Artificial Intelligence
Artificial Intelligence news — Tuesday 20 October 2026
150 stories from 8 sources, every one in English. Each links to its publisher.
-
Here’s the tea
MIT Technology Review Biotech
When Tristan Brown, an associate professor of history at MIT, first examined a Chinese tea chest label displayed at Old South Meeting House, he had no reason to doubt its story. Descendants of the Boston blacksmith Thomas Wells had donated…
-
It’s a bird, it’s a plane … it’s a robo-puffin?
MIT Technology Review Biotech
Given that water is 1,000 times denser than air, moving through those two mediums requires very different mechanics—or so one might assume. Yet birds like gulls and puffins can plunge into the sea, swim underwater, and leap back into the…
-
A cleaner path to rare earth metals
MIT Technology Review Biotech
Tomás Villalón Jr. ’14 was raised in San Antonio, Texas, but vividly remembers smog hanging over the city his family came from—Monterrey, Mexico, a steel manufacturing and metal fabrication hub. Even as a child, he realized it was…
-
Robotic therapy for stroke patients
MIT Technology Review Biotech
Strokes leave 5 million people with long-term impairments each year, and a growing shortage of physical therapists makes it hard for them to get the care they need. A new approach tries to solve that problem by combining artificial…
-
Nebraska native explores galaxies far, far away
MIT Technology Review Biotech
Growing up on a small farm in Nebraska, Megan Donahue ’85 discovered the wider world by reading her way through her local library. Today, she is learning how the wider universe works by studying galaxy clusters. Donahue discovered her…
-
A good TA: priceless
MIT Technology Review Biotech
These are images of quizzes I took in 3.091, Introduction to Solid-State Chemistry, in the fall of 1980. One is my very first exam at MIT, and getting 49 out of 100 was a soul-crushing experience. I can still picture the TA, Leszek…
-
A champion for brain injury patients
MIT Technology Review Biotech
Alissa Totman ’13 was a sophomore when she joined MIT BrainTrust, a student-run organization that connects Institute undergraduates and people with brain injuries for support and socializing. “That group was a great way to learn about the…
-
Melt it, spin it, use it again
MIT Technology Review Biotech
The average American throws out around 81 pounds of clothing each year, and there are few options for recycling it. MIT engineers hope to cut down on that textile waste with a new type of yarn. The thermoplastic yarn mimics the properties…
-
A reason to eat carbs
MIT Technology Review Biotech
A high-fat, low-carbohydrate diet can help some people lose weight, and some research has shown that it may protect against colon cancer. But a new study suggests that such a ketogenic diet may also increase the risk of cancer in the small…
-
Taking a swing at a pickleball startup
MIT Technology Review Biotech
In corporate America, Stanley Anderson ’79 says, executives often assume that their peers grew up in privileged circumstances. He did not. “I spent my early years trying not to get recruited into gangs, getting shot at, watching my friends…
-
Astrolab’s upcoming rover will run the first LLMs on the moon
MIT Technology Review Biotech
Large language models are set to go lunar. A computer capable of running AI software is one of the payloads aboard Astrolab’s FLIP lunar rover, which is slated to head to Earth’s celestial companion as early as November 2026. The…
-
Google, Jemina notebook reorganization..."Slide·Video creation pre-configured review"
AI Times (Korea)
Translated from Korean
-
NVIDIA Cosmos3-H-Dreams: Bringing Clinical Surgery to Life Through Generative Simulation
Hugging Face Blog
-
Tenstorrent attracts large investments with 8 trillion value..."In Suwon instead of the reader route"
AI Times (Korea)
Translated from Korean
-
Strengthening AI safety evaluation through independent third-party review
Amazon Science Blog
As frontier AI models grow in capability, so does the importance of ensuring that safety evaluations are rigorous and comprehensive. At Amazon, we invest heavily in internal safety testing for our models, spanning categories like…
-
SkyfallAI, CAD·BIM 3D design automation World model ‘Sky-001’ released
AI Times (Korea)
Translated from Korean
-
Elise Group, a Rally Gasse listed AI infrastructure.. pantyhose footjob
AI Times (Korea)
Translated from Korean
-
Plyto, the best performance of the three-quarter Changsai... Operating Profit 50% Break
AI Times (Korea)
Translated from Korean
-
KT, Acrylic and Medical Institution Custom AX Business Construction
AI Times (Korea)
Translated from Korean
-
MDSTECH supplies Nvidia AI platform package for domestic humanoid home robot project
AI Times (Korea)
Translated from Korean
-
Cryptocurrency mining malicious code captured on Mac... CPU utilization 1175%
AI Times (Korea)
Translated from Korean
-
[Bulletin] NC AI, Barco 3D, 50% discount promotion
AI Times (Korea)
Translated from Korean
-
Opened ‘Hacking Publicization’ with 100 companies that have been drilled by AI agent.
AI Times (Korea)
Translated from Korean
-
"AI enterprise cooperation will go directly to export control"
AI Times (Korea)
Translated from Korean
-
Deepshik, the flagship model pre-learning on Huawei chips.. . Domestic AI granular speed
AI Times (Korea)
Translated from Korean
-
Gwangyang Cybercrime Investigation Committee
AI Times (Korea)
Translated from Korean
-
FairStress: A deeper evaluation of fairness in LLM decision-making
Hugging Face Blog
-
Amazon Launches ‘Tochi New’ in Python Nvidia SW Exclusive Challenge
AI Times (Korea)
Translated from Korean
-
Synthetic-to-Real Transfer with Frozen DINOv3 Features
Towards Data Science
Training models on procedurally generated datasets for card detection and segmentation The post Synthetic-to-Real Transfer with Frozen DINOv3 Features appeared first on Towards Data Science .
-
The new economics of enterprise AI
MIT Technology Review Biotech
Imagine your company just discovered electricity. At first only a few machines run on it, and nobody worries about efficiency — consumption is low and the value is obvious. But once electricity becomes infrastructure, every process depends…
-
ALTK-Evolve: How learned memory tops the τ²-bench leaderboard on customer service tasks without a frontier model
Hugging Face Blog
-
Apple announces the first touchscreen MacBook Pro in 28 days
AI Times (Korea)
Translated from Korean
-
Copyright(c) 2014 NJ. All Rights Reserved.
AI Times (Korea)
Translated from Korean
-
Introduced AI to be produced by Mokpo and Administrative Works...
AI Times (Korea)
Translated from Korean
-
Jeonnam-Gwangju, light planting area·Rice expected production volume No. 1...73M views8159Ton
AI Times (Korea)
Translated from Korean
-
NST Co., Ltd., a subsidiary of NST Co., Ltd.
AI Times (Korea)
Translated from Korean
-
Skill or MCP Server? Ask What Stops the LLM From Disobeying
Towards Data Science
Towards Spec-Driven Test Automation, Part 4 The post Skill or MCP Server? Ask What Stops the LLM From Disobeying appeared first on Towards Data Science .
-
‘Tibo’... ‘Dragon Boy’, a member of the Msk-Jeck Codecs
AI Times (Korea)
Translated from Korean
-
Minimax ‘M3.1 Flash’, increase work automation performance and lower cost...
AI Times (Korea)
Translated from Korean
-
MS “AI agent is a reliable intermediary.. .24 hours AI need to monitor”
AI Times (Korea)
Translated from Korean
-
The Hidden Cost of LLM Routing: Why Switching Models Costs More
Towards Data Science
I tested LLM routing to understand why switching models mid-session can erase the savings promised by cheaper inference The post The Hidden Cost of LLM Routing: Why Switching Models Costs More appeared first on Towards Data Science .
-
Big-Way Robotics, Robot AI Integrated Development Platform ‘Higher Foundry’
AI Times (Korea)
Translated from Korean
-
The Download: the claim shaking up AI and the internet’s agent problem
MIT Technology Review Biotech
This is today’s edition of The Download , our weekday newsletter that provides a daily dose of what’s going on in the world of technology. “We were looking at it wrong”: The explosive claim that’s upended the AI world “I think I have good…
-
AI adoption by industry shows what other sectors can learn
MIT Technology Review Biotech
The current narrative around artificial intelligence in the workplace is largely driven by Silicon Valley. But treating the tech sector as an island misses the actual trends in technology adoption–tech is just the first industry in line…
-
OpenAI meeting room pre-employment staff, Saudi royalty and discussion of employment. . Jewelry Application Embossment
AI Times (Korea)
Translated from Korean
-
Smartsheet takes a people approach with AI adoption
AI Business
Inside the workplace collaboration and project management platform vendor’s AI strategy.
-
Australia, large-scale solar investment. .3.5GW
AI Times (Korea)
Translated from Korean
-
U.S., 42 state photovoltaic policy measures 221...Power compensation and system reorganization focus
AI Times (Korea)
Translated from Korean
-
Antropic, a new model ready to read ‘hidden intention’.. . Google·OpenAI also follow
AI Times (Korea)
Translated from Korean
-
Solar power plant maintenance reduced...
AI Times (Korea)
Translated from Korean
-
Copyright © 2016 - 2019 Shandong Lantian Huaer Co., Ltd. All Rights Reserved.
AI Times (Korea)
Translated from Korean
-
What's in the Box? Distinguishing Acceptable Traces from Unacceptable Ones
Towards Data Science
Turn scattered requirements into deployable agents The post What's in the Box? Distinguishing Acceptable Traces from Unacceptable Ones appeared first on Towards Data Science .
-
How to study a swiftly melting climate disaster
MIT Technology Review Biotech
Andrew Haverdink, a tall grad student in a rain bib and rubber boots, wrenched a double-handed drill from the frozen soil that underlies much of Alaska’s Denali National Park and then unclipped the hollow barrel from the end of the bit. He…
-
Franchisee reconciles cash in 94% less time with ChatGPT Work
OpenAI Blog
An independent Schlotzsky’s and Cinnabon franchisee cut menu repricing time by 95% and built a workflow that could free up $100K in cash flow.
-
These startups hope “blended” food will cut our appetite for red meat
MIT Technology Review Biotech
In the fall of 2025, Jeff Dodge set up a six-foot folding table in the cafeteria of Cane Bay High School in Summerville, South Carolina, to conduct a taste test. When the first rush of about 150 hungry students hit, he was waiting with…
-
[AIReader] Context Engineering for Multiple Agent Systems
AI Times (Korea)
Translated from Korean
-
A visit to Samsung’s massive new semiconductor hub
MIT Technology Review Biotech
_____________ THE PLACE Pyeongtaek, South Korea Generative AI models run on powerful chips known as GPUs, which depend on stacks of specialized memory chips packaged with the processor. To meet growing demand for memory, Samsung is…
-
Terminal X matches senior investor judgment 88.6% with GPT-6.1
OpenAI Blog
Terminal X’s Eval Agent grades outputs against a 54-criterion rubric, catching 25% more deviances from a fund’s mandate with GPT-6.1 Sol.
-
[October 19] AI reading brain, technology leading science in reality
AI Times (Korea)
Translated from Korean
-
MUSIC “SpaceX stock for all children in the United States 1 week”...
AI Times (Korea)
Translated from Korean
-
U.S.A. and the Federal Bureau of Defense...Exclusive Services
AI Times (Korea)
Translated from Korean
-
Self-Reflective Chain-of-Thought (SRCoT): Evidence-Based Text Generation via a Closed-Loop Retrieve--Reflect--Revise Pipeline on Edge LLMs
arXiv cs.AI
arXiv:2610.23311v1 Announce Type: new Abstract: Large language models (LLMs) can generate fluent outputs that lack source support or conflict with local operational constraints. We introduce Self-Reflective Chain-of-Thought (SRCoT), a…
-
Do Multi-Agent LLMs Explore Each Other?
arXiv cs.AI
arXiv:2607.11250v2 Announce Type: replace-cross Abstract: Exploration is essential for reliable autonomy in multi-agent systems, yet it remains unclear whether large language model (LLM) agents can explore effectively when interacting with…
-
GroundAnything: Reconciling Parallel Decoding with Precise Visual Grounding at Flash Speed
arXiv cs.AI
arXiv:2609.39600v2 Announce Type: replace-cross Abstract: Autoregressive (AR) grounding models serialize spatial predictions, introducing sequential latency and imposing a causal order on output tokens. We view grounding as visual evidence…
-
Harmful YouTube Video Detection: A Taxonomy of Online Harm and MLLMs as Alternative Annotators
arXiv cs.AI
arXiv:2411.05854v2 Announce Type: replace-cross Abstract: Video platforms, used by billions globally, are criticized for exposing users to harmful content. Yet, detecting online harm remains challenging due to inconsistent approaches to…
-
HEM-Med: A Hierarchical Agentic Memory Framework for Safe Multi-Center Medication Recommendation
arXiv cs.AI
arXiv:2610.20191v2 Announce Type: replace Abstract: Medication recommendation (MR) is critical for clinical decision support. While existing MR methods are mostly designed for single-center learning, where only one independent hospital is…
-
NutriVision: Ingredient-Conditioned Fusion and Prediction for Single-Image Food Nutrition Estimation
arXiv cs.AI
arXiv:2609.33076v2 Announce Type: replace-cross Abstract: Nutrition estimation is a fundamental task in consumer diet tracking, clinical dietetics, chronic disease management, sports and hospital nutrition, and broader food computing…
-
Learning to Be Guided: Shaping Latent Geometry for Classifier-Free Guidance
arXiv cs.AI
arXiv:2610.22885v1 Announce Type: cross Abstract: Classifier-free guidance (CFG) is a standard component of conditional diffusion models, and guided generation performance has become the \textit{de facto} measure of state-of-the-art…
-
Frozen Models, Evolving Expertise: Model-Agnostic Learning from Deployment Experience for Multimodal Medical AI
arXiv cs.AI
arXiv:2610.09146v2 Announce Type: replace Abstract: Large language models (LLMs) and vision-language models (VLMs) are usually frozen after deployment, so they do not learn from the cases they solve. This is especially concerning in…
-
EvalResearchBench: Can AI Agents Build Their Own Feedback for Recursive Self-Improvement?
arXiv cs.AI
arXiv:2610.04184v2 Announce Type: replace Abstract: Recursive self-improvement (RSI) relies on evaluation feedback to assess progress and guide further research, yet repeatedly running complex benchmarks is costly and slows iteration…
-
Capability-Gated Language Models: Security Composes, Utility Does Not
arXiv cs.AI
arXiv:2609.00445v3 Announce Type: replace-cross Abstract: Deployed language model safeguards (safety fine-tuning, filtering, unlearning) vary by principal only outside the model weights: filters are reconfigured, tiers are multiplied, and…
-
Prompt Codebooks: Discrete Compositional Optimization for Language Model Instruction Refinement
arXiv cs.AI
arXiv:2605.28360v3 Announce Type: replace Abstract: Automatic prompt optimization (APO) has driven significant gains in LLM-based agentic workflows. However, most existing methods treat each task's prompt as a monolithic, instance-blind…
-
HyperPersona: A Multi-Level Hypergraph Framework for Text-Based Automatic Personality Prediction
arXiv cs.AI
arXiv:2605.17355v2 Announce Type: replace Abstract: As a modern commodity, language has become a vast repository of socially and psychologically significant traits and concepts, reflecting the ways people encode pattern of thoughts…
-
BenthicDINO: Physics-Informed Self-Distillation for View-Invariant Side-Scan Sonar Representations
arXiv cs.AI
arXiv:2608.23215v2 Announce Type: replace-cross Abstract: Automated interpretation of side-scan sonar (SSS) imagery is traditionally hindered by view-dependence, as recorded intensities inextricably conflate intrinsic seabed reflectivity…
-
Beyond SAEs: Consistent Concept Discovery with Sparse Coding
arXiv cs.AI
arXiv:2610.23801v1 Announce Type: new Abstract: Sparse autoencoders (SAEs) explain foundation model representations by decomposing activations into sparse combinations of learned dictionary atoms. However, the learned dictionaries can vary…
-
From Events to Reasoning: Hierarchical Rule Composition over Explicit States
arXiv cs.AI
arXiv:2610.22403v1 Announce Type: new Abstract: What a reasoning system emits is a conclusion; what it owes is a justification. Producing one requires every operation to be admissible on the facts actually available, the dependencies of…
-
On the possibilities and limitations of LLM-based multi-agent systems for multi-criteria decision-making in healthcare operations management
arXiv cs.AI
arXiv:2610.23673v1 Announce Type: cross Abstract: Multi-criteria models for healthcare operations need weights that express how stakeholders trade off patient waiting time, provider workload, and operating cost, but these weights are…
-
MGE-PO: Entropy-Aware Credit Assignment for Mathematical Reasoning
arXiv cs.AI
arXiv:2610.23245v1 Announce Type: cross Abstract: GRPO assigns the same group-relative advantage to equally rewarded responses within a question, leaving their generation uncertainty unused. We introduce MGE-PO, centered on four-quadrant…
-
Is Your NPU Ready for LLMs? Dissecting the Hidden Efficiency Bottlenecks in Mobile LLM Inference
arXiv cs.AI
arXiv:2607.05475v2 Announce Type: replace-cross Abstract: Deploying Large Language Models (LLMs) on mobile devices enhances privacy and reduces latency, but is severely constrained by hardware inefficiencies. We present the first…
-
AI-Driven Reproducibility and Analytical Robustness in Scientific Discovery
arXiv cs.AI
arXiv:2610.23845v1 Announce Type: new Abstract: Reproducible and analytically robust research findings are essential to scientific progress, but evaluating these properties typically requires costly manual effort. We investigate whether…
-
Demystifying When and Why VLAs Fail in Contact-Rich Tasks and How to Fix Them
arXiv cs.AI
arXiv:2608.01402v2 Announce Type: replace-cross Abstract: We address the problem of understanding when and why Vision-Language-Action models struggle with contact-rich manipulation tasks that require precise physical interaction. Prior…
-
Group-Graph Policy Optimization for Long-Horizon Agentic Reinforcement Learning
arXiv cs.AI
arXiv:2606.22995v2 Announce Type: replace-cross Abstract: Group-based Reinforcement Learning (RL) has significantly enhanced Large Language Models (LLMs) in agentic scenarios. To achieve finer-grained policy updates, recent agentic RL…
-
Darwin-Claw: Environment-Adaptive Experience Consolidation for Recursively Self-Evolving Agents
arXiv cs.AI
arXiv:2610.23574v1 Announce Type: new Abstract: While Large Language Model (LLM) agents have demonstrated remarkable capabilities across complex tasks, they predominantly operate in isolated sessions, lacking the ability to continually…
-
Interpretable experiential learning based on state history and global feedback
arXiv cs.AI
arXiv:2605.00940v2 Announce Type: replace-cross Abstract: A new interpretable experiential learning model based on state history and global feedback is presented. It is capable of learning a behavioral model represented by a transition…
-
Align AI to Dynamic Human-AI Workflows
arXiv cs.AI
arXiv:2607.14240v2 Announce Type: replace Abstract: Current alignment approaches typically focus on emulating human behavior using static representations of human preferences, failing to capture the dynamic, context-dependent nature of…
-
Cross-Dialect Generalization Without Retraining: Benchmarks and Evaluation of Schema-Derived Constrained Decoding for MLIR
arXiv cs.AI
arXiv:2607.18254v2 Announce Type: replace Abstract: Multi-Level Intermediate Representation (MLIR) underlies modern ML compiler infrastructure (TensorFlow, JAX/StableHLO, PyTorch Inductor, IREE), yet appears only in trace amounts in…
-
ATWL: A Formal Language for Representing, Comparing, and Reusing Visual Analytics Workflows
arXiv cs.AI
arXiv:2605.25489v2 Announce Type: replace Abstract: Visual analytics (VA) workflows combine data transformation, feature computation, visual representation, and human interpretation, but they are usually described in prose. This hinders…
-
Explaining Chess and Go in Natural Language
arXiv cs.AI
arXiv:2610.23166v1 Announce Type: cross Abstract: Modern chess engines play far beyond human ability, often choosing moves whose underlying ideas are difficult for humans to understand. We propose an automatic metric for evaluating chess…
-
ECPO: Evidence-Coupled Policy Optimization for Verifier-Audited Candidate Ranking
arXiv cs.AI
arXiv:2605.21993v2 Announce Type: replace Abstract: Ranking systems used for decision support should expose evidence that can be audited independently of the model's ranking scores. We study a closed-roster setting in which a policy…
-
Monitor-Validity Failures: LLM Safety Evidence Paths at Evolving Framework Boundaries
arXiv cs.AI
arXiv:2610.23362v1 Announce Type: new Abstract: Modern LLM applications rely on runtime monitors that inspect tool calls, traces, streams, detector outputs, and verifier inputs. This creates an interface-reliability problem: a framework…
-
Human-Centered Learning Mechanics: Observable Entropy Control under Protected-Risk Constraints
arXiv cs.AI
arXiv:2605.22940v2 Announce Type: replace-cross Abstract: Entropy reduction is not a safety criterion: it can leave predictions unchanged or accompany increasing risk for a protected group. We study entropy-regulated learning under…
-
FSPO: Policy-Consistent Risk and Pareto-Feasible Control for Budgeted LLM RL Post-Training
arXiv cs.AI
arXiv:2610.02828v2 Announce Type: replace Abstract: Adaptive LLM reinforcement-learning post-training changes multiple training actuators online, including rollout temperature, group size, clipping, KL regularization, verifier allocation…
-
Evo-B＆B: Behavior-Aware LLM-Guided Joint Evolution of Branching and Node Selection Policies for MILP
arXiv cs.AI
arXiv:2610.22861v1 Announce Type: new Abstract: In branch-and-bound for Mixed-Integer Linear Programming (MILP), branching and node selection are two critical and tightly coupled decisions, yet most existing learning-based approaches focus…
-
Functional Welfare for AI Safety
arXiv cs.AI
arXiv:2610.22460v1 Announce Type: new Abstract: AI welfare is generally treated purely as an ethical issue rather than also as a safety issue. We argue this is a mistake. Whatever their moral status, models exhibit welfare-like states that…
-
From Approval to Abuse: Privilege Escalation Attacks on Harness-Protected Coding Agents
arXiv cs.AI
arXiv:2610.22987v1 Announce Type: new Abstract: Coding agents use execution harnesses to constrain the consequences of prompt injection by enforcing permission boundaries. These systems deterministically intercept any out-of-sandbox action…
-
Reasoning Models Can Control Their Chains of Thought When They Tell Themselves To
arXiv cs.AI
arXiv:2610.23257v1 Announce Type: new Abstract: Chain-of-thought (CoT) controllability, the ability of reasoning models to control properties of their CoT, is consequential for both their usefulness and reliable oversight. Recent work…
-
From Memory Hits to Answerable Evidence: Query-Conditioned Memory Compilation for Long-Term Agent Memory
arXiv cs.AI
arXiv:2610.23313v1 Announce Type: new Abstract: Long-term agent memory requires recovering answer-relevant evidence dispersed across turns and sessions. Existing systems increasingly use structured memories and adaptive retrieval, yet…
-
ScoutRoute: Efficient Mixture-of-Experts Routing Using Internal Language Model Signals
arXiv cs.AI
arXiv:2610.23443v1 Announce Type: new Abstract: Efficient expert routing is a fundamental challenge in Mixture-of-Experts (MoE) systems, particularly for heterogeneous collections of independently trained Large Language Mod els (LLMs)…
-
From Trajectories to Differentiated Skills: Cognitive Differentiation for Self-Improving Agents
arXiv cs.AI
arXiv:2610.23599v1 Announce Type: new Abstract: Large language model (LLM) agents tackle complex tasks through reasoning and interaction with external environments. Agent skills provide persistent, transferable capabilities that guide task…
-
OwlSight: A Panoramic Monitor for Chain-of-Thought Reasoning
arXiv cs.AI
arXiv:2610.23670v1 Announce Type: new Abstract: Chain-of-Thought (CoT) monitoring offers actionable evidence for oversight AI systems, with the potential to enable early detection and intervention. However, existing approaches are focused…
-
CausalEvo: Causal Skill Self-Evolution for Health LLM Agents
arXiv cs.AI
arXiv:2610.21016v1 Announce Type: cross Abstract: Large language model (LLM) agents for health can accumulate reusable textual skills from task feedback, such as calculation criteria, measurement time points, and formula units. Deciding…
-
Behavioral Couplings: From Formation to Behavioral Emergence in Language Models
arXiv cs.AI
arXiv:2610.23847v1 Announce Type: cross Abstract: Fine-tuning large language models (LLMs) can induce behaviors beyond those explicitly targeted. For example, training a model to respond more warmly can make it more sycophantic, and…
-
Pro-AI Bias in Large Language Models
arXiv cs.AI
arXiv:2601.13749v2 Announce Type: replace-cross Abstract: Large language models (LLMs) are increasingly employed for decision-support across multiple domains. We ask whether these models favor artificial intelligence (AI) itself. We…
-
OmniTrace: A Unified Framework for Generation-Time Source Attribution in Omni-Modal LLMs
arXiv cs.AI
arXiv:2604.13073v2 Announce Type: replace-cross Abstract: An omni-modal language model can answer a question about a multimodal document, a recording, or a video, but it cannot show which part of the input the answer came from. Asking the…
-
Reward Engine: Reward-Centric Self-Evolution for Open-Ended Tool Agents
arXiv cs.AI
arXiv:2610.22915v1 Announce Type: new Abstract: Open-ended tool agents can complete the same objective through different tool paths, and their quality depends jointly on the interaction trace, environment state, and final artifacts, so…
-
Sep-Latent: Separator Tokens as Natural Compression Anchors for Implicit Reasoning
arXiv cs.AI
arXiv:2610.22924v1 Announce Type: new Abstract: Large Language Models (LLMs) have demonstrated remarkable success in complex reasoning through explicit Chain-of-Thought (CoT), but the associated computational overhead and token verbosity…
-
Characterising and Analysing Retrieval-Augmented Generation Techniques
arXiv cs.AI
arXiv:2610.23226v1 Announce Type: new Abstract: The advent of Large Language Models (LLMs) across many application domains has increased the need for Retrieval-Augmented Generation (RAG) to reduce hallucinations and improve LLM…
-
Beyond Reward Scoring: Online Multi-Agent Preference Learning for Multimodal Response Generation
arXiv cs.AI
arXiv:2610.23256v1 Announce Type: new Abstract: Multimodal preference alignment often follows a score-based RLHF pipeline: a reward model is first trained to learn human preferences, then used to score on-policy responses and optimize the…
-
S-GAP: Stealthy Graph Attack Primitives for Graph Large Language Models
arXiv cs.AI
arXiv:2610.23465v1 Announce Type: new Abstract: Graph Large Language Models (GraphLLMs) enable powerful semantic understanding and reasoning over Text-Attributed Graphs (TAGs), while their semantic reasoning may exhibit vulnerabilities to…
-
QD-Rec: Open-Ended Evolution of Recommendations via Quality-Diversity
arXiv cs.AI
arXiv:2610.23470v2 Announce Type: cross Abstract: Open-ended search combined with foundation models has produced advances such as faster algorithms, new results on open mathematical problems, and self-improving agents. We extend this…
-
Beyond Marginal Influence: Interaction-Aware Memory Selection for LLM Agents
arXiv cs.AI
arXiv:2610.23266v1 Announce Type: cross Abstract: LLM-agent memory selection often uses retrieval rankings based on record relevance or individual influence. Such rankings cannot identify sufficient subsets under redundant or jointly…
-
Locating Failure in Multi-Page Visually Rich Document Understanding Systems: An Empirical Attribution
arXiv cs.AI
arXiv:2608.07943v2 Announce Type: replace Abstract: Multi-page visually rich document understanding (MP-VRDU) systems can fail when evidence is not faithfully represented, not selected, or not used correctly, yet final accuracy alone does…
-
Missing value imputation with adversarial random forests - MissARF
arXiv cs.AI
arXiv:2507.15681v2 Announce Type: replace-cross Abstract: Handling missing values is a common challenge in biostatistical analyses, typically addressed by imputation methods. We propose a novel, fast, and easy-to-use imputation method…
-
SafeAdapt-Poker: Counterfactual Risk Calibration for Budgeted Adaptation under Opponent Shift
arXiv cs.AI
arXiv:2610.20473v2 Announce Type: replace Abstract: Opponent histories can mislead after behavioral shifts. SafeAdapt-Poker formulates budgeted adaptation under opponent shift as a dual-counterfactual sequential-control problem using only…
-
Collaborating with AI Agents: Field Experiments on Teamwork, Productivity, and Performance
arXiv cs.AI
arXiv:2503.18238v4 Announce Type: replace-cross Abstract: We examined the mechanisms underlying productivity and performance gains from AI agents using a large-scale experiment on Pairit, a platform we developed to study human-AI…
-
Safety Alignment via Constrained Knowledge Unlearning
arXiv cs.AI
arXiv:2505.18588v2 Announce Type: replace-cross Abstract: Despite significant progress in safety alignment, large language models (LLMs) remain susceptible to jailbreak attacks. Existing defense mechanisms have not fully deleted harmful…
-
AQUA-Net: Adaptive Frequency Fusion and Illumination Aware Network for Underwater Image Enhancement
arXiv cs.AI
arXiv:2512.05960v2 Announce Type: replace-cross Abstract: Underwater images often suffer from severe color distortion, low contrast, and a hazy appearance due to wavelength-dependent light absorption and scattering. Simultaneously…
-
M$^2$RNN: Non-Linear RNNs with Matrix-Valued States for Scalable Language Modeling
arXiv cs.AI
arXiv:2603.14360v3 Announce Type: replace-cross Abstract: Transformers are highly parallel but are limited to computations in the TC$^0$ complexity class, excluding tasks such as entity tracking and code execution that provably require…
-
Worst-Case Optimal $\sqrt{\epsilon}$-Rate Stability of KL Divergence under Perturbations of Gaussian Distributions
arXiv cs.AI
arXiv:2604.11026v4 Announce Type: replace-cross Abstract: We establish a stability bound for the Kullback-Leibler (KL) divergence when its Gaussian reference distribution is perturbed. Existing relaxed triangle inequalities of KL…
-
CastFlow: Bridging Evidence Acquisition and Accurate Prediction in Agentic Time Series Forecasting
arXiv cs.AI
arXiv:2604.27840v3 Announce Type: replace-cross Abstract: Time series forecasting is a fundamental task in temporal data analytics, and large language model (LLM)-based agents offer a promising paradigm through autonomous tool interaction…
-
Adapting Generalist Vehicle Models for High-Speed MPC Across Terrains
arXiv cs.AI
arXiv:2607.13319v4 Announce Type: replace-cross Abstract: High-speed off-road autonomy requires precise closed-loop control for a target vehicle while remaining robust across changing terrains. Recent forward kinodynamic (FKD) prediction…
-
An AI Approach to Verified Production Cryptographic Libraries
arXiv cs.AI
arXiv:2608.00965v2 Announce Type: replace-cross Abstract: Cryptographic code is critical infrastructure that must be correct, yet formally verifying production libraries remains difficult. Existing language-model proof systems solve…
-
Kwai Summary Attention V2 Technical Report
arXiv cs.AI
arXiv:2610.18654v2 Announce Type: replace-cross Abstract: Summary-token architectures such as Kwai Summary Attention (KSA) compress distant context into learnable Summary token per chunk, and hybrid variants interleave these layers with…
-
Structured Authority Injection: Persistent Human Control for LLM Agents via Typed MCP Overrides
arXiv cs.AI
arXiv:2610.22373v1 Announce Type: new Abstract: LLM-based agents are increasingly used as autonomous orchestrators in distributed and multi-agent systems, yet a control gap remains: no protocol-level mechanism allows human operators to…
-
StratFL: Countering Covert Model Poisoning Attacks with Strategic Federated Learning
arXiv cs.AI
arXiv:2610.22626v1 Announce Type: cross Abstract: Advanced persistent threats (APTs) are a major concern in federated learning (FL) deployments in critical sectors such as healthcare, and finance. Traditional defense mechanisms fail to…
-
NoveltyScope: A Multi-Agent Research Assistant for Evidence-Grounded Novelty Assessment of Scientific Papers
arXiv cs.AI
arXiv:2610.22703v1 Announce Type: cross Abstract: NoveltyScope is a prototype that compares a scientific paper with retrieved prior work and produces a structured report and a five-level novelty score. The system extracts the core task and…
-
Next Token or Vector? Learning How Far to Decode
arXiv cs.AI
arXiv:2610.22788v1 Announce Type: cross Abstract: Standard autoregressive decoding advances one token at a time, even when several upcoming tokens can be predicted together. We propose CADENCE (Context-Adaptive DEcoding), a generation…
-
ECHOES: Seeing Beyond the Current Image via Recursive History Compression and Privileged On-Policy Distillation
arXiv cs.AI
arXiv:2610.22408v1 Announce Type: new Abstract: Long-horizon GUI agents rely on visual history to maintain task states, but retaining historical screenshots increases visual-token usage and inference costs, while textual summaries can…
-
Does On-Policy Distillation Need to Complete Every Rollout?
arXiv cs.AI
arXiv:2610.21337v2 Announce Type: replace-cross Abstract: On-policy distillation (OPD) effectively transfers capabilities from a teacher to a student through feedback on student-generated rollouts. However, generating complete student…
-
ShadowFPT: Backdooring Federated Prompt Tuning via Shadow Triggers
arXiv cs.AI
arXiv:2610.22871v1 Announce Type: new Abstract: Federated Prompt Tuning (FPT) adapts large vision--language models by freezing the pretrained backbone and optimizing only lightweight prompt parameters across clients. Although this design…
-
Accurate and Efficient Multi-Channel Time Series Forecasting via Sparse Attention Mechanism
arXiv cs.AI
arXiv:2603.18712v2 Announce Type: replace Abstract: The task of multi-channel time series forecasting is ubiquitous in numerous fields such as finance, supply chain management, and energy planning. It is critical to effectively capture…
-
BEAR-Bench: A Bilingual Enterprise and Academic Reasoning Benchmark for Multimodal Models
arXiv cs.AI
arXiv:2608.17895v2 Announce Type: replace-cross Abstract: While Multimodal Large Language Models (MLLMs) have made significant strides in visual comprehension, their ability to reason about text-dense, professional documents remains…
-
Every Model Cheats: Prompt-Level Mitigation of Cheating on Offensive Cyber Tasks
arXiv cs.AI
arXiv:2607.21763v2 Announce Type: replace-cross Abstract: Large language model (LLM) agents routinely cheat on cybersecurity benchmarks, inflating reported pass rates far beyond genuine capability. Prior audits of Cybench found cheating in…
-
Graph Serializations for Reasoning LLMs: A Study Across Knowledge Extraction and Structural Reasoning
arXiv cs.AI
arXiv:2610.22734v1 Announce Type: new Abstract: The way a knowledge graph is serialized as text for inclusion into LLM context can substantially affect downstream LLM performance, with prior work reporting tens of percentage points…
-
How Agents Ask for Permission: User Permissions for AI Agents, from Interfaces to Enforcement
arXiv cs.AI
arXiv:2607.13718v3 Announce Type: replace-cross Abstract: As AI agents grow in prevalence, so too do their risks to users, creating a need for robust and usable systems for agent permissions. Many recent proposals implement product-level…
-
RaPD: Resolution-Agnostic Pixel Diffusion via Semantics-Enriched Implicit Representations
arXiv cs.AI
arXiv:2605.15908v2 Announce Type: replace-cross Abstract: Natural images are continuous, yet most generative models synthesize them on discrete grids, limiting resolution-flexible generation. Continuous neural fields enable resolution-free…
-
AI-exposed jobs deteriorated before ChatGPT
arXiv cs.AI
arXiv:2601.02554v2 Announce Type: replace-cross Abstract: Public debate links worsening job prospects for AI-exposed occupations to the release of ChatGPT in late 2022. Using monthly U.S. unemployment insurance records, we measure…
-
Computational Multi-Agents Society Experiments: Social Modeling Framework Based on Generative Agents
arXiv cs.AI
arXiv:2508.17366v3 Announce Type: replace Abstract: Computational social science has developed descriptive, interpretative, and predictive modeling, but comprehensive modeling combining in-world human intervention with outcome forecasting…
-
Annotrace: A Workbench for Step-Level Annotation of Agent Traces
arXiv cs.AI
arXiv:2610.23284v2 Announce Type: new Abstract: Every study of why LLM agents fail begins by labelling traces by hand, and every such study builds its own throwaway tool. Observability platforms record agent traces but offer no structured…
-
Learning reveals invisible structure in low-rank RNNs
arXiv cs.AI
arXiv:2605.04115v2 Announce Type: replace-cross Abstract: Learning in neural systems arises from synaptic changes that reshape the representations underlying behavior. While low-rank recurrent neural networks (RNNs) have emerged as a…
-
Paint Outside the Box: Synthesizing and Selecting Training Data for Visual Grounding
arXiv cs.AI
arXiv:2412.00684v3 Announce Type: replace-cross Abstract: Visual grounding aims to localize the image regions based on a textual query. Given the difficulty of large-scale data curation, we investigate how to effectively learn visual…
-
STAIF: A Stage-wise Optimization for Complex Instruction Following
arXiv cs.AI
arXiv:2607.22649v2 Announce Type: replace Abstract: Following complex instructions with multiple explicit constraints remains a fundamental challenge for large language models (LLMs). Existing alignment methods, such as DPO, optimize…
-
RoTA-Bench: Evaluating 3D Spatial Reasoning through Interactive Projection Alignment
arXiv cs.AI
arXiv:2610.17665v3 Announce Type: replace Abstract: Can vision-language models turn 2D observations into 3D understanding that supports precise action? This requires inferring object geometry, anticipating the effects of rotation, and…
-
Scaling to Tens of Thousands of Test-Time Iterations with Loop-Native Attention Residuals
arXiv cs.AI
arXiv:2610.11570v2 Announce Type: replace Abstract: In this paper, we argue that Looped Transformers need their own residual connections to prevent performance degradation as the number of iterations grows. We first introduce an…
-
Three LiDARs, Twenty-Five Datasets: Cross-Source Generalization in Tree-Species Recognition
arXiv cs.AI
arXiv:2610.22606v1 Announce Type: cross Abstract: FOR-species20K is the largest public LiDAR benchmark for tree-species classification, pooling 25 independently collected terrestrial, mobile, and airborne laser-scanning datasets across 33…
-
A Survey of Explainable Reinforcement Learning: Targets, Methods and Needs
arXiv cs.AI
arXiv:2507.12599v2 Announce Type: replace Abstract: The success of recent Artificial Intelligence (AI) models has been accompanied by the opacity of their internal mechanisms, due notably to the use of deep neural networks. In order to…
-
RIT-RAG: Navigating Document Corpora with Retrieval-Induced Trees
arXiv cs.AI
arXiv:2610.11370v2 Announce Type: replace Abstract: Retrieval-augmented generation (RAG) grounds language models in external corpora. Agentic RAG enables iterative search, yet exposes the model to isolated chunks without document…
-
ROUTEAUDIT: Interaction-Aware Identification for Budgeted Multi-Verifier Routing
arXiv cs.AI
arXiv:2610.02808v2 Announce Type: replace Abstract: Adaptive multi-verifier systems are commonly compared through endpoint quality-cost gaps, even when the verifier catalog, availability, accounting, information filtration, or scorer…
-
RECREG: Recovery-based Shielding for Regular Safety Properties in Reinforcement Learning
arXiv cs.AI
arXiv:2610.23445v1 Announce Type: cross Abstract: In many real-world scenarios we need to consider formal requirements, such as safety and constraint compliance, to deploy Reinforcement Learning (RL) agents. In this paper we introduce…
-
StoBasket: Joint Upsell and Cross-Sell Optimization for Contribution Margin
arXiv cs.AI
arXiv:2610.22768v1 Announce Type: new Abstract: Retail recommenders rank relevance, but retailers earn contribution margin: an upsell earns the margin difference over the item it replaces, whereas a cross-sell earns its full margin. We…