arXiv:2511.17644v1 Announce Type: new Abstract: Artificial intelligence deployed in risk-sensitive domains such as healthcare, finance, and security must not only achieve predictive accuracy but also ensure transparency, ethical alignment, and compliance with regulatory expectations. Hybrid neuro symbolic models combine the pattern-recognition strengths of neural networks with the interpretability and logical rigor of symbolic reasoning, making them well-suited for these contexts. This paper surveys hybrid architectures, ethical design considerations, and deployment patterns that balance accuracy with accountability. We highlight techniques for integrating knowledge graphs with deep inference, embedding fairness-aware rules, and generating human-readable explanations. Through case studies in healthcare decision support, financial risk management, and autonomous infrastructure, we show how hybrid systems can deliver reliable and auditable AI. Finally, we outline evaluation protocols and future directions for scaling neuro symbolic frameworks in complex, high stakes environments.

arXiv:2511.18298v1 Announce Type: new Abstract: The exponential growth of scientific knowledge has created significant barriers to cross-disciplinary knowledge discovery, synthesis and research collaboration. In response to this challenge, we present BioSage, a novel compound AI architecture that integrates LLMs with RAG, orchestrated specialized agents and tools to enable discoveries across AI, data science, biomedical, and biosecurity domains. Our system features several specialized agents including the retrieval agent with query planning and response synthesis that enable knowledge retrieval across domains with citation-backed responses, cross-disciplinary translation agents that align specialized terminology and methodologies, and reasoning agents that synthesize domain-specific insights with transparency, traceability and usability. We demonstrate the effectiveness of our BioSage system through a rigorous evaluation on scientific benchmarks (LitQA2, GPQA, WMDP, HLE-Bio) and introduce a new cross-modal benchmark for biology and AI, showing that our BioSage agents outperform vanilla and RAG approaches by 13\%-21\% powered by Llama 3.1. 70B and GPT-4o models. We perform causal investigations into compound AI system behavior and report significant performance improvements by adding RAG and agents over the vanilla models. Unlike other systems, our solution is driven by user-centric design principles and orchestrates specialized user-agent interaction workflows supporting scientific activities including but not limited to summarization, research debate and brainstorming. Our ongoing work focuses on multimodal retrieval and reasoning over charts, tables, and structured scientific data, along with developing comprehensive multimodal benchmarks for cross-disciplinary discovery. Our compound AI solution demonstrates significant potential for accelerating scientific advancement by reducing barriers between traditionally siloed domains.

arXiv:2511.19115v1 Announce Type: new Abstract: In AI, the existential risk denotes the hypothetical threat posed by an artificial system that would possess both the capability and the objective, either directly or indirectly, to eradicate humanity. This issue is gaining prominence in scientific debate due to recent technical advancements and increased media coverage. In parallel, AI progress has sparked speculation and studies about the potential emergence of artificial consciousness. The two questions, AI consciousness and existential risk, are sometimes conflated, as if the former entailed the latter. Here, I explain that this view stems from a common confusion between consciousness and intelligence. Yet these two properties are empirically and theoretically distinct. Arguably, while intelligence is a direct predictor of an AI system's existential threat, consciousness is not. There are, however, certain incidental scenarios in which consciousness could influence existential risk, in either direction. Consciousness could be viewed as a means towards AI alignment, thereby lowering existential risk; or, it could be a precondition for reaching certain capabilities or levels of intelligence, and thus positively related to existential risk. Recognizing these distinctions can help AI safety researchers and public policymakers focus on the most pressing issues.

arXiv:2511.17515v1 Announce Type: cross Abstract: Systems analysis students increasingly use Generative AI, yet current pedagogy lacks systematic approaches for teaching responsible AI orchestration that fosters critical thinking whilst meeting educational outcomes. Students risk accepting AI suggestions blindly or uncritically without assessing alignment with user needs or contextual appropriateness. SAGE (Structured AI-Guided Education) addresses this gap by embedding GenAI into curriculum design, training students when to accept, modify, or reject AI contributions. Implementation with 18 student groups across four Australian universities revealed how orchestration skills develop. Most groups (84\%) moved beyond passive acceptance, showing selective judgment, yet none proactively identified gaps overlooked by both human and AI analysis, indicating a competency ceiling. Students strong at explaining decisions also performed well at integrating sources, and those with deep domain understanding consistently considered accessibility considerations. Accessibility awareness proved fragile. When writing requirements, 85\% of groups explicitly considered elderly users and cultural needs. Notably, 55\% of groups struggled identifying when AI misclassified system boundaries (what belongs inside versus outside the system), 45\% missed data management errors (how information is stored and updated), and 55\% overlooked missing exception handling. Three implications emerge for educators: (i) require students to document why they accepted, modified, or rejected each AI suggestion, making reasoning explicit; (ii) embed accessibility prompts at each development stage because awareness collapses without continuous scaffolding; and (iii) have students create their own specifications before using AI, then compare versions, and anchor to research or standards to identify gaps.

arXiv:2511.17655v1 Announce Type: cross Abstract: Our study provides a full deep learning system for automated classification of brain tumors from MRI images, includes six benchmarked architectures (five ImageNet-pre-trained models (VGG-16, Inception V3, ResNet-50, Inception-ResNet V2, Xception) and a custom built, compact CNN (1.31M params)). The study moves the needle forward in a number of ways, including (1) full standardization of assessment with respect to preprocessing, training sets/protocols (optimizing networks with the AdamW optimizer, CosineAnnealingLR, patiene for early stopping = 7), and metrics to assess performance were identical along all models; (2) a high level of confidence in the localizations based on prior studies as both Grad-CAM and GradientShap explanation were used to establish anatomically important and meaningful attention regions and address the black-box issue; (3) a compact 1.31 million parameter CNN was developed that achieved 96.49% testing accuracy and was 100 times smaller than Inception-ResNet V2 while permitting real-time inference (375ms) on edge devices; (4) full evaluation beyond accuracy reporting based on measures of intersection over union, Hausdorff distance, and precision-recall curves, and confusion matrices across all splits. Inception-ResNet V2 reached state-of-the-art performance, achieving a 99.53% accuracy on testing and obtaining a precision, recall, and F1-score of at least 99.50% dominant performance based on metrics of recent studies. We demonstrated a lightweight model that is suitable to deploy on devices that do not have multi-GPU infrastructure in under-resourced settings. This end-to-end solution considers accuracy, interpretability, and deployability of trustworthy AI to create the framework necessary for performance assessment and deployment within advance and low-resource healthcare systems to an extent that enabled participation at the clinical screening and triage level.

arXiv:2511.17658v1 Announce Type: cross Abstract: As digital communication grows in importance when connecting with healthcare providers, traditional behavioral and content message features are imbued with renewed significance. If one is to meaningfully connect with them, it is crucial to understand what drives them to engage and respond. In this study, the authors analyzed several million text messages sent through the Impiricus platform to learn which factors influenced whether or not a doctor clicked on a link in a message. Several key insights came to light through the use of logistic regression, random forest, and neural network models, the details of which the authors discuss in this paper.

arXiv:2511.17669v1 Announce Type: cross Abstract: High-performance computing (HPC) and parallel computing increasingly rely on global collaboration among diverse teams, yet traditional computing curricula inadequately prepare students for cross-cultural teamwork essential in modern computational research environments. This paper presents Empa, an AI-powered virtual mentor that integrates intercultural collaboration training into undergraduate computing education. Built using large language models and deployed through a progressive web application, Empa guides students through structured activities covering cultural dimensions, communication styles, and conflict resolution that are critical for effective multicultural teamwork. Our system addresses the growing need for culturally competent HPC professionals by helping computing students develop skills to collaborate effectively in international research teams, contribute to global computational projects, and navigate the cultural complexities inherent in distributed computing environments. Pilot preparation for deployment in computing courses demonstrates the feasibility of AI-mediated intercultural training and provides insights into scalable approaches for developing intercultural collaboration skills essential for HPC workforce development.

arXiv:2511.17828v1 Announce Type: cross Abstract: Foundation models hold promise for specialized medical imaging tasks, though their effectiveness in breast imaging remains underexplored. This study leverages BiomedCLIP as a foundation model to address challenges in model generalization. BiomedCLIP was adapted for automated BI-RADS breast density classification using multi-modality mammographic data (synthesized 2D images, digital mammography, and digital breast tomosynthesis). Using 96,995 images, we compared single-modality (s2D only) and multi-modality training approaches, addressing class imbalance through weighted contrastive learning. Both approaches achieved similar accuracy (multi-modality: 0.74, single-modality: 0.73), with the multi-modality model offering broader applicability across different imaging modalities and higher AUC values consistently above 0.84 across BI-RADS categories. External validation on the RSNA and EMBED datasets showed strong generalization capabilities (AUC range: 0.80-0.93). GradCAM visualizations confirmed consistent and clinically relevant attention patterns, highlighting the models interpretability and robustness. This research underscores the potential of foundation models for breast imaging applications, paving the way for future extensions for diagnostic tasks.

arXiv:2511.17959v1 Announce Type: cross Abstract: As AI agents attempt to autonomously act on users' behalf, they raise transparency and control issues. We argue that permission-based access control is indispensable in providing meaningful control to the users, but conventional permission models are inadequate for the automated agentic execution paradigm. We therefore propose automated permission management for AI agents. Our key idea is to conduct a user study to identify the factors influencing users' permission decisions and to encode these factors into an ML-based permission management assistant capable of predicting users' future decisions. We find that participants' permission decisions are influenced by communication context but importantly individual preferences tend to remain consistent within contexts, and align with those of other participants. Leveraging these insights, we develop a permission prediction model achieving 85.1% accuracy overall and 94.4% for high-confidence predictions. We find that even without using permission history, our model achieves an accuracy of 66.9%, and a slight increase of training samples (i.e., 1-4) can substantially increase the accuracy by 10.8%.

arXiv:2511.18076v1 Announce Type: cross Abstract: This research proposes an enhancement to the innovative portfolio optimization approach using the G-Learning algorithm, combined with parametric optimization via the GIRL algorithm (G-learning approach to the setting of Inverse Reinforcement Learning) as presented by. The goal is to maximize portfolio value by a target date while minimizing the investor's periodic contributions. Our model operates in a highly volatile market with a well-diversified portfolio, ensuring a low-risk level for the investor, and leverages reinforcement learning to dynamically adjust portfolio positions over time. Results show that we improved the Sharpe Ratio from 0.42, as suggested by recent studies using the same approach, to a value of 0.483 a notable achievement in highly volatile markets with diversified portfolios. The comparison between G-Learning and GIRL reveals that while GIRL optimizes the reward function parameters (e.g., lambda = 0.0012 compared to 0.002), its impact on portfolio performance remains marginal. This suggests that reinforcement learning methods, like G-Learning, already enable robust optimization. This research contributes to the growing development of reinforcement learning applications in financial decision-making, demonstrating that probabilistic learning algorithms can effectively align portfolio management strategies with investor needs.

arXiv:2511.18084v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly adopted in clinical decision support, yet aligning them with the multifaceted reasoning pathways of real-world medicine remains a major challenge. Using more than 8,000 infertility treatment records, we systematically evaluate four alignment strategies: Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), Group Relative Policy Optimization (GRPO), and In-Context Learning (ICL) through a dual-layer framework combining automatic benchmarks with blinded doctor-in-the-loop assessments. GRPO achieves the highest algorithmic accuracy across multiple decision layers, confirming the value of reinforcement-based optimization for structured prediction tasks. However, clinicians consistently prefer the SFT model, citing clearer reasoning processes (p = 0.035) and higher therapeutic feasibility (p = 0.019). In blinded pairwise comparisons, SFT attains the highest winning rate (51.2%), outperforming both GRPO (26.2%) and even physicians' original decisions (22.7%). These results reveal an alignment paradox: algorithmic improvements do not necessarily translate into higher clinical trust, and may diverge from human-centered preferences. Our findings highlight the need for alignment strategies that prioritize clinically interpretable and practically feasible reasoning, rather than solely optimizing decision-level accuracy.

arXiv:2511.18274v1 Announce Type: cross Abstract: Digital health interventions are increasingly used in physical and occupational therapy to deliver home exercise programs via sensor equipped devices such as smartphones, enabling remote monitoring of adherence and performance. However, digital interventions are typically programmed as software before clinical encounters as libraries of parametrized exercise modules targeting broad patient populations. At the point of care, clinicians can only select modules and adjust a narrow set of parameters like repetitions, so patient specific needs that emerge during encounters, such as distinct movement limitations, and home environments, are rarely reflected in the software. We evaluated a digital intervention paradigm that uses large language models (LLMs) to translate clinicians' exercise prescriptions into intervention software. In a prospective single arm feasibility study with 20 licensed physical and occupational therapists and a standardized patient, clinicians created 40 individualized upper extremity programs (398 instructions) that were automatically translated into executable software. Our results show a 45% increase in the proportion of personalized prescriptions that can be implemented as software compared with a template based benchmark, with unanimous consensus among therapists on ease of use. The LLM generated software correctly delivered 99.78% (397/398) of instructions as prescribed and monitored performance with 88.4% (352/398) accuracy, with 90% (18/20) of therapists judged it safe to interact with patients, and 75% (15/20) expressed willingness to adopt it. To our knowledge, this is the first prospective evaluation of clinician directed intervention software generation with LLMs in healthcare, demonstrating feasibility and motivating larger trials to assess clinical effectiveness and safety in real patient populations.

arXiv:2511.18334v1 Announce Type: cross Abstract: Urinary tract infection (UTI) flare-ups pose a significant health risk for older adults with chronic conditions. These infections often go unnoticed until they become severe, making early detection through innovative smart home technologies crucial. Traditional machine learning (ML) approaches relying on simple binary classification for UTI detection offer limited utility to nurses and practitioners as they lack insight into prediction uncertainty, hindering informed clinical decision-making. This paper presents a clinician-in-the-loop (CIL) smart home system that leverages ambient sensor data to extract meaningful behavioral markers, train robust predictive ML models, and calibrate them to enable uncertainty-aware decision support. The system incorporates a statistically valid uncertainty quantification method called Conformal-Calibrated Interval (CCI), which quantifies uncertainty and abstains from making predictions ("I don't know") when the ML model's confidence is low. Evaluated on real-world data from eight smart homes, our method outperforms baseline methods in recall and other classification metrics while maintaining the lowest abstention proportion and interval width. A survey of 42 nurses confirms that our system's outputs are valuable for guiding clinical decision-making, underscoring their practical utility in improving informed decisions and effectively managing UTIs and other condition flare-ups in older adults.

arXiv:2511.18622v1 Announce Type: cross Abstract: We present OpenGloss, a synthetic encyclopedic dictionary and semantic knowledge graph for English that integrates lexicographic definitions, encyclopedic context, etymological histories, and semantic relationships in a unified resource. OpenGloss contains 537K senses across 150K lexemes, on par with WordNet 3.1 and Open English WordNet, while providing more than four times as many sense definitions. These lexemes include 9.1M semantic edges, 1M usage examples, 3M collocations, and 60M words of encyclopedic content. Generated through a multi-agent procedural generation pipeline with schema-validated LLM outputs and automated quality assurance, the entire resource was produced in under one week for under $1,000. This demonstrates that structured generation can create comprehensive lexical resources at cost and time scales impractical for manual curation, enabling rapid iteration as foundation models improve. The resource addresses gaps in pedagogical applications by providing integrated content -- definitions, examples, collocations, encyclopedias, etymology -- that supports both vocabulary learning and natural language processing tasks. As a synthetically generated resource, OpenGloss reflects both the capabilities and limitations of current foundation models. The dataset is publicly available on Hugging Face under CC-BY 4.0, enabling researchers and educators to build upon and adapt this resource.

arXiv:2511.18635v1 Announce Type: cross Abstract: Large Language Models (LLMs) inherit societal biases from their training data, potentially leading to harmful or unfair outputs. While various techniques aim to mitigate these biases, their effects are often evaluated only along the dimension of the bias being targeted. This work investigates the cross-category consequences of targeted bias mitigation. We study four bias mitigation techniques applied across ten models from seven model families, and we explore racial, religious, profession- and gender-related biases. We measure the impact of debiasing on model coherence and stereotypical preference using the StereoSet benchmark. Our results consistently show that while targeted mitigation can sometimes reduce bias in the intended dimension, it frequently leads to unintended and often negative consequences in others, such as increasing model bias and decreasing general coherence. These findings underscore the critical need for robust, multi-dimensional evaluation tools when examining and developing bias mitigation strategies to avoid inadvertently shifting or worsening bias along untargeted axes.

arXiv:2511.18640v1 Announce Type: cross Abstract: Frontier artificial intelligence (AI) models, such as OpenAI's GPT-5 and Meta's DINOv3, have advanced rapidly through training on internet-scale public data, yet such systems lack access to private clinical data. Neuroimaging, in particular, is underrepresented in the public domain due to identifiable facial features within MRI and CT scans, fundamentally restricting model performance in clinical medicine. Here, we show that frontier models underperform on neuroimaging tasks and that learning directly from uncurated data generated during routine clinical care at health systems, a paradigm we call health system learning, yields high-performance, generalist neuroimaging models. We introduce NeuroVFM, a visual foundation model trained on 5.24 million clinical MRI and CT volumes using a scalable volumetric joint-embedding predictive architecture. NeuroVFM learns comprehensive representations of brain anatomy and pathology, achieving state-of-the-art performance across multiple clinical tasks, including radiologic diagnosis and report generation. The model exhibits emergent neuroanatomic understanding and interpretable visual grounding of diagnostic findings. When paired with open-source language models through lightweight visual instruction tuning, NeuroVFM generates radiology reports that surpass frontier models in accuracy, clinical triage, and expert preference. Through clinically grounded visual understanding, NeuroVFM reduces hallucinated findings and critical errors, offering safer clinical decision support. These results establish health system learning as a paradigm for building generalist medical AI and provide a scalable framework for clinical foundation models.

arXiv:2511.19220v1 Announce Type: cross Abstract: Large vision language models (VLMs) have achieved impressive performance on medical visual question answering benchmarks, yet their reliance on visual information remains unclear. We investigate whether frontier VLMs demonstrate genuine visual grounding when answering Italian medical questions by testing four state-of-the-art models: Claude Sonnet 4.5, GPT-4o, GPT-5-mini, and Gemini 2.0 flash exp. Using 60 questions from the EuropeMedQA Italian dataset that explicitly require image interpretation, we substitute correct medical images with blank placeholders to test whether models truly integrate visual and textual information. Our results reveal striking variability in visual dependency: GPT-4o shows the strongest visual grounding with a 27.9pp accuracy drop (83.2% [74.6%, 91.7%] to 55.3% [44.1%, 66.6%]), while GPT-5-mini, Gemini, and Claude maintain high accuracy with modest drops of 8.5pp, 2.4pp, and 5.6pp respectively. Analysis of model-generated reasoning reveals confident explanations for fabricated visual interpretations across all models, suggesting varying degrees of reliance on textual shortcuts versus genuine visual analysis. These findings highlight critical differences in model robustness and the need for rigorous evaluation before clinical deployment.

Background: Preventive strategies integrated with digital health and artificial intelligence (AI), have significant potential to mitigate the global burden of atherosclerotic cardiovascular disease (ASCVD). AI-enabled clinical decision support (CDS) systems increasingly provide patient-specific insights beyond traditional risk factors. Despite these advances, their capacity to enhance clinical decision-making in resource-constrained settings remains largely unexplored. Objective: We conducted a randomised controlled study to assess the effect of AI-based CDS on 10-year ASCVD risk assessment and management in primary prevention. Methods: In a three-way within-subject randomised design, doctors completed nine clinical vignettes representative of primary care presentations in a resource-constrained outpatient setting. For each vignette, participants assessed 10-year ASCVD risk and made management decisions using either a conceptual prototype of AI-based CDS, automated CDS, or no decision support. The conceptual prototype represented contemporary risk calculators based on traditional machine learning models (e.g., random forest, neural networks, logistic regression) that incorporate additional predictors alongside traditional risk factors. Primary outcomes were correct risk assessment and patient management (prescription of aspirin, statins, and anti-hypertensives; referral for advanced examinations). Decision-making time and perceptions about AI utility were also measured. Results: 102 doctors from all seven geographical regions of Indonesia participated. Most participants were 26–35 years (83%), 56% male, with a median of six years of clinical experience (IQR=4.75). AI-based CDS improved risk assessment by 27% (2(2, n=102) = 48.875, P<.001 compared to unassisted or one additional correct risk classification for every patients where doctors use ai needed treat nnt="3.7;" ci the prescription of statins also improved by n="102)=" p in pairwise comparisons assisted with ai-based cds correctly assessed significantly more cases adjusted and prescribed appropriate statin often medium effect size r=".35)" control. ai-assisted required less time marginal means sec vs f however improvements aspirin anti-hypertensives did not reach statistical significance. no improvement was observed referral decisions. participants generally viewed positively agreeing strongly that they would follow its recommendations indicating it if given access. believed could enhance efficiency assessment particularly high-volume primary care settings while noting need verify against clinical guidelines each patient. conclusions: coupled reduced decision-making highlight potential utility ascvd resource-constrained efficient healthcare resources is crucial. further research ascertain whether this online study translate real-world low-resource settings.>

Science, Ahead of Print.

Nature, Published online: 25 November 2025; doi:10.1038/d41586-025-03856-1

Blending ‘old-fashioned’ logic systems with the neural networks that power large language models is one of the hottest trends in artificial intelligence.