The vast majority of Web pages fail to comply with established Web accessibility guidelines, excluding a range of users with diverse abilities from interacting with their content. Making Web pages accessible to all users requires dedicated expertise and additional manual efforts from Web page providers. To lower their efforts and promote inclusiveness, we aim to automatically detect and correct Web accessibility violations in HTML code. While previous work has made progress in detecting certain types of accessibility violations, the problem of automatically detecting and correcting accessibility violations remains an open challenge that we address. We introduce a novel taxonomy classifying Web accessibility violations into three key categories - Syntactic, Semantic, and Layout. This taxonomy provides a structured foundation for developing our detection and correction method and redefining evaluation metrics. We propose a novel method, AccessGuru, which combines existing accessibility testing tools and Large Language Models (LLMs) to detect violations and applies taxonomy-driven prompting strategies to correct all three categories. To evaluate these capabilities, we develop a benchmark of real-world Web accessibility violations. Our benchmark quantifies syntactic and layout compliance and judges semantic accuracy through comparative analysis with human expert corrections. Evaluation against our benchmark shows that AccessGuru achieves up to 84% average violation score decrease, significantly outperforming prior methods that achieve at most 50%.

Text-attributed graphs (TAGs) have emerged as a powerful representation for modeling complex relationships across diverse domains. With the rise of large language models (LLMs), there is growing interest in leveraging their capabilities for graph learning. However, current approaches face significant challenges in embedding structural information into LLM-compatible formats, requiring either computationally expensive alignment mechanisms or manual graph verbalization techniques that often lose critical structural details. Moreover, these methods typically require labeled data from source domains for effective transfer learning, significantly constraining their adaptability. We propose STAG, a novel self-supervised framework that directly quantizes graph structural information into discrete tokens using a frozen codebook. Unlike traditional quantization approaches, our method employs soft assignment and KL divergence guided quantization to address the unique challenges of graph data, which lacks natural tokenization structures. Our framework enables both LLM-based and traditional learning approaches, supporting true zero-shot transfer learning without requiring labeled data even in the source domain. Extensive experiments demonstrate state-of-the-art performance across multiple node classification benchmarks while maintaining compatibility with different LLM architectures, offering an elegant solution to bridging graph learning with LLMs.

Funding The study was supported by the Start - up Fund for RAPs under the Strategic Hiring Scheme (P0048623) from HKSAR, the Global STEM Professorship Scheme (P0046113) and Henry G. Leong Endowed Professorship in Elderly Vision Health. 2 Abstract Large language models (LLMs) show promise for tailored healthcare communication but face challenges in interpretability and multi - task integration particularly for domain - specific needs like myopia, a nd their real - world effectiveness as patient education tools has yet to be demonstrated . Here, we introduce ChatMyopia, an LLM - based AI agent designed to address text and image - based inquiries related to myopia. To achieve this, ChatMyopia integrates an image classification tool and a retrieval - augmented knowledge base built from literature, expert consensus, and clinical guidelines. M yopic maculopathy grading task, single question examination and human evaluations validated its ability to deliver personalized, accurate, and safe responses to myopia - related inquirie s with high scalability and interpretability . In a randomized controlled trial (n=70, NCT06607822), ChatMyopia significantly improved patient satisfaction compared to traditional leaflets, enhancing patient education in accuracy, empathy, disease awareness, and patient - eye care practitioner communication. These findings highlight ChatMyopia ' s potential as a valuable supplement to enhance patient education and improve satisfaction with medical services in primary eye care settings . Keywords: Large language model, Medical a gent, Myopia, Patient education, Randomized controlled trial. Introduction For patients, a lack of basic understanding of their condition before initial consultations can hinder communication, as clinicians may spend time explaining fundamental concepts instead of critical issues, resulting in poor decisions and noncompliance [1, 2] . Therefore, patients require professional information and support to enhance their healthcare experiences.

Generative agents have made significant progress in simulating human behavior, but existing frameworks often simplify emotional modeling and focus primarily on specific tasks, limiting the authenticity of the simulation. Our work proposes the Psychological-mechanism Agent (PSYA) framework, based on the Cognitive Triangle (Feeling-Thought-Action), designed to more accurately simulate human behavior. The PSYA consists of three core modules: the Feeling module (using a layer model of affect to simulate changes in short-term, medium-term, and long-term emotions), the Thought module (based on the Triple Network Model to support goal-directed and spontaneous thinking), and the Action module (optimizing agent behavior through the integration of emotions, needs and plans). To evaluate the framework's effectiveness, we conducted daily life simulations and extended the evaluation metrics to self-influence, one-influence, and group-influence, selection five classic psychological experiments for simulation. The results show that the PSYA framework generates more natural, consistent, diverse, and credible behaviors, successfully replicating human experimental outcomes. Our work provides a richer and more accurate emotional and cognitive modeling approach for generative agents and offers an alternative to human participants in psychological experiments.

Scalable oversight protocols aim to empower evaluators to accurately verify AI models more capable than themselves. However, human evaluators are subject to biases that can lead to systematic errors. We conduct two studies examining the performance of simple oversight protocols where evaluators know that the model is "correct most of the time, but not all of the time". We find no overall advantage for the tested protocols, although in Study 1, showing arguments in favor of both answers improves accuracy in cases where the model is incorrect. In Study 2, participants in both groups become more confident in the system's answers after conducting online research, even when those answers are incorrect. We also reanalyze data from prior work that was more optimistic about simple protocols, finding that human evaluators possessing knowledge absent from models likely contributed to their positive results--an advantage that diminishes as models continue to scale in capability. These findings underscore the importance of testing the degree to which oversight protocols are robust to evaluator biases, whether they outperform simple deference to the model under evaluation, and whether their performance scales with increasing problem difficulty and model capability.

AI systems now function as cognitive extensions, evolving from tools to active cognitive collaborators within human-AI integrated systems. While these systems can amplify cognition - enhancing problem-solving, learning, and creativity - they present a fundamental "comfort-growth paradox": AI's user-friendly nature may foster intellectual stagnation by minimizing cognitive friction necessary for development. As AI aligns with user preferences and provides frictionless assistance, it risks inducing cognitive complacency rather than promoting growth. We introduce Enhanced Cognitive Scaffolding to resolve this paradox - reconceptualizing AI from convenient assistant to dynamic mentor. Drawing from Vygotskian theories, educational scaffolding principles, and AI ethics, our framework integrates three dimensions: (1) Progressive Autonomy, where AI support gradually fades as user competence increases; (2) Adaptive Personalization, tailoring assistance to individual needs and learning trajectories; and (3) Cognitive Load Optimization, balancing mental effort to maximize learning while minimizing unnecessary complexity. Research across educational, workplace, creative, and healthcare domains supports this approach, demonstrating accelerated skill acquisition, improved self-regulation, and enhanced higher-order thinking. The framework includes safeguards against risks like dependency, skill atrophy, and bias amplification. By prioritizing cognitive development over convenience in human-AI interaction, Enhanced Cognitive Scaffolding offers a pathway toward genuinely amplified cognition while safeguarding autonomous thought and continuous learning.

We present a SemiEmpirical Theory of Learning (SETOL) that explains the remarkable performance of State-Of-The-Art (SOTA) Neural Networks (NNs). We provide a formal explanation of the origin of the fundamental quantities in the phenomenological theory of Heavy-Tailed Self-Regularization (HTSR): the heavy-tailed power-law layer quality metrics, alpha and alpha-hat. In prior work, these metrics have been shown to predict trends in the test accuracies of pretrained SOTA NN models, importantly, without needing access to either testing or training data. Our SETOL uses techniques from statistical mechanics as well as advanced methods from random matrix theory and quantum chemistry. The derivation suggests new mathematical preconditions for ideal learning, including a new metric, ERG, which is equivalent to applying a single step of the Wilson Exact Renormalization Group. We test the assumptions and predictions of SETOL on a simple 3-layer multilayer perceptron (MLP), demonstrating excellent agreement with the key theoretical assumptions. For SOTA NN models, we show how to estimate the individual layer qualities of a trained NN by simply computing the empirical spectral density (ESD) of the layer weight matrices and plugging this ESD into our SETOL formulas. Notably, we examine the performance of the HTSR alpha and the SETOL ERG layer quality metrics, and find that they align remarkably well, both on our MLP and on SOTA NNs.

To understand and identify the unprecedented risks posed by rapidly advancing artificial intelligence (AI) models, this report presents a comprehensive assessment of their frontier risks. Drawing on the E-T-C analysis (deployment environment, threat source, enabling capability) from the Frontier AI Risk Management Framework (v1.0) (SafeWork-F1-Framework), we identify critical risks in seven areas: cyber offense, biological and chemical risks, persuasion and manipulation, uncontrolled autonomous AI R\&D, strategic deception and scheming, self-replication, and collusion. Guided by the "AI-$45^\circ$ Law," we evaluate these risks using "red lines" (intolerable thresholds) and "yellow lines" (early warning indicators) to define risk zones: green (manageable risk for routine deployment and continuous monitoring), yellow (requiring strengthened mitigations and controlled deployment), and red (necessitating suspension of development and/or deployment). Experimental results show that all recent frontier AI models reside in green and yellow zones, without crossing red lines. Specifically, no evaluated models cross the yellow line for cyber offense or uncontrolled AI R\&D risks. For self-replication, and strategic deception and scheming, most models remain in the green zone, except for certain reasoning models in the yellow zone. In persuasion and manipulation, most models are in the yellow zone due to their effective influence on humans. For biological and chemical risks, we are unable to rule out the possibility of most models residing in the yellow zone, although detailed threat modeling and in-depth assessment are required to make further claims. This work reflects our current understanding of AI frontier risks and urges collective action to mitigate these challenges.

Chain-of-Thought (CoT) reasoning has emerged as a powerful tool for enhancing the problem-solving capabilities of large language models (LLMs). However, the theoretical foundations of learning from CoT data remain underdeveloped, and existing approaches -- such as Supervised Fine-Tuning (SFT), Reinforcement Learning (RL), Tree-of-Thoughts (ToT), and Monte Carlo Tree Search (MCTS) -- often fail on complex reasoning tasks. In this work, we identify core obstacles that hinder effective CoT learning, including distribution drift, lack of embedded search, and exponential inference costs. We introduce the Diligent Learner, a new learning paradigm that explicitly models reasoning as a depth-first search guided by a validator and supports backtracking upon failure. Under two mild and realistic assumptions, we prove that the Diligent Learner can efficiently learn from CoT data while existing methods fail to do so. This framework offers a path toward building scalable and reliable reasoning systems trained on naturally occurring, incomplete data -- paving the way for the development of Large Reasoning Models (LRMs) with robust, interpretable problem-solving abilities.

In many applications, one needs to learn a dynamical system from its solutions sampled at a finite number of time points. The learning problem is often formulated as an optimization problem over a chosen function class. However, in the optimization procedure, it is necessary to employ a numerical scheme to integrate candidate dynamical systems and assess how their solutions fit the data. This paper reveals potentially serious effects of a chosen numerical scheme on the learning outcome. In particular, our analysis demonstrates that a damped oscillatory system may be incorrectly identified as having "anti-damping" and exhibiting a reversed oscillation direction, despite adequately fitting the given data points.