AI news archive.
10,661 articles filed under Research, newest first, from the outlets listed on the sources page.
- arXiv cs.AIresearchToward a Graded Measure of Belief Stability in Large Language Models
arXiv:2609.34158v1 Announce Type: cross Abstract: Large language models (LLMs) increasingly mediate how people access and reason with information, yet factual reliability is usually evaluated one judgment at a time. We i…
- arXiv cs.AIresearchWorldGraph: Graph-Native World Modeling
arXiv:2609.34159v1 Announce Type: cross Abstract: World models infer latent states of an environment to capture its underlying dynamics and predict future evolution. Many real-world environments, however, are inherently…
- arXiv cs.AIresearchGradLev: Token-Parallel Test-Time Training Via Costate Prediction
arXiv:2609.34174v2 Announce Type: replace-cross Abstract: Test-time training (TTT) allows a model to improve its predictions at inference time by updating weights after every observed token. However, sequential gra- dien…
- arXiv cs.AIresearchUnified Visual-Tactile-Action Modeling from Human Demonstrations for Dexterous Manipulation
arXiv:2609.34182v2 Announce Type: replace-cross Abstract: Dexterous manipulation requires tactile feedback. However, robot tactile demonstrations are difficult to scale,because dexterous-hand teleoperation provides limit…
- arXiv cs.AIresearchEntroPack: Fast and Accurate Entropy-Coded Weight Compression at Arbitrary Bitrates
arXiv:2609.34185v1 Announce Type: cross Abstract: Weight compression helps large neural networks fit deployment memory budgets, but common fixed-width formats offer only coarse storage choices. Entropy coding supports fi…
- arXiv cs.AIresearchLLMs are not stochastic parrots: Evidence for meaning-mediated abstraction from conlang-like tasks
arXiv:2609.34187v4 Announce Type: replace-cross Abstract: The strong version of the stochastic parrot argument claims that, although large language models (LLMs) may exceed rote regurgitation, they cannot move beyond sta…
- arXiv cs.AIresearchAlphaPareto: Formulaic Alpha Discovery with LLM-Guided Multi-Objective Reinforcement Learning
arXiv:2609.34188v1 Announce Type: cross Abstract: Formulaic alpha discovery is a core challenge in quantitative trading, as identifying alphas that work well together remains difficult. Recent reinforcement learning (RL)…
- arXiv cs.AIresearchReGDiff: Guided Diffusion in Regulated Latent Space for Exploring Metamaterial Voxel Geometry
arXiv:2609.34231v1 Announce Type: cross Abstract: Metamaterials are artificially engineered structures whose mechanical and physical behaviors are strongly shaped by geometry rather than composition. Voxel representation…
- arXiv cs.AIresearchCoherence-Aware Distributional Evaluation of Open-Ended Text Generation
arXiv:2609.34240v2 Announce Type: replace-cross Abstract: Existing open-ended generation metrics measure likelihood, lexical diversity, or distributional similarity in generic representation space, yet can miss fundament…
- arXiv cs.AIresearchWAM-OPD: Sharpening World Action Models via On-Policy Distillation
arXiv:2609.34250v1 Announce Type: cross Abstract: Pretrained world action models (WAMs) provide generalist capabilities across diverse robotic manipulation tasks, yet improving target-task performance to an expert level…
- arXiv cs.AIresearchInvestigating Human--AI Discrepancies via Multiple-Solution Problems
arXiv:2609.34258v1 Announce Type: cross Abstract: Frontier artificial intelligence (AI) models are benchmarked on whether they reach a correct answer. Yet many problems admit several correct answers and repeated attempts…
- arXiv cs.AIresearchBayesian Active Learning for Intent Disambiguation in Interactive Robot Planning
arXiv:2609.34270v1 Announce Type: cross Abstract: Interactive robot planning requires robots to infer and execute human intentions from natural language instructions that are often ambiguous, incomplete, or underspecifie…
- arXiv cs.AIresearchSee, Measure, and Reason: Learning Visually Grounded Reasoning in Pathology
arXiv:2609.34277v1 Announce Type: cross Abstract: Pathological assessment relies on recognizing fine-grained visual details in histological images. Vision-language models (VLMs) increasingly support pathology interpretat…
- arXiv cs.AIresearchDirect Self-Evolving Optimization: Evolving LLMs without Challenger Training
arXiv:2609.34279v1 Announce Type: cross Abstract: Self-evolving language models improve by generating tasks and learning from their own feedback, but adapting the task generator often requires a separate challenger-train…
- arXiv cs.AIresearchReScraper: Unified Scraping and Cleaning of Web Data for Effective LLM Pretraining
arXiv:2609.34287v1 Announce Type: cross Abstract: LLM pretraining corpora are normally cleaned by a stack of hand-written heuristics. A heuristic scraper extracts the main content from HTML, and dozens of rule-based filt…
- arXiv cs.AIresearchWhen World Models Lie: Adaptive Safety Analysis Under Wrong Imaginations
arXiv:2609.34300v1 Announce Type: cross Abstract: World models offer a powerful substrate for safety reasoning in high-dimensional robotic systems, but they are also fallible: their predictions can be biased, miscalibrat…
- arXiv cs.AIresearchOne Sequence, Many Decodings: CAGenMol-2 Recasts Drug Design as Masked Molecular Inference
arXiv:2609.34301v1 Announce Type: cross Abstract: Drug design couples property evaluation, conditional generation, structure-based design, and local optimization, yet machine learning systems typically address these capa…
- arXiv cs.AIresearchPlaylistEval: Can Video-Language Judges Be Trusted at Day Scale and Beyond?
arXiv:2609.34314v1 Announce Type: cross Abstract: Video-language models are increasingly used as judges of video understanding, both for evaluating model outputs and for training reward models. Whether their judgments re…
- arXiv cs.AIresearchSkillPE: Creativity-Oriented Cinematic Skill Evolution for Text-to-Video Prompt Engineering
arXiv:2609.34335v1 Announce Type: cross Abstract: Achieving high-quality, cinematic results in text-to-video generation remains challenging for non-experts, whose prompts often lack professional narrative and creative de…
- arXiv cs.AIresearchLearning to Steer, Steering to See: Unveiling the Geometry of RLVR in Large Language Models via Trainable Vectors
arXiv:2609.34344v1 Announce Type: cross Abstract: Reinforcement learning (RL) has become a key paradigm for enhancing the reasoning of large language models, yet the high dimensionality of parameter updates makes its tra…
- arXiv cs.AIresearchCommutator Memory: Sparse, Path-Local Reading and Steering in Language Models
arXiv:2609.34348v1 Announce Type: cross Abstract: Gradient updates on different data generally do not commute: training a language model on two data sources in opposite orders gives different weights, even with the same…
- arXiv cs.AIresearchPredictive Semantic Safety: From Visual Physical Reasoning to Safety-Critical Control
arXiv:2609.34356v1 Announce Type: cross Abstract: Physical interactions can create future hazards that are not apparent from the robot's current geometric surroundings. We present a framework termed Predictive Semantic S…
- arXiv cs.AIresearchSyncRA: Learning Temporal Correspondence in Omni-Modal Models
arXiv:2609.34363v1 Announce Type: cross Abstract: Recent omni-modal models demonstrate strong perception of audio and visual inputs, yet often struggle to connect what they hear with what they see at the same moment. Thi…
- arXiv cs.AIresearchWhen Harness Beats Scale, and When Reading Beats Both
arXiv:2609.34366v1 Announce Type: cross Abstract: We describe our system for DocSem, the document-grounded quantitative reasoning shared task at DocInsights 2026, and analyze why it succeeded on labeled data and failed o…
- arXiv cs.AIresearchFrom Static to Dynamic: On-Policy Distillation from Image to Video Diffusion Models
arXiv:2609.34371v1 Announce Type: cross Abstract: On-policy distillation (OPD) specializes pretrained video diffusion models through teacher supervision along the student's own generation trajectory. Although large video…
- arXiv cs.AIresearchABC-Align: Prediction-Powered Alignment with Adaptive Bias Control
arXiv:2609.34374v1 Announce Type: cross Abstract: Language model post-training is often bottlenecked by the need for human-collected preference data, which is expensive and difficult to scale. Reinforcement learning from…
- arXiv cs.AIresearchLRC-JEPA: Disentangling Dynamics and Residual Context for Efficient World Models
arXiv:2609.34375v1 Announce Type: cross Abstract: Compact JEPA world models enable efficient latent-space planning, but low-dimensional representation trained under reward-free self-supervision must encode both action-co…
- arXiv cs.AIresearchDPS: Dual-Mode Precision LLM Serving with Semi-Unified Memory
arXiv:2609.34380v1 Announce Type: cross Abstract: Existing LLM serving systems virtualize and optimize KV-cache memory, but treat model-weight memory as fixed throughout execution. Recent work on multi-precision model re…
- arXiv cs.AIresearchP2P: Cross-View Population Denoising for Unpaired Single-Cell Perturbation Response Prediction
arXiv:2609.34391v1 Announce Type: cross Abstract: AIVC (AI Virtual Cell) is a learned simulator of cellular behavior across conditions. Predicting how a cell population responds transcriptionally to a genetic perturbatio…
- arXiv cs.AIresearchEval4DiRec: A Unified and Systematic Evaluation Framework for Diffusion-based Recommender Systems
arXiv:2609.34404v2 Announce Type: replace-cross Abstract: Leveraging the strong generative capabilities and stable training dynamics of diffusion models, diffusion-based recommender systems (RSs) have recently emerged as…
- arXiv cs.AIresearchMASCIT: A Mask-Aware State Space Classifier for Naturally Irregular Time Series
arXiv:2609.34409v2 Announce Type: replace-cross Abstract: Naturally irregular time series combine asynchronous observations, missing values, unequal lengths, and nonuniform sampling, while dense adapters can discard temp…
- arXiv cs.AIresearchZero-Shot Cue-Grounded Topic Segmentation of Spoken Documents
arXiv:2609.34425v1 Announce Type: cross Abstract: Topic segmentation structures spoken documents into coherent sections, facilitating navigation and downstream understanding. The appropriate granularity can vary substant…
- arXiv cs.AIresearchModeling Whole-Slide Images as Dynamic Tumor Microenvironment Fields
arXiv:2609.34451v1 Announce Type: cross Abstract: Due to the gigapixel-scale nature of whole-slide images (WSIs), weakly supervised WSI analysis is commonly formulated as a multiple instance learning (MIL) problem, where…
- arXiv cs.AIresearchSPACE-LoRA: Allocating Activation-Subspace Protection for Continual Learning
arXiv:2609.34453v1 Announce Type: cross Abstract: This study addresses the catastrophic forgetting problem that occurs when sequentially learning successive tasks using Low-Rank Adaptation (LoRA) from a lifelong learning…
- arXiv cs.AIresearchUnderstanding Generalization Requires Universal Induction
arXiv:2609.34458v1 Announce Type: cross Abstract: Classical statistical theory is insufficient to explain the successes of general-purpose AI models, because it depends on handcrafted inductive biases that it cannot just…
- arXiv cs.AIresearchPrecise Editing and Flexible Referencing for Interactable Worlds
arXiv:2609.34470v1 Announce Type: cross Abstract: We present EditWorld, a video world model for precise editing and flexible referencing in interactable worlds. Existing video world models primarily focus on navigation,…
- arXiv cs.AIresearchVL-AcneSeg: A Vision-Language Framework for Region-Aware Acne Lesion Segmentation
arXiv:2609.34472v1 Announce Type: cross Abstract: Acne assessment is crucial for clinical decision-making, yet traditional grading and counting are subjective and fail to account for lesion size. While area-based assessm…
- arXiv cs.AIresearchCausal Routing for Unlearning
arXiv:2609.34475v1 Announce Type: cross Abstract: LLMs cannot forget the way we delete a file. Strangely, we are asked to remove something that was never put anywhere in particular. What the model took from a piece of te…
- arXiv cs.AIresearchLearn Here, Move Less Elsewhere: Input-Conditioned Plasticity from Retained-Domain Activation Atlases
arXiv:2609.34478v1 Announce Type: cross Abstract: Task-specific fine-tuning can rewrite a language model's answers beyond the training task, complicating updates that must preserve existing behavior. We introduce ATLAS,…
- arXiv cs.AIresearchSentZero: An Enhanced Sentence-Centric Vision-Language Pretraining for Multi-Task Zero-Shot Chest X-Ray Analysis
arXiv:2609.34479v1 Announce Type: cross Abstract: Vision-language (VL) pretraining using paired chest X-ray (CXR) images and radiology reports has shown strong potential for medical image understanding. However, existing…
- arXiv cs.AIresearchFlexLoop: Depth-Elastic Looped Policies for Adaptive Test-Time Computation in Deep RL
arXiv:2609.34488v1 Announce Type: cross Abstract: Looped architectures scale computation by reusing the same parameters across recurrent steps, and recent work shows that they substantially improve deep reinforcement lea…
- arXiv cs.AIresearchScalable GNN-based Knowledge Graph Representation Learning with Efficient Message Passing
arXiv:2609.34499v1 Announce Type: cross Abstract: Graph neural networks (GNNs) excel at representation learning on Knowledge Graphs (KGs), achieving stateof-the-art performance on tasks like link prediction or entity cla…
- arXiv cs.AIresearchShallow Queries, Mature Values: Depth-Asynchronous Self-Speculation for Looped Transformers
arXiv:2609.34538v1 Announce Type: cross Abstract: Looped Transformers reuse a shared block across recurrent depths, making autoregressive decoding expensive because every generated token requires many sequential recurren…
- arXiv cs.AIresearchActionLens: Diagnosing Spatial-Temporal Binding Failures in Vision-Language Models
arXiv:2609.34547v1 Announce Type: cross Abstract: Video-capable vision-language models score above 80\% on popular benchmarks yet struggle with spatial-temporal binding: associating the right action with the right person…
- arXiv cs.AIresearchGenNVS: Geometry-enhanced Novel View Synthesis via Disentangled 3D Prior
arXiv:2609.34579v2 Announce Type: replace-cross Abstract: Single-image novel view synthesis remains challenging because the underlying 3D geometry is highly ambiguous. Recent diffusion-based approaches produce plausible…
- arXiv cs.AIresearchIn-game Toxic Detection: Bi-directional Representations with Attention Residuals
arXiv:2609.34584v1 Announce Type: cross Abstract: In-game toxic language has emerged as a critical concern in the gaming industry and community. While several frameworks and models for online game toxicity analysis have…
- arXiv cs.AIresearchReinforcement Learning from Intermediate Renders for Image-to-Code Generation
arXiv:2609.34587v1 Announce Type: cross Abstract: Reinforcement learning is increasingly used to post-train vision-language models for image-to-code generation, such as generating SVG code from a reference image, by opti…
- arXiv cs.AIresearchSummarize Before Grounding: Query-Guided Chunk Condensation for Long-Video Temporal Grounding
arXiv:2609.34598v1 Announce Type: cross Abstract: Video temporal grounding (VTG) aims to localize the video interval corresponding to a language query. Recent large vision-language models (LVLMs) show great potential in…
- arXiv cs.AIresearchEfficient World Action Model Inference with Adaptive Intermediate States
arXiv:2609.34608v1 Announce Type: cross Abstract: World Action Models (WAMs) enable future-aware control by jointly modeling actions and environment dynamics. However, iterative diffusion or flow inference incurs substan…
- arXiv cs.AIresearchNereus: Adaptive Parallelism for LLM Post-Training
arXiv:2609.34645v1 Announce Type: cross Abstract: Reinforcement learning (RL) post-training for large language models (LLMs) coordinates multiple models across generation, inference, and training on GPU clusters. Several…