Unmasking Societal Biases in Respiratory Support for ICU Patients through Social Determinants of Health Mira Moukheiber 1, Lama Moukheiber 1, Dana Moukheiber 1 and Hyung-Chul Lee 2, 1 Massachusetts Institute of Technology 2 Seoul National University College of Medicine, Seoul National University Hospital, Department of Anesthesiology and Pain Medicine vital@snu.ac.kr Abstract In critical care settings, where precise and timely interventions are crucial for health outcomes, evaluating disparities in patient outcomes is important. Current approaches often fall short in comprehensively understanding and evaluating the impact of respiratory support interventions on individuals affected by social determinants of health. Attributes such as gender, race, and age are commonly assessed and essential, but provide only a partial view of the complexities faced by diverse populations. In this study, we focus on two clinically motivated tasks: prolonged mechanical ventilation and successful weaning. We also perform fairness audits on the models' predictions across demographic groups and social determinants of health to better understand the health inequities in respiratory interventions in the intensive care unit. We also release a temporal benchmark dataset, verified by clinical experts, to enable benchmarking of clinical respiratory intervention tasks. 1 Introduction Critically-ill patients often find themselves in the intensive care unit (ICU) seeking specialized support for respiratory distress [ Doyle et al., 1995; Ware and Matthay, 2000 ] . Despite advances in supportive treatments, the in-hospital mortality rate remains 40% for conditions such as acute lung injury and acute respiratory distress syndrome [ Rubenfeld et al., 2005; Sweatt and Levitt, 2014 ] .

Retrieval-augmented generation (RAG) has shown impressive capabilities in mitigating hallucinations in large language models (LLMs). However, LLMs struggle to handle misleading retrievals and often fail to maintain their own reasoning when exposed to conflicting or selectively-framed evidence, making them vulnerable to real-world misinformation. In such real-world retrieval scenarios, misleading and conflicting information is rampant, particularly in the political domain, where evidence is often selectively framed, incomplete, or polarized. However, existing RAG benchmarks largely assume a clean retrieval setting, where models succeed by accurately retrieving and generating answers from gold-standard documents. This assumption fails to align with real-world conditions, leading to an overestimation of RAG system performance. To bridge this gap, we introduce RAGuard, a fact-checking dataset designed to evaluate the robustness of RAG systems against misleading retrievals. Unlike prior benchmarks that rely on synthetic noise, our dataset constructs its retrieval corpus from Reddit discussions, capturing naturally occurring misinformation. It categorizes retrieved evidence into three types: supporting, misleading, and irrelevant, providing a realistic and challenging testbed for assessing how well RAG systems navigate different retrieval information. Our benchmark experiments reveal that when exposed to misleading retrievals, all tested LLM-powered RAG systems perform worse than their zero-shot baselines (i.e., no retrieval at all), highlighting their susceptibility to noisy environments. To the best of our knowledge, RAGuard is the first benchmark to systematically assess RAG robustness against misleading evidence. We expect this benchmark will drive future research toward improving RAG systems beyond idealized datasets, making them more reliable for real-world applications.

Humor is prevalent in online communications and it often relies on more than one modality (e.g., cartoons and memes). Interpreting humor in multimodal settings requires drawing on diverse types of knowledge, including metaphorical, sociocultural, and commonsense knowledge. However, identifying the most useful knowledge remains an open question. We introduce \method{}, a method inspired by the information bottleneck principle that elicits relevant world knowledge from vision and language models which is iteratively refined for generating an explanation of the humor in an unsupervised manner. Our experiments on three datasets confirm the advantage of our method over a range of baselines. Our method can further be adapted in the future for additional tasks that can benefit from eliciting and conditioning on relevant world knowledge and open new research avenues in this direction.

The current discourse on large language models (LLMs) and copyright largely takes a "behavioral" perspective, focusing on model outputs and evaluating whether they are substantially similar to training data. However, substantial similarity is difficult to define algorithmically and a narrow focus on model outputs is insufficient to address all copyright risks. In this interdisciplinary work, we take a complementary "structural" perspective and shift our focus to how LLMs are trained. We operationalize a notion of "fair learning" by measuring whether any training decision substantially affected the model's memorization. As a case study, we deconstruct Pythia, an open-source LLM, and demonstrate the use of causal and correlational analyses to make factual determinations about Pythia's training decisions. By proposing a legal standard for fair learning and connecting memorization analyses to this standard, we identify how judges may advance the goals of copyright law through adjudication. Finally, we discuss how a fair learning standard might evolve to enhance its clarity by becoming more rule-like and incorporating external technical guidelines.

Recent advancements in Neural Audio Codec (NAC) models have inspired their use in various speech processing tasks, including speech enhancement (SE). In this work, we propose a novel, efficient SE approach by leveraging the pre-quantization output of a pretrained NAC encoder. Unlike prior NAC-based SE methods, which process discrete speech tokens using Language Models (LMs), we perform SE within the continuous embedding space of the pretrained NAC, which is highly compressed along the time dimension for efficient representation. Our lightweight SE model, optimized through an embedding-level loss, delivers results comparable to SE baselines trained on larger datasets, with a significantly lower real-time factor of 0.005. Additionally, our method achieves a low GMAC of 3.94, reducing complexity 18-fold compared to Sepformer in a simulated cloud-based audio transmission environment. This work highlights a new, efficient NAC-based SE solution, particularly suitable for cloud applications where NAC is used to compress audio before transmission. Copyright 20XX IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

The success of large language models (LLMs) has attracted many individuals to fine-tune them for domain-specific tasks by uploading their data. However, in sensitive areas like healthcare and finance, privacy concerns often arise. One promising solution is to sample synthetic data with Differential Privacy (DP) guarantees to replace private data. However, these synthetic data contain significant flawed data, which are considered as noise. Existing solutions typically rely on naive filtering by comparing ROUGE-L scores or embedding similarities, which are ineffective in addressing the noise. To address this issue, we propose RewardDS, a novel privacy-preserving framework that fine-tunes a reward proxy model and uses reward signals to guide the synthetic data generation. Our RewardDS introduces two key modules, Reward Guided Filtering and Self-Optimizing Refinement, to both filter and refine the synthetic data, effectively mitigating the noise. Extensive experiments across medical, financial, and code generation domains demonstrate the effectiveness of our method.

Conversational agents are increasingly woven into individuals' personal lives, yet users often underestimate the privacy risks involved. The moment users share information with these agents (e.g., LLMs), their private information becomes vulnerable to exposure. In this paper, we characterize the notion of contextual privacy for user interactions with LLMs. It aims to minimize privacy risks by ensuring that users (sender) disclose only information that is both relevant and necessary for achieving their intended goals when interacting with LLMs (untrusted receivers). Through a formative design user study, we observe how even "privacy-conscious" users inadvertently reveal sensitive information through indirect disclosures. Based on insights from this study, we propose a locally-deployable framework that operates between users and LLMs, and identifies and reformulates out-of-context information in user prompts. Our evaluation using examples from ShareGPT shows that lightweight models can effectively implement this framework, achieving strong gains in contextual privacy while preserving the user's intended interaction goals through different approaches to classify information relevant to the intended goals.

Autonomous driving is at a critical stage in revolutionizing transportation systems and reshaping societal norms. More than 1,400 self-driving cars, trucks, and other vehicles are currently in operation or testing in the U.S. (Etherington 2019), and 4.5 million autonomous vehicles are expected to run on U.S. roads by 2030 (Meyer 2023). While autonomous driving is promising in improving traffic efficiency and personal mobility, safety is a prerequisite of all possible achievements and is becoming the first priority in practice (Du et al. 2020). In October 2023, Cruise, one of the leading autonomous driving companies, was ordered by California to stop operations of driverless cars in the state after one of Cruise's cars struck a pedestrian in San Francisco (Kerr 2023). The rare incident involved a woman who was first hit by a human driver and then thrown onto the road in front of a Cruise vehicle. The Cruise vehicle then rolled over the pedestrian and finally stopped on top of her, causing serious injuries. Such an accident reflects one of the greatest challenges in autonomous driving: the safety of an autonomous car is largely determined by the ability to detect and react to rare scenarios rather than common normal situations, which have been well considered during development. Although rare in a long-tailed distribution, unusual driving scenarios do happen and can have large impact on driving safety. To mitigate the impact of abnormal ego behaviors when outside the design domains, a detection system for anomalous driving scenarios is necessary, the output of which can be potentially used as a high-level decision for motion planning.

Large Language Models (LLMs) are revolutionizing information retrieval, with chatbots becoming an important source for answering user queries. As by their design, LLMs prioritize generating correct answers, the value of highly plausible yet incorrect answers (candidate answers) tends to be overlooked. However, such answers can still prove useful, for example, they can play a crucial role in tasks like Multiple-Choice Question Answering (MCQA) and QA Robustness Assessment (QARA). Existing QA datasets primarily focus on correct answers without explicit consideration of the plausibility of other candidate answers, limiting opportunity for more nuanced evaluations of models. To address this gap, we introduce PlausibleQA, a large-scale dataset comprising 10,000 questions and 100,000 candidate answers, each annotated with plausibility scores and justifications for their selection. Additionally, the dataset includes 900,000 justifications for pairwise comparisons between candidate answers, further refining plausibility assessments. We evaluate PlausibleQA through human assessments and empirical experiments, demonstrating its utility in MCQA and QARA analysis. Our findings show that plausibility-aware approaches are effective for MCQA distractor generation and QARA. We release PlausibleQA as a resource for advancing QA research and enhancing LLM performance in distinguishing plausible distractors from correct answers.

The recent application of LLMs to the legal field has spurred the creation of benchmarks across various jurisdictions and languages. However, no benchmark has yet been specifically designed for the Portuguese legal system. In this work, we present LegalBench.PT, the first comprehensive legal benchmark covering key areas of Portuguese law. To develop LegalBench.PT, we first collect long-form questions and answers from real law exams, and then use GPT-4o to convert them into multiple-choice, true/false, and matching formats. Once generated, the questions are filtered and processed to improve the quality of the dataset. To ensure accuracy and relevance, we validate our approach by having a legal professional review a sample of the generated questions. Although the questions are synthetically generated, we show that their basis in human-created exams and our rigorous filtering and processing methods applied result in a reliable benchmark for assessing LLMs' legal knowledge and reasoning abilities. Finally, we evaluate the performance of leading LLMs on LegalBench.PT and investigate potential biases in GPT-4o's responses. We also assess the performance of Portuguese lawyers on a sample of questions to establish a baseline for model comparison and validate the benchmark.