Real-time AI-curated news from 93,297+ articles across 50+ sources. Sentiment analysis, importance scoring, and key takeaways — updated every 15 minutes.
Researchers present a data-centric optimization framework for AI coding agents operating on branching lakehouses, demonstrating that agent skills can be systematically improved through task-verifier pairs and sandboxed execution. The approach treats agent evaluation as state verification rather than output matching, achieving 31.9% accuracy improvements on preliminary tasks.
Researchers propose 'Model Science,' a systematic discipline for understanding AI models beyond traditional benchmarking. The framework consolidates analysis around four functional perspectives—Verify, Explore, Steer, and Refine—and emphasizes deep study of individual models rather than population-level comparisons, drawing lessons from established sciences like neuroscience and medicine.
Researchers introduce TaskWeave, a hierarchical framework that enables large language model agents to maintain coherent behavior in complex organizational simulations over extended periods. The system uses memory-centered coordination and dependency-aware tracking to sustain long-horizon tasks, demonstrating viability for enterprise-level multi-agent applications through year-long IT company simulations.
Researchers analyzed how language models make decisions by tracing answer scores across neural network layers in 9,000 MMLU trajectories, finding that correct answers are often unstable and that attention mechanisms better preserve correctness than MLP layers. The study reveals decision-making is a distributed process rather than a final-layer phenomenon, with implications for understanding model reliability and interpretability.
Researchers have developed an AI system using multimodal data analysis to predict at-risk mathematics students and provide early academic warnings. The framework combines knowledge graphs with temporal modeling to identify students struggling with complex concepts and enable timely interventions that improve learning outcomes.
Researchers developed an integrated algorithmic platform combining Building Information Modeling, sensor data, and multi-objective optimization to design energy-efficient buildings. Testing on a mid-rise office building achieved a 29.3% reduction in annual energy consumption while limiting lifecycle cost increases to 3.7%, demonstrating practical scalability for green building design.
HomeFlow introduces a data flywheel system for training large language model agents in smart home environments, using procedural generation and Monte Carlo tree search to create diverse, verifiable training trajectories. The approach achieves 87.03% task success rates on a new SmartHome-Bench benchmark, outperforming GPT-5.5 by 1.23 percentage points.
Researchers propose GCAN, a novel deep learning framework that uses counterfactual generation and brain atlas constraints to improve the explainability of cognitive decline diagnosis from brain imaging data. The method achieves competitive classification performance on mild cognitive impairment and subjective cognitive decline detection while providing interpretable insights into disease-related connectivity changes.
SIRIUS-SQL introduces a multi-candidate approach to Text-to-SQL generation that addresses redundancy, execution error classification, and selector limitations through difficulty-smoothing reinforcement learning, targeted repair mechanisms, and hybrid confidence-gated selection. The system achieves 75.88% accuracy on BIRD dev and 91.20% on SPIDER test, surpassing previous state-of-the-art multi-candidate systems.
SkillSmith introduces a co-evolution framework where AI agent skills and tools develop together rather than independently, using ecological dynamics to model skill interactions and anti-pattern tracking to prevent repeated failures. The system demonstrates consistent improvements across multiple benchmarks and model scales, particularly as task complexity increases.
Researchers propose a Mean-Field Entropy Dynamics framework to analyze failure modes in Large Language Model multi-agent systems, identifying a "Reasoning Trap" where sophisticated reasoning models paradoxically perform poorly as orchestrators due to context limitations. The study introduces Inverse Workflow Generation for benchmarking and provides physically interpretable parameters for predicting system stability.
Researchers introduce a failure-aware observability framework to diagnose wasted computation in multi-agent LLM systems, identifying six failure modes through online trace signals. Testing on 165 GAIA validation traces reveals 41% failure rates across difficulty levels and token consumption ranging from 8,152 to 16,389 tokens, positioning observability as a diagnostic layer between execution logs and accuracy.
Researchers propose GovAI-Pipe, a technical governance framework that operationalizes AI policy principles into auditable deployment checkpoints for Turkey's e-Government Gateway, which serves 68 million users. The four-layer pipeline addresses the gap between high-level regulatory frameworks like the EU AI Act and the practical implementation of AI systems in citizen-facing government applications.
Researchers demonstrate that deterministic post-retrieval aggregation using serial numbers outperforms LLM-based conflict resolution in memory systems by 10-28 percentage points. The study reveals that the bottleneck in fact-consolidation tasks is assembly logic rather than storage, with implications for building more reliable AI agents that track evolving information.
Researchers present a category-theoretic framework for agentic AI systems that can revise their own representational structures during scientific discovery, rather than merely generating answers within fixed assumptions. The work demonstrates how self-revising discovery systems can be engineered for materials science through two instantiated systems: Builder/Breaker and CategoryScienceClaw.
Researchers present graph-coupled causal Bayesian optimization, a method that improves expensive system optimization by sharing information across related interventions through a causal kernel. The approach demonstrates logarithmic information gains and cleanly separates optimization, causal estimation, and intervention selection errors, with strongest performance when direct interventions are unavailable.
Researchers developed a brain-computer musical interface (BCMI) that translates EEG signals into real-time adaptive music based on emotional states. Testing with 22 participants revealed that frontal alpha asymmetry—a common neurophysiological marker—failed to reliably distinguish intentional emotional states, with individual differences like musical training explaining more variance than actual emotional manipulation.
TERRA introduces a theoretical framework for transferring machine learning representations across structurally similar but unrelated domains—from driving scenes to robot workspaces to financial markets. The research formalizes when and how well a model trained in one domain generalizes to another through mathematical constructs like Markov decision process homomorphisms and Gromov-Wasserstein distances, presenting a preregistered experimental program without empirical validation.
Researchers introduce RoleCDE, a benchmark for evaluating role-playing agents in large language models, revealing a 'Role Value Decoupling' phenomenon where LLMs default to alignment-oriented decisions over role-specific values when conflicts arise. Fine-tuning with RoleCDE data effectively mitigates this behavior while preserving general performance.
Researchers propose S-SPPO, an improved framework for aligning large language models with human preferences that addresses instability issues in Self-Play Preference Optimization. The method uses semantic calibration techniques to prevent policy degradation when the model generates semantically similar responses, achieving competitive performance on AlpacaEval 2.0 without additional human annotations.
Researchers propose Joint Neighborhood Optimization (JNO), a new framework for knowledge editing in large language models that simultaneously manages desired information propagation and prevents unintended disruption to related facts. The method uses Pressure-Aware Coordination to jointly optimize coupled constraints and achieves 7% improvement in both propagation and preservation metrics across different model architectures.
Researchers introduce ReSkill, an RL-in-the-loop framework that improves how AI agents create and refine reusable skills during policy learning. The method synchronizes skill evolution with policy optimization, enabling agents to automatically develop, test, and prune strategies that generalize across tasks more effectively than existing approaches.
Researchers introduce MobEvolve, an AI framework that generates realistic human mobility patterns by combining interpretable heuristics with LLM agents that self-evolve through iterative learning. The system outperforms existing deep learning and LLM approaches while maintaining computational efficiency and behavioral plausibility across Singapore and Montreal datasets.
Researchers introduced GAIATrace, a token-level trace dataset documenting how state-of-the-art agentic AI systems (MiroThinker and OWL) execute general tasks, alongside Vidur-Agent, a simulator enabling reproducible system evaluation. This work addresses the black-box nature of agentic AI by providing unprecedented visibility into reasoning processes and system-level behavior.
Researchers propose a framework for incorporating Large Language Model (LLM) priors into multi-objective Bayesian optimization while maintaining robustness against miscalibrated advice. Using an objective-wise reputation mechanism and counterfactual gating, the approach dynamically adjusts trust in LLM suggestions based on observed performance rather than accepting them blindly, with empirical validation across molecular optimization tasks.
