Minqian Liu, Ioana Baldini, David Rabinowitz, David S Rosenberg, Sebastian Gehrmann, Mark Dredze. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Margaret Mitchell, Giuseppe Attanasio, Ioana Baldini, Miruna Clinciu, Jordan Clive, Pieter Delobelle, Manan Dey, Sil Hamilton, Timm Dill, Jad Doughman, Ritam Dutt, Avijit Ghosh, Jessica Zosa Forde, Carolin Holtermann, Lucie-Aimée Kaffee, Tanmay Laud, Anne Lauscher, Roberto L Lopez-Davila, Maraim Masoud, Nikita Nangia, Anaelia Ovalle, Giada Pistilli, Dragomir Radev, Beatrice Savoldi, Vipul Raheja, Jeremy Qin, Esther Ploeger, Arjun Subramonian, Kaustubh Dhole, Kaiser Sun, Amirbek Djanibekov, Jonibek Mansurov, Kayo Yin, Emilio Villa Cueva, Sagnik Mukherjee, Jerry Huang, Xudong Shen, Jay Gala, Hamdan Al-Ali, Tair Djanibekov, Nurdaulet Mukhituly, Shangrui Nie, Shanya Sharma, Karolina Stanczak, Eliza Szczechla, Tiago Timponi Torrent, Deepak Tunuguntla, Marcelo Viridiano, Oskar Van Der Wal, Adina Yakefu, Aurélie Névéol, Mike Zhang, Sydney Zink, Zeerak Talat. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Current datasets for unwanted social bias auditing are limited to studying protected demographic features such as race and gender. In this work, we introduce a comprehensive benchmark that is meant to capture the amplification of social bias, via stigmas, in generative language models. Taking inspiration from social science research, we start with a documented list of 93 US-centric stigmas and curate a question-answering (QA) dataset which involves simple social situations. Our benchmark, SocialStigmaQA, contains roughly 10K prompts, with a variety of prompt styles, carefully constructed to systematically test for both social bias and model robustness. We present results for SocialStigmaQA with two open source generative language models and we find that the proportion of socially biased output ranges from 45% to 59% across a variety of decoding strategies and prompting styles. We demonstrate that the deliberate design of the templates in our benchmark (e.g., adding biasing text to the prompt or using different verbs that change the answer that indicates bias) impacts the model tendencies to generate socially biased output. Additionally, through manual evaluation, we discover problematic patterns in the generated chain-of-thought output that range from subtle bias to lack of reasoning. Warning: This paper contains examples of text which are toxic, biased, and potentially harmful.

The increasing size of large language models (LLMs) has introduced challenges in their training and inference. Removing model components is perceived as a solution to tackle the large model sizes, however, existing pruning methods solely focus on performance, without considering an essential aspect for the responsible use of LLMs: model fairness. It is crucial to address the fairness of LLMs towards diverse groups, such as women, Black people, LGBTQ+, Jewish communities, among others, as they are being deployed and available to a wide audience. In this work, first, we investigate how attention heads impact fairness and performance in pre-trained transformer-based language models. We then propose a novel method to prune the attention heads that negatively impact fairness while retaining the heads critical for performance, i.e. language modeling capabilities. Our approach is practical in terms of time and resources, as it does not require fine-tuning the final pruned, and fairer, model. Our findings demonstrate a reduction in gender bias by 19%, 19.5%, 39.5%, 34.7%, 23%, and 8% for DistilGPT-2, GPT-2, GPT-Neo of two different sizes, GPT-J, and Llama 2 models, respectively, in comparison to the biased model, with only a slight decrease in performance. WARNING: This work uses language that is offensive in nature.

The widespread use of large language models has brought up essential questions about the potential biases these models might learn.This led to the development of several metrics aimed at evaluating and mitigating these biases.In this paper, we first demonstrate that prompt-based fairness metrics exhibit poor agreement, as measured by correlation, raising important questions about the reliability of fairness assessment using prompts.Then, we outline six relevant reasons why such a low correlation is observed across existing metrics.Based on these insights, we propose a method called Correlated Fairness Output (CAIRO) to enhance the correlation between fairness metrics.CAIRO augments the original prompts of a given fairness metric by using several pretrained language models and then selects the combination of the augmented prompts that achieves the highest correlation across metrics.We show a significant improvement in Pearson correlation from 0.3 and 0.18 to 0.90 and 0.98 across metrics for gender and religion biases, respectively.

Large language models (LLMs) have been rapidly adopted, as showcased by ChatGPT's overnight popularity, and are integrated in products used by millions of people every day, such as search engines and productivity suites. Yet the societal impact of LLMs, encompassing both benefits and harms, is not well understood. Inspired by cybersecurity practices, red-teaming is emerging as a technique to uncover model vulnerabilities. Despite increasing attention from industry, academia, and government centered around red-teaming LLMs, such efforts are still limited in the diversity of the red-teaming focus, approaches and participants. Importantly, given that LLMs are becoming ubiquitous, it is imperative that red-teaming efforts are scaled out to include large segments of the research, practitioners and the people whom are directly affected by the deployment of these systems. The goal of this tutorial is two fold. First, we introduce the topic of LLM red-teaming by reviewing the state of the art for red-teaming practices, from participatory events to automatic AI-focused approaches, exposing the gaps in both the techniques and coverage of the targeted harms. Second, we plan to engage the audience in a hands-on and interactive exercise in LLM red-teaming to showcase the ease (or difficulty) of exposing model vulnerabilities, contingent on both the targeted harm and model capabilities. We believe that the KDD community of researchers and practitioners are in a unique position to address the existing gaps in red-teaming approaches, given their longstanding research and practice of extracting knowledge from data.

To address the growing concern of unfairness in Artificial Intelligence (AI), several bias mitigation algorithms have been introduced in prior research. Their capabilities are often evaluated on certain overly-used datasets without rigorously stress-testing them under simultaneous train and test distribution shifts. To address this, we investigate the fairness vulnerabilities of these algorithms across several distribution shift scenarios using synthetic data, to highlight scenarios where these algorithms do and don’t work to encourage their trustworthy use. The paper makes three important contributions. Firstly, we propose a flexible pipeline called the Fairness Auditor to systematically stress-test bias mitigation algorithms using multiple synthetic datasets with shifts. Secondly, we introduce the Deviation Metric for measuring the fairness and utility performance of these algorithms under such shifts. Thirdly, we propose an interactive reporting tool for comparing algorithmic performance across various synthetic datasets, mitigation algorithms and metrics called the Fairness Report.

One way to help users navigate debated topics online is to apply stance detection in web search. Automatically identifying whether search results are against, neutral, or in favor could facilitate diversification efforts and support interventions that aim to mitigate cognitive biases. To be truly useful in this context, however, stance detection models not only need to make accurate (cross-topic) predictions but also be sufficiently explainable to users when applied to search results – an issue that is currently unclear. This paper presents a study into the feasibility of using current stance detection approaches to assist users in their web search on debated topics. We train and evaluate 10 stance detection models using a stance-annotated data set of 1204 search results. In a preregistered user study (N = 291), we then investigate the quality of stance detection explanations created using different explainability methods and explanation visualization techniques. The models we implement predict stances of search results across topics with satisfying quality (i.e., similar to the state-of-the-art for other data types). However, our results reveal stark differences in explanation quality (i.e., as measured by users’ ability to simulate model predictions and their attitudes towards the explanations) between different models and explainability methods. A qualitative analysis of textual user feedback further reveals potential application areas, user concerns, and improvement suggestions for such explanations. Our findings have important implications for the development of user-centered solutions surrounding web search on debated topics.

Machine learning (ML) models often perform differently under distribution shifts, in terms of utility, fairness, and other dimensions.We propose the Adversarial Auditor for measuring the utility and fairness performance of ML models under compound shifts of outcome and protected attributes.We use Multi-Objective Bayesian Optimization (MOBO) to account for multiple metrics and identify shifts where model performance is extreme, both good and bad.Using two case studies, we show that MOBO performed better than random and grid-based approaches in identifying scenarios by adversarially optimizing objectives, highlighting the value of such an auditor for developing fair, accurate and shift-robust models.

Clinical notes are becoming an increasingly important data source for machine learning (ML) applications in healthcare. Prior research has shown that deploying ML models can perpetuate existing biases against racial minorities, as bias can be implicitly embedded in data. In this study, we investigate the level of implicit race information available to ML models and human experts and the implications of model-detectable differences in clinical notes. Our work makes three key contributions. First, we find that models can identify patient self-reported race from clinical notes even when the notes are stripped of explicit indicators of race. Second, we determine that human experts are not able to accurately predict patient race from the same redacted clinical notes. Finally, we demonstrate the potential harm of this implicit information in a simulation study, and show that models trained on these race-redacted clinical notes can still perpetuate existing biases in clinical treatment decisions.

Thousands of scientific publications discuss evidence on the efficacy of non-cancer generic drugs being tested for cancer. However, trying to manually identify and extract such evidence is intractable at scale. We introduce a natural language processing pipeline to automate the identification of relevant studies and facilitate the extraction of therapeutic associations between generic drugs and cancers from PubMed abstracts. We annotate datasets of drug-cancer evidence and use them to train models to identify and characterize such evidence at scale. To make this evidence readily consumable, we incorporate the results of the models in a web application that allows users to browse documents and their extracted evidence. Users can provide feedback on the quality of the evidence extracted by our models. This feedback is used to improve our datasets and the corresponding models in a continuous integration system. We describe the natural language processing pipeline in our application and the steps required to deploy services based on the machine learning models.

More than 200 generic drugs approved by the U.S. Food and Drug Administration for non-cancer indications have shown promise for treating cancer. Due to their long history of safe patient use, low cost, and widespread availability, repurposing of these drugs represents a major opportunity to rapidly improve outcomes for cancer patients and reduce healthcare costs. In many cases, there is already evidence of efficacy for cancer, but trying to manually extract such evidence from the scientific literature is intractable. In this emerging applications paper, we introduce a system to automate non-cancer generic drug evidence extraction from PubMed abstracts. Our primary contribution is to define the natural language processing pipeline required to obtain such evidence, comprising the following modules: querying, filtering, cancer type entity extraction, therapeutic association classification, and study type classification. Using the subject matter expertise on our team, we create our own datasets for these specialized domain-specific tasks. We obtain promising performance in each of the modules by utilizing modern language processing techniques and plan to treat them as baseline approaches for future improvement of individual components.

In recent years there has been an increasing trend in which data scientists and domain experts work together to tackle complex scientific questions. However, such collaborations often face challenges. In this paper, we aim to decipher this collaboration complexity through a semi-structured interview study with 22 interviewees from teams of bio-medical scientists collaborating with data scientists. In the analysis, we adopt the Olsons' four-dimensions framework proposed in Distance Matters to code interview transcripts. Our findings suggest that besides the glitches in the collaboration readiness, technology readiness, and coupling of work dimensions, the tensions that exist in the common ground building process influence the collaboration outcomes, and then persist in the actual collaboration process. In contrast to prior works' general account of building a high level of common ground, the breakdowns of content common ground together with the strengthen of process common ground in this process is more beneficial for scientific discovery. We discuss why that is and what the design suggestions are, and conclude the paper with future directions and limitations.

Life on earth presents elegant solutions to many of the challenges innovators and entrepreneurs across disciplines face every day. To facilitate innovations inspired by nature, there is an emerging need for systems that bring relevant biological information to this application-oriented market. In this paper, we discuss our approach to assembling a system that uses machine learning techniques to assess a scientific article's potential usefulness to innovators, and classifies these articles in a way that helps innovators find information relevant to the challenges they are attempting to solve.

Your computer is continuously executing programs, but does it really understand them? Not in any meaningful sense. That burden falls upon human knowledge workers, who are increasingly asked to write and understand code. They would benefit greatly from intelligent tools that reveal the connections between their code and its subject matter. Towards this prospect, we present an AI system that forms semantic representations of computer programs, using techniques from knowledge representation and program analysis. These representations are created through a novel algorithm for the semantic enrichment of dataflow graphs. We illustrate its workings with examples from the field of data science. The algorithm is undergirded by a new ontology language for modeling computer programs and a new ontology about data science, written in this language.

Canary testing is an emerging technique that offers to minimize the risk of deploying a new version of software. It does so by slowly transferring load from the current to the new ("canary") version. As this ramp-up progresses, a human compares the performance and correctness of the two versions, and assesses whether to abort the canary version. For canary testing to be effective, a plethora of metrics must be analyzed, including CPU utilization and logged errors, across hundreds to thousands of machines. Performing this analysis manually is both time consuming and error prone. In this paper, we present CanaryAdvisor, a tool for automatic canary testing of cloud-based applications. CanaryAdvisor continuously monitors the deployed versions of an application and detects degradations in correctness, performance, and/or scalability. We describe our design and implementation of the CanaryAdvisor and outline open challenges.

This work revisits precomputation prefetching targeting long access latency loads with access patterns that are hard to predict. It presents Ekivolos, a precomputation prefetcher system that automatically builds prefetching slices that contain enough control flow instructions to faithfully and autonomously recreate the program's access behavior without inducing monitoring and execution overhead on the main thread. Ekivolos departs from the traditional notion of creating optimized short slices. In contrast, it shows that even longer slices can run ahead of the main thread and perform useful prefetches as long as they are sufficiently accurate. Ekivolos operates on arbitrary application binaries and takes advantage of the observed execution paths in creating its slices. On a set of emerging workloads Ekivolos outperforms three state-of-the-art hardware prefetchers and previously proposed precomputation-based prefetchers.

Graphics processing units (GPUs) can deliver considerable performance gains over general purpose processors. However, GPU performance improvement vary considerably across applications. Porting applications to GPUs by rewriting code with GPU-specific languages requires significant effort. In consequence, it is desirable to predict which applications would benefit most before porting to the GPU. This paper shows that machine learning techniques can build accurate predictive models for GPU acceleration. This study presents an approach which applies supervised learning algorithms to infer predictive models, based on dynamic profile data collected via instrumented runs on general purpose processors. For a set of 18 parallel benchmarks, the results show that a small set of easily-obtainable features can predict the magnitude of GPU speedups on two different high-end GPUs, with accuracies varying between 77% and 90%, depending on the prediction mechanism and scenario. For already-ported applications, similar models can predict the best device to run an application with an effective accuracy of 91%.

This paper describes the Liquid Metal entry in the 2013 ICFPT Design Competition. The Liquid Metal system provides a high-level language called Lime and a toolchain targeting FPGAs. Lime allowed us to use standard software development processes for programming, debugging, and performance tuning our FPGA design. We believe such iteration and refinement are far more challenging with low-level languages and design tools commonly used for FPGA development.