as well as by automatic quality metrics that are used to gauge their performance. In this paper, we introduce an authentic dataset of machine translation critical errors to point to the ethical and safety issues involved in the common use of MT. The dataset comprises mistranslations of Arabic mental health postings manually annotated with critical error types. We also show how the commonly used quality metrics do not penalise critical errors and highlight this as a critical issue that merits further attention from researchers.

This study explores the use of Google Translate (GT) for translating mental healthcare (MHealth) information and evaluates its accuracy, comprehensibility, and implications for multilingual healthcare communication through analysing GT output in the MHealth domain from English to Persian, Arabic, Turkish, Romanian, and Spanish. Two datasets comprising MHealth information from the UK National Health Service website and information leaflets from The Royal College of Psychiatrists were used. Native speakers of the target languages manually assessed the GT translations, focusing on medical terminology accuracy, comprehensibility, and critical syntactic/semantic errors. GT output analysis revealed challenges in accurately translating medical terminology, particularly in Arabic, Romanian, and Persian. Fluency issues were prevalent across various languages, affecting comprehension, mainly in Arabic and Spanish. Critical errors arose in specific contexts, such as bullet-point formatting, specifically in Persian, Turkish, and Romanian. Although improvements are seen in longer-text translations, there remains a need to enhance accuracy in medical and mental health terminology and fluency, whilst also addressing formatting issues for a more seamless user experience. The findings highlight the need to use customised translation engines for Mhealth translation and the challenges when relying solely on machine-translated medical content, emphasising the crucial role of human reviewers in multilingual healthcare communication.

Assessing performance in Natural Language Processing is becoming increasingly complex. One particular challenge is the potential for evaluation datasets to overlap with training data, either directly or indirectly, which can lead to skewed results and overestimation of model performance. As a consequence, human evaluation is gaining increasing interest as a means to assess the performance and reliability of models. One such method is the red teaming approach, which aims to generate edge cases where a model will produce critical errors. While this methodology is becoming standard practice for generative AI, its application to the realm of conditional AI remains largely unexplored. This paper presents the first study on human-based red teaming for Machine Translation (MT), marking a significant step towards understanding and improving the performance of translation models. We delve into both human-based red teaming and a study on automation, reporting lessons learned and providing recommendations for both translation models and red teaming drills. This pioneering work opens up new avenues for research and development in the field of MT.

Explainable NLP techniques primarily explain by answering "Which tokens in the input are responsible for this prediction?''. We argue that for NLP models that make predictions by comparing two input texts, it is more useful to explain by answering "What differences between the two inputs explain this prediction?''. We introduce a technique to generate contrastive highlights that explain the predictions of a semantic divergence model via phrase-alignment-guided erasure. We show that the resulting highlights match human rationales of cross-lingual semantic differences better than popular post-hoc saliency techniques and that they successfully help people detect fine-grained meaning differences in human translations and critical machine translation errors.

A major challenge in the practical use of Machine Translation (MT) is that users lack guidance to make informed decisions about when to rely on outputs. Progress in quality estimation research provides techniques to automatically assess MT quality, but these techniques have primarily been evaluated in vitro by comparison against human judgments outside of a specific context of use. This paper evaluates quality estimation feedback in vivo with a human study simulating decision-making in high-stakes medical settings. Using Emergency Department discharge instructions, we study how interventions based on quality estimation versus backtranslation assist physicians in deciding whether to show MT outputs to a patient. We find that quality estimation improves appropriate reliance on MT, but backtranslation helps physicians detect more clinically harmful errors that QE alone often misses.

Widely used learned metrics for machine translation evaluation, such as COMET and BLEURT, estimate the quality of a translation hypothesis by providing a single sentence-level score. As such, they offer little insight into translation errors (e.g., what are the errors and what is their severity). On the other hand, generative large language models (LLMs) are amplifying the adoption of more granular strategies to evaluation, attempting to detail and categorize translation errors. In this work, we introduce xCOMET, an open-source learned metric designed to bridge the gap between these approaches. xCOMET integrates both sentence-level evaluation and error span detection capabilities, exhibiting state-of-the-art performance across all types of evaluation (sentence-level, system-level, and error span detection). Moreover, it does so while highlighting and categorizing error spans, thus enriching the quality assessment. We also provide a robustness analysis with stress tests, and show that xCOMET is largely capable of identifying localized critical errors and hallucinations.

ML APIs have greatly relieved application developers of the burden to design and train their own neural network models -- classifying objects in an image can now be as simple as one line of Python code to call an API. However, these APIs offer the same pre-trained models regardless of how their output is used by different applications. This can be suboptimal as not all ML inference errors can cause application failures, and the distinction between inference errors that can or cannot cause failures varies greatly across applications. To tackle this problem, we first study 77 real-world applications, which collectively use six ML APIs from two providers, to reveal common patterns of how ML API output affects applications' decision processes. Inspired by the findings, we propose ChameleonAPI, an optimization framework for ML APIs, which takes effect without changing the application source code. ChameleonAPI provides application developers with a parser that automatically analyzes the application to produce an abstract of its decision process, which is then used to devise an application-specific loss function that only penalizes API output errors critical to the application. ChameleonAPI uses the loss function to efficiently train a neural network model customized for each application and deploys it to serve API invocations from the respective application via existing interface. Compared to a baseline that selects the best-of-all commercial ML API, we show that ChameleonAPI reduces incorrect application decisions by 43%.

In this paper, we focus on how current Machine Translation (MT) tools perform on the translation of emotion-loaded texts by evaluating outputs from Google Translate according to a framework proposed in this paper. We propose this evaluation framework based on the Multidimensional Quality Metrics (MQM) and perform a detailed error analysis of the MT outputs. From our analysis, we observe that about 50% of the MT outputs fail to preserve the original emotion. After further analysis of the errors, we find that emotion carrying words and linguistic phenomena such as polysemous words, negation, abbreviation etc., are common causes for these translation errors.