Real-time AI-curated news from 88,764+ articles across 50+ sources. Sentiment analysis, importance scoring, and key takeaways — updated every 15 minutes.
Researchers find that large language models make decisions based on systematic behavioral patterns but struggle to accurately articulate their reasoning. The study reveals a disconnect between what LLMs claim influences their choices and the attributes that actually drive their decisions, suggesting models operate with 'superficial beliefs' rather than fully understood decision frameworks.
Researchers introduce Workflow-GYM, a benchmark for evaluating AI agents on complex, long-horizon professional GUI tasks across specialized software environments. Testing reveals that even state-of-the-art models achieve only 30% success rates, exposing significant limitations in agent consistency, error handling, and domain-specific software comprehension.
Researchers demonstrate that successful machine learning strategies remain highly compressible and generalizable even when trained on held-out benchmarks, suggesting overfitting in benchmark-driven ML is rare because effective strategies occupy a low-complexity region of strategy space. Using LLM-driven research agents, they show that short prompts and minimal feedback suffice to reproduce high-performance models across diverse domains.
Researchers introduce Monte Carlo Pass Search (MCPS), a novel AI system that evaluates football passes by simulating counterfactual scenarios using trajectory generation and value prediction models. The work combines existing machine learning techniques with a new public Bundesliga dataset featuring 3D ball tracking, enabling distribution-aware analysis of pass execution quality and decision-making.
ReasonAlloc is a training-free framework that optimizes key-value cache memory allocation during LLM inference for reasoning tasks by using hierarchical, non-uniform budget distribution across layers and attention heads. The method significantly reduces memory bottlenecks in chain-of-thought reasoning while maintaining performance, outperforming existing compression approaches on mathematical reasoning benchmarks.
Researchers demonstrate that self-distillation in language models improves significantly when feedback is structurally aligned with the model's reasoning trace rather than using binary rewards or reference solutions. Step-aligned critique, which targets only tokens where reasoning fails, outperforms alternative approaches by 5-16 points, suggesting that feedback design fundamentally shapes model learning efficiency.
Researchers developed and tested visual disclosure methods for communicating human-AI collaboration in journalism, finding that simple text labels fail to convey nuance while interactive formats like chatbot interfaces provide more transparency. The study reveals that visualization design significantly influences reader perception of AI's actual role in news production, raising concerns about how disclosure formats can misrepresent collaborative contribution ratios.
This academic paper presents a framework for culturally-aware artificial intelligence in education (AIED) developed by undergraduate students working on community-based learning projects across Asia-Pacific regions. The research bridges computational science and social work to create AI solutions for cultural heritage preservation and sustainable development, emphasizing human-centered design and cultural context in educational technology.
This academic research applies AI-driven speech processing to analyze team-teaching dynamics in university classrooms across 36 sessions. The study reveals that experienced teachers, undergraduate instruction, and collaborative learning tasks correlate with greater loudness variation, suggesting strategic vocal modulation to enhance engagement and highlight key information.
Researchers introduce the Agentic Social Affordance Framework (ASAF), a theoretical model examining how agent identity design in multi-agent AI systems influences human collaboration outcomes. The framework proposes that agent social identity functions as a collaboration interface distinct from technical orchestration, operating through identity signaling, behavioral priming, and collaborative governance mechanisms.
Researchers introduce CollabSkill, a framework for evaluating how AI agents perform when collaborating with real human workers on occupational tasks. Using data from 93 workers across 386 sessions, the study reveals that Claude Code outperforms Codex in practical collaboration scenarios—diverging from autonomous benchmark rankings—and identifies hands-on experience as the primary driver of effective human-AI teamwork.
Researchers propose Self-EmoQ, an emotion-planning framework that determines emotional context before text generation to improve streaming emotional text-to-speech synthesis. The system uses reinforcement learning with Plutchik's emotion theory and demonstrates superior performance on multiple dialogue datasets, with a functional real-time deployment pipeline.
Researchers have developed a lightweight, real-time human pose estimation (HPE) system using MediaPipe that enables practical athletic performance analysis without expensive marker-based motion capture equipment. The work surveys existing HPE approaches and contributes a modular prototype delivering AI-powered feedback for sports training with minimal computational overhead.
CANVAS is an automated AI system that generates rich, multi-sensory art descriptions and synchronized audio narration for museum collections and digital art, addressing accessibility gaps for blind and low-vision audiences. The system processes images through large language models and text-to-speech services via Zapier, producing detailed captions faster and cheaper than human alternatives while demonstrating superior lexical diversity compared to baseline alt-text.
Researchers introduce a framework for designing human-AI coordination in everyday products, addressing the gap between high-level AI design principles and practical UI implementation. The framework identifies three key dimensions—salience, involvement, and activity—and provides mid-level tools including coordination zones and design patterns applicable to commercial AI applications.
Researchers developed an AI framework using eight large language models to automatically generate high-quality source code documentation, with a novel multi-LLM evaluation system assessing outputs across nine quality criteria. Testing on a medical physics library revealed a 42% performance gap between top and bottom models, demonstrating the framework's effectiveness in reducing manual documentation effort for safety-critical software.
A theoretical computer science paper formalizes decision-making under information constraints as action-sufficient compression, where systems need only preserve distinctions relevant to choosing optimal actions rather than reconstructing full state information. The framework applies rate-distortion theory to support states with regret-based distortion, offering a mathematical foundation for robust single-cycle arbitration.
Researchers propose MGAP, a training-free decoding method that reduces hallucinations in multimodal large language models (MLLMs) by selectively suppressing language priors while preserving semantic structure. Unlike previous approaches that blindly penalize language biases, MGAP uses geometry-aware subspace projection to distinguish between helpful and harmful language priors, achieving improved hallucination suppression without degrading model coherence.
Researchers developed a machine-learning framework combining gradient-boosted decision trees with conformal prediction to improve non-alcoholic fatty liver disease (NAFLD) risk screening. The model achieved 91.2% internal and 89.1% external validation accuracy while identifying six key metabolic biomarkers, enabling better population-level disease stratification.
Researchers introduce UniTok, a universal tokenizer that converts continuous time series data into discrete tokens, enabling UniTok-FM—a foundation model pretrained via next-token prediction. This unified approach supports forecasting, generation, and classification tasks without task-specific modifications, achieving competitive performance with specialized models while enabling zero-shot and few-shot inference capabilities.
Researchers introduce Blurry Window Attention (BLA), a novel attention mechanism that addresses the quadratic complexity and memory limitations of traditional Transformer models by reconstructing sparse key-value history through Dirichlet kernel interpolation. BLA demonstrates 8x state efficiency improvements over sliding window attention while maintaining competitive performance on information retrieval tasks, positioning it as a viable alternative for long-context language modeling.
Researchers propose DualSelect, a framework for fine-tuning large language models that simultaneously selects relevant safety references and compatible task samples to preserve safety alignment while improving task performance. The method achieves significant safety improvements (5.10+ points) across models from 1B to 8B parameters without sacrificing utility.
QSplitFL introduces a Deep Q-Network framework that optimizes split point selection in federated learning by considering device heterogeneity, using lightweight hardware metrics instead of model weights. The approach demonstrates improved convergence and accuracy across multiple datasets and neural network architectures while adapting to varying client capabilities.
Researchers propose SD-GRPO, a new machine learning technique that improves how multimodal AI systems generate long-form responses by analyzing outputs in semantic segments rather than as a single unit. The method addresses a fundamental limitation in existing GRPO frameworks when applied to vision-language tasks, showing consistent performance improvements across controlled and real-world benchmarks.
Researchers introduce PatchSTG, a new graph Transformer architecture that addresses scalability challenges in traffic forecasting by partitioning unevenly distributed sensors into geographic patches. The model reduces computational complexity from quadratic to near-linear while maintaining competitive forecasting accuracy across multiple prediction horizons.
