arXiv:2512.22334v1 Announce Type: new Abstract: We introduce SciEvalKit, a unified benchmarking toolkit designed to evaluate AI models for science across a broad range of scientific disciplines and task capabilities. Unlike general-purpose evaluation platforms, SciEvalKit focuses on the core competencies of scientific intelligence, including Scientific Multimodal Perception, Scientific Multimodal Reasoning, Scientific Multimodal Understanding, Scientific Symbolic Reasoning, Scientific Code Generation, Science Hypothesis Generation and Scientific Knowledge Understanding. It supports six major scientific domains, spanning from physics and chemistry to astronomy and materials science. SciEvalKit builds a foundation of expert-grade scientific benchmarks, curated from real-world, domain-specific datasets, ensuring that tasks reflect authentic scientific challenges. The toolkit features a flexible, extensible evaluation pipeline that enables batch evaluation across models and datasets, supports custom model and dataset integration, and provides transparent, reproducible, and comparable results. By bridging capability-based evaluation and disciplinary diversity, SciEvalKit offers a standardized yet customizable infrastructure to benchmark the next generation of scientific foundation models and intelligent agents. The toolkit is open-sourced and actively maintained to foster community-driven development and progress in AI4Science.

arXiv:2512.23184v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used to simulate human behavior, but common practices to use LLM-generated data are inefficient. Treating an LLM's output ("model choice") as a single data point underutilizes the information inherent to the probabilistic nature of LLMs. This paper introduces and formalizes "model belief," a measure derived from an LLM's token-level probabilities that captures the model's belief distribution over choice alternatives in a single generation run. The authors prove that model belief is asymptotically equivalent to the mean of model choices (a non-trivial property) but forms a more statistically efficient estimator, with lower variance and a faster convergence rate. Analogous properties are shown to hold for smooth functions of model belief and model choice often used in downstream applications. The authors demonstrate the performance of model belief through a demand estimation study, where an LLM simulates consumer responses to different prices. In practical settings with limited numbers of runs, model belief explains and predicts ground-truth model choice better than model choice itself, and reduces the computation needed to reach sufficiently accurate estimates by roughly a factor of 20. The findings support using model belief as the default measure to extract more information from LLM-generated data.

arXiv:2512.23545v1 Announce Type: cross Abstract: Recent pathological foundation models have substantially advanced visual representation learning and multimodal interaction. However, most models still rely on a static inference paradigm in which whole-slide images are processed once to produce predictions, without reassessment or targeted evidence acquisition under ambiguous diagnoses. This contrasts with clinical diagnostic workflows that refine hypotheses through repeated slide observations and further examination requests. We propose PathFound, an agentic multimodal model designed to support evidence-seeking inference in pathological diagnosis. PathFound integrates the power of pathological visual foundation models, vision-language models, and reasoning models trained with reinforcement learning to perform proactive information acquisition and diagnosis refinement by progressing through the initial diagnosis, evidence-seeking, and final decision stages. Across several large multimodal models, adopting this strategy consistently improves diagnostic accuracy, indicating the effectiveness of evidence-seeking workflows in computational pathology. Among these models, PathFound achieves state-of-the-art diagnostic performance across diverse clinical scenarios and demonstrates strong potential to discover subtle details, such as nuclear features and local invasions.

Background: Digital therapeutics (DTx) for children and adolescents with mental health problems have been developed in the health care industry. Despite reports of side effects from DTx for children and adolescents, there have been no guidelines to address the prevention of DTx overdependence among young users. Objective: This study aimed to identify the requirements for guidelines to prevent DTx overdependence in children and adolescents and to develop and evaluate these guidelines. Methods: We conducted 2 phases. This study first involved a phase I survey to develop guidelines, including assessments of smartphone usage and mental health conditions. The second phase evaluated the guidelines’ effectiveness, reliability, necessity, and satisfaction using a visual analog scale through a randomized controlled trial. Participants—45 children and adolescents aged 9-16 years and 42 caregivers—were randomly assigned to the experimental and control groups. Results: Phase I revealed that blocking mobile applications and notifications (mean 8.5, SD 1.8) and parental monitoring (mean 8.5, SD 2.1) were effective preventive features. Caregivers, children, and adolescents expressed concerns about the side effects and overdependence of DTx and decreased effects due to nonindividualized guidelines in subjective responses to the phase I survey. Based on these insights, personalized guidelines for phase II were developed, in which overall mean visual analog scale scores for guideline evaluation were higher in the experimental group, except for necessity among caregivers (mean 8.5, SD 1.3 versus mean 8.7, SD 1.2). Conclusions: Both caregivers and children and adolescents demonstrated the need for guidelines to prevent overdependence on DTx distinct from smartphone usage. Tailored guidelines may be acceptable for use in real-world therapeutic protocols. Guidelines to prevent overdependence on DTx in children and adolescents and to achieve a balance between their benefits and risks need to be established. Trial Registration: Clinical Research Information Service (CRiS) of the Republic of Korea KCT0008893; https://cris.nih.go.kr/cris/search/detailSearch.do?seq=25609

npj Digital Medicine, Published online: 26 December 2025; doi:10.1038/s41746-025-02277-8

A novel evaluation benchmark for medical LLMs illuminating safety and effectiveness in clinical domains

arXiv:2512.16856v1 Announce Type: new Abstract: AI safety and alignment research has predominantly been focused on methods for safeguarding individual AI systems, resting on the assumption of an eventual emergence of a monolithic Artificial General Intelligence (AGI). The alternative AGI emergence hypothesis, where general capability levels are first manifested through coordination in groups of sub-AGI individual agents with complementary skills and affordances, has received far less attention. Here we argue that this patchwork AGI hypothesis needs to be given serious consideration, and should inform the development of corresponding safeguards and mitigations. The rapid deployment of advanced AI agents with tool-use capabilities and the ability to communicate and coordinate makes this an urgent safety consideration. We therefore propose a framework for distributional AGI safety that moves beyond evaluating and aligning individual agents. This framework centers on the design and implementation of virtual agentic sandbox economies (impermeable or semi-permeable), where agent-to-agent transactions are governed by robust market mechanisms, coupled with appropriate auditability, reputation management, and oversight to mitigate collective risks.

arXiv:2512.15769v1 Announce Type: cross Abstract: The increasing use of generative models such as diffusion models for synthetic data augmentation has greatly reduced the cost of data collection and labeling in downstream perception tasks. However, this new data source paradigm may introduce important security concerns. This work investigates backdoor propagation in such emerging generative data supply chains, namely Data-Chain Backdoor (DCB). Specifically, we find that open-source diffusion models can become hidden carriers of backdoors. Their strong distribution-fitting ability causes them to memorize and reproduce backdoor triggers during generation, which are subsequently inherited by downstream models, resulting in severe security risks. This threat is particularly concerning under clean-label attack scenarios, as it remains effective while having negligible impact on the utility of the synthetic data. Furthermore, we discover an Early-Stage Trigger Manifestation (ESTM) phenomenon: backdoor trigger patterns tend to surface more explicitly in the early, high-noise stages of the diffusion model's reverse generation process before being subtly integrated into the final samples. Overall, this work reveals a previously underexplored threat in generative data pipelines and provides initial insights toward mitigating backdoor risks in synthetic data generation.

arXiv:2512.16455v1 Announce Type: cross Abstract: In this paper, we describe a federated compute platform dedicated to support Artificial Intelligence in scientific workloads. Putting the effort into reproducible deployments, it delivers consistent, transparent access to a federation of physically distributed e-Infrastructures. Through a comprehensive service catalogue, the platform is able to offer an integrated user experience covering the full Machine Learning lifecycle, including model development (with dedicated interactive development environments), training (with GPU resources, annotation tools, experiment tracking, and federated learning support) and deployment (covering a wide range of deployment options all along the Cloud Continuum). The platform also provides tools for traceability and reproducibility of AI models, integrates with different Artificial Intelligence model providers, datasets and storage resources, allowing users to interact with the broader Machine Learning ecosystem. Finally, it is easily customizable to lower the adoption barrier by external communities.

arXiv:2512.09566v2 Announce Type: replace Abstract: Drug discovery is a time-consuming and expensive process, with traditional high-throughput and docking-based virtual screening hampered by low success rates and limited scalability. Recent advances in generative modelling, including autoregressive, diffusion, and flow-based approaches, have enabled de novo ligand design beyond the limits of enumerative screening. Yet these models often suffer from inadequate generalization, limited interpretability, and an overemphasis on binding affinity at the expense of key pharmacological properties, thereby restricting their translational utility. Here we present Trio, a molecular generation framework integrating fragment-based molecular language modeling, reinforcement learning, and Monte Carlo tree search, for effective and interpretable closed-loop targeted molecular design. Through the three key components, Trio enables context-aware fragment assembly, enforces physicochemical and synthetic feasibility, and guides a balanced search between the exploration of novel chemotypes and the exploitation of promising intermediates within protein binding pockets. Experimental results show that Trio reliably achieves chemically valid and pharmacologically enhanced ligands, outperforming state-of-the-art approaches with improved binding affinity (+7.85%), drug-likeness (+11.10%) and synthetic accessibility (+12.05%), while expanding molecular diversity more than fourfold. By combining generalization, plausibility, and interpretability, Trio establishes a closed-loop generative paradigm that redefines how chemical space can be navigated, offering a transformative foundation for the next era of AI-driven drug discovery.

arXiv:2511.07392v3 Announce Type: replace-cross Abstract: In robotic surgery, surgeons fully engage their hands and visual attention in procedures, making it difficult to access and manipulate multimodal patient data without interrupting the workflow. To overcome this problem, we propose a Voice-Interactive Surgical Agent (VISA) built on a hierarchical multi-agent framework consisting of an orchestration agent and three task-specific agents driven by Large Language Models (LLMs). These LLM-based agents autonomously plan, refine, validate, and reason to interpret voice commands and execute tasks such as retrieving clinical information, manipulating CT scans, or navigating 3D anatomical models within surgical video. We construct a dataset of 240 user commands organized into hierarchical categories and introduce the Multi-level Orchestration Evaluation Metric (MOEM) that evaluates the performance and robustness at both the command and category levels. Experimental results demonstrate that VISA achieves high stage-level accuracy and workflow-level success rates, while also enhancing its robustness by correcting transcription errors, resolving linguistic ambiguity, and interpreting diverse free-form expressions. These findings highlight the strong potential of VISA to support robotic surgery and its scalability for integrating new functions and agents.

arXiv:2512.01241v2 Announce Type: replace-cross Abstract: Large language models (LLMs) are routinely used by physicians and patients for medical advice, yet their clinical safety profiles remain poorly characterized. We present NOHARM (Numerous Options Harm Assessment for Risk in Medicine), a benchmark using 100 real primary care-to-specialist consultation cases to measure frequency and severity of harm from LLM-generated medical recommendations. NOHARM covers 10 specialties, with 12,747 expert annotations for 4,249 clinical management options. Across 31 LLMs, potential for severe harm from LLM recommendations occurs in up to 22.2% (95% CI 21.6-22.8%) of cases, with harm of omission accounting for 76.6% (95% CI 76.4-76.8%) of errors. Safety performance is only moderately correlated (r = 0.61-0.64) with existing AI and medical knowledge benchmarks. The best models outperform generalist physicians on safety (mean difference 9.7%, 95% CI 7.0-12.5%), and a diverse multi-agent approach improves safety compared to solo models (mean difference 8.0%, 95% CI 4.0-12.1%). Therefore, despite strong performance on existing evaluations, widely used AI models can produce severely harmful medical advice at nontrivial rates, underscoring clinical safety as a distinct performance dimension necessitating explicit measurement.

arXiv:2512.13716v1 Announce Type: new Abstract: Personalized decision-making is essential for human-AI interaction, enabling AI agents to act in alignment with individual users' value preferences. As AI systems expand into real-world applications, adapting to personalized values beyond task completion or collective alignment has become a critical challenge. We address this by proposing a value-driven approach to personalized decision-making. Human values serve as stable, transferable signals that support consistent and generalizable behavior across contexts. Compared to task-oriented paradigms driven by external rewards and incentives, value-driven decision-making enhances interpretability and enables agents to act appropriately even in novel scenarios. We introduce ValuePilot, a two-phase framework consisting of a dataset generation toolkit (DGT) and a decision-making module (DMM). DGT constructs diverse, value-annotated scenarios from a human-LLM collaborative pipeline. DMM learns to evaluate actions based on personal value preferences, enabling context-sensitive, individualized decisions. When evaluated on previously unseen scenarios, DMM outperforms strong LLM baselines, including GPT-5, Claude-Sonnet-4, Gemini-2-flash, and Llama-3.1-70b, in aligning with human action choices. Our results demonstrate that value-driven decision-making is an effective and extensible engineering pathway toward building interpretable, personalized AI agents.

arXiv:2512.14329v1 Announce Type: cross Abstract: Dynamic prediction of locomotor capacity after stroke is crucial for tailoring rehabilitation, yet current assessments provide only static impairment scores and do not indicate whether patients can safely perform specific tasks such as slope walking or stair climbing. Here, we develop a data-physics hybrid generative framework that reconstructs an individual stroke survivor's neuromuscular control from a single 20 m level-ground walking trial and predicts task-conditioned locomotion across rehabilitation scenarios. The system combines wearable-sensor kinematics, a proportional-derivative physics controller, a population Healthy Motion Atlas, and goal-conditioned deep reinforcement learning with behaviour cloning and generative adversarial imitation learning to generate physically plausible, patient-specific gait simulations for slopes and stairs. In 11 stroke survivors, the personalized controllers preserved idiosyncratic gait patterns while improving joint-angle and endpoint fidelity by 4.73% and 12.10%, respectively, and reducing training time to 25.56% relative to a physics-only baseline. In a multicentre pilot involving 21 inpatients, clinicians who used our locomotion predictions to guide task selection and difficulty obtained larger gains in Fugl-Meyer lower-extremity scores over 28 days of standard rehabilitation than control clinicians (mean change 6.0 versus 3.7 points). These findings indicate that our generative, task-predictive framework can augment clinical decision-making in post-stroke gait rehabilitation and provide a template for dynamically personalized motor recovery strategies.

arXiv:2410.07553v5 Announce Type: replace Abstract: The rapid advances of multimodal agents built on large foundation models have largely overlooked their potential for language-based communication between agents in collaborative tasks. This oversight presents a critical gap in understanding their effectiveness in real-world deployments, particularly when communicating with humans. Existing agentic benchmarks fail to address key aspects of inter-agent communication and collaboration, particularly in scenarios where agents have unequal access to information and must work together to achieve tasks beyond the scope of individual capabilities. To fill this gap, we introduce COMMA: a novel puzzle benchmark designed to evaluate the collaborative performance of multimodal multi-agent systems through language communication. Our benchmark features a variety of multimodal puzzles, providing a comprehensive evaluation across four key categories of agentic capability in a communicative collaboration setting. Our findings reveal surprising weaknesses in state-of-the-art models, including strong proprietary models like GPT-4o and reasoning models like o4-mini. Many chain of thought reasoning models such as R1-Onevision and LLaVA-CoT struggle to outperform even a random baseline in agent-agent collaboration, indicating a potential growth area in their communication abilities.

arXiv:2508.15811v2 Announce Type: replace-cross Abstract: Generative query suggestion using large language models offers a powerful way to enhance conversational systems, but aligning outputs with nuanced user preferences remains a critical challenge. To address this, we introduce a multi-stage framework designed for progressive alignment between the generation policy and user intent. Our pipeline begins with prompt engineering as a cold-start strategy, followed by the Supervised Fine-Tuning stage, in which we introduce a distillation method on click logs to create a robust foundational model. To better model user preferences while capturing their inherent uncertainty, we develop a Gaussian Reward Model (GaRM) that represents user preferences as probability distributions rather than point estimates. Finally, we employ reinforcement learning to align the generation policy with these preferences, guided by a composite reward function that integrates GaRM with auxiliary heuristics to mitigate reward hacking. To maintain training stability, this process is enhanced by a novel out-of-distribution regularization method and a two-stage reward fusion technique. Extensive experiments demonstrate that our framework significantly outperforms baselines on both automatic and human evaluations and yields a 34\% relative increase in user engagement as measured by click-through rate in live A/B tests.

arXiv:2510.02967v3 Announce Type: replace-cross Abstract: This paper presents the development and evaluation of a Retrieval-Augmented Generation (RAG) system for querying the United Kingdom's National Institute for Health and Care Excellence (NICE) clinical guidelines using Large Language Models (LLMs). The extensive length and volume of these guidelines can impede their utilisation within a time-constrained healthcare system, a challenge this project addresses through the creation of a system capable of providing users with precisely matched information in response to natural language queries. The system's retrieval architecture, composed of a hybrid embedding mechanism, was evaluated against a corpus of 10,195 text chunks derived from three hundred guidelines. It demonstrates high performance, with a Mean Reciprocal Rank (MRR) of 0.814, a Recall of 81% at the first chunk and of 99.1% within the top ten retrieved chunks, when evaluated on 7901 queries. The most significant impact of the RAG system was observed during the generation phase. When evaluated on a manually curated dataset of seventy question-answer pairs, RAG-enhanced models showed substantial gains in performance. Faithfulness, the measure of whether an answer is supported by the source text, was increased by 64.7 percentage points to 99.5% for the RAG-enhanced O4-Mini model and significantly outperformed the medical-focused Meditron3-8B LLM, which scored 43%. Clinical evaluation by seven Subject Matter Experts (SMEs) further validated these findings, with GPT-4.1 achieving 98.7% accuracy while reducing unsafe responses by 67% compared to O4-Mini (from 3.0 to 1.0 per evaluator). This study thus establishes RAG as an effective, reliable, and scalable approach for applying generative AI in healthcare, enabling cost-effective access to medical guidelines.

npj Digital Medicine, Published online: 15 December 2025; doi:10.1038/s41746-025-02218-5

H&E-based MSI/MMR testing with AI in colorectal cancer: a multi-centred blinded evaluation

arXiv:2512.11661v1 Announce Type: cross Abstract: Large Language Models (LLMs) are increasingly embedded in academic writing practices. Although numerous studies have explored how researchers employ these tools for scientific writing, their concrete implementation, limitations, and design challenges within the literature review process remain underexplored. In this paper, we report a user study with researchers across multiple disciplines to characterize current practices, benefits, and \textit{pain points} in using LLMs to investigate related work. We identified three recurring gaps: (i) lack of trust in outputs, (ii) persistent verification burden, and (iii) requiring multiple tools. This motivates our proposal of six design goals and a high-level framework that operationalizes them through improved related papers visualization, verification at every step, and human-feedback alignment with generation-guided explanations. Overall, by grounding our work in the practical, day-to-day needs of researchers, we designed a framework that addresses these limitations and models real-world LLM-assisted writing, advancing trust through verifiable actions and fostering practical collaboration between researchers and AI systems.

Background: Computer perception (CP) technologies hold significant promise for advancing precision mental health care systems, given their ability to leverage algorithmic analysis of continuous, passive sensing data from wearables and smartphones (eg, behavioral activity, geolocation, vocal features, and ambient environmental data) to infer clinically meaningful behavioral and physiological states. However, successful implementation critically depends on cultivating well-founded stakeholder trust. Objective: This study aims to investigate, across adolescents, caregivers, clinicians, and developers, the contingencies under which CP technologies are deemed trustworthy in health care. Methods: We conducted 80 semistructured interviews with a purposive sample of adolescents (n=20) diagnosed with autism, Tourette syndrome, anxiety, obsessive-compulsive disorder, or attention-deficit/hyperactivity disorder and their caregivers (n=20); practicing clinicians across psychiatry, psychology, and pediatrics (n=20); and CP system developers (n=20). Interview transcripts were coded by 2 independent coders and analyzed using multistage, inductive thematic content analysis to identify prominent themes. Results: Across stakeholder groups, 5 core criteria emerged as prerequisites for trust in CP outputs: (1) epistemic alignment—consistency between system outputs, personal experience, and existing diagnostic frameworks; (2) demonstrable rigor—training on representative data and validation in real-world contexts; (3) explainability—transparent communication of input variables, thresholds, and decision logic; (4) sensitivity to complexity—the capacity to accommodate heterogeneity and comorbidity in symptom expression; and (5) a nonsubstitutive role—technologies must augment, rather than supplant, clinical judgment. A novel and cautionary finding was that epistemic alignment—whether outputs affirmed participants’ preexisting beliefs, diagnostic expectations, or internal states—was a dominant factor in determining whether the tool was perceived as trustworthy. Participants also expressed relational trust, placing confidence in CP systems based on endorsements from respected peers, academic institutions, or regulatory agencies. However, both trust strategies raise significant concerns: confirmation bias may lead users to overvalue outputs that align with their assumptions, while surrogate trust may be misapplied in the absence of robust performance validation. Conclusions: This study advances empirical understanding of how trust is formed and calibrated around artificial intelligence–based CP technologies. While trust is commonly framed as a function of technical performance, our findings show that it is deeply shaped by cognitive heuristics, social relationships, and alignment with entrenched epistemologies. These dynamics can facilitate intuitive verification but may also constrain the transformative potential of CP systems by reinforcing existing beliefs. To address this, we recommend a dual strategy: (1) embedding CP tools within institutional frameworks that uphold rigorous validation, ethical oversight, and transparent design; and (2) providing clinicians with training and interface designs that support critical appraisal and minimize susceptibility to cognitive bias. Recalibrating trust to reflect actual system capacities—rather than familiarity or endorsement—is essential for ethically sound and clinically meaningful integration of CP technologies.