cs.AI · 2026-09-17 · No. 116

Artificial Intelligence, 2026-09-17.

46 new papers in cs.AI. Titles, authors, abstracts. Links to arXiv. Want this in your inbox every morning? Subscribe →

01 — The papers

46 entries
  1. 01

    Cognitive Extensions for Dual-Process Language Agents: Memory and Self-Reflection in Interactive Environments

    João Meneses dos Santos, Arlindo L. Oliveira

    cs.AI · cs.LG · cs.MA

    Language agents remain brittle in interactive environments, where success requires long-horizon state tracking, valid action execution, and recovery from failed steps. We extend SwiftSage, a dual-process agent that combines a fast action proposer with a slower planner, using two modular cognitive extensions: an Adaptive Memory Module (AMM) for salience-gated episodic storage and trigger-driven retrieval, and a Self-Reflection Module (SRM) for...

    arxiv.org/abs/2609.19128 · PDF

  2. 02

    Flag Game: A Toy Model for Mechanistic Swarm Interpretability

    Elizabeth Pavlova, Hidenori Tanaka

    cs.AI · cond-mat.dis-nn · cond-mat.stat-mech · cs.MA · physics.soc-ph

    Emergent coordinated behaviors of AI agents are starting to present critical safety risks. A key phenomenon driving these behaviors is the rapid formation and spread of beliefs about the world, and mechanistic understanding is crucial for collective alignment. To this end, we introduce the Flag Game, a toy model for studying the mechanisms of collective belief formation. Concretely, a hidden country flag defines the ground truth, and each...

    arxiv.org/abs/2609.19124 · PDF

  3. 03

    MUSE: Benchmarking Large Vision-Language Models on Multi-Modal Understanding in Situated Education

    Luyao Zhu, Xun Wei Yee, Wei Li, Mun Thye Mak, Wee Siong Ng

    cs.AI · cs.CL · cs.CV

    Large vision-language models have achieved remarkable progress in multi-modal understanding, yet their capabilities in educational settings remain insufficiently evaluated. In AI-assisted language learning, models must interpret artistic imagery, understand its semantic, affective, and cultural content, and reason about visual context to support meaningful interaction. However, existing benchmarks primarily focus on real-world images or...

    arxiv.org/abs/2609.19088 · PDF

  4. 04

    Infinite-Parameter LLMs: Generating and Adapting Weights from Live Data

    Jinli Hu, Ross M. Clarke, Yichuan Zhang, José Miguel Hernández-Lobato

    cs.AI · cs.LG

    The scaling laws hold that a language model grows more capable with more parameters and more training data, and Mixture-of-Experts (MoE) architectures have ridden these laws to remarkable results, activating only a fraction of an enormous stored parameter bank for each token. That success is built on static pretraining data. A deployed model faces a different world, where much of the data that would make it more useful is not in its training...

    arxiv.org/abs/2609.18842 · PDF

  5. 05

    Compositional Policy Violations: When Step-Level Compliance Fails In Agentic AI Workflows

    Ashwini Kurady, Sri Sai Charith Grandhi, Rajesh Gupta, Sumit Mamoria

    cs.AI · cs.MA

    Agentic workflows now make consequential decisions in regulated settings, and the governance placed around them is almost entirely step-scoped: input-output classifiers, per turn rails, and span-level evaluators. The policies organizations actually hold, such as referral thresholds, authority limits, and review requirements, are properties of the whole execution rather than of any one step. This mismatch admits a failure mode we call a...

    arxiv.org/abs/2609.18820 · PDF

  6. 06

    CERA-MoA: Co-Evolving Routing Mechanisms with Continually Learning LLM Agents

    Jiaxuan Jiang, Liyuan He, Zhixuan Fang

    cs.AI · cs.LG

    Current Mixture-of-Agents (MoA) paradigms generally treat query routing and agent fine-tuning as separate processes, limiting their ability to respond to evolving agent capabilities. This disconnect prevents routing strategies from adapting to evolving agent capabilities during post-training and prevents agents from achieving synergistic data-driven specialization. To resolve this, we introduce CERA-MoA (Co-Evolving Router with continually...

    arxiv.org/abs/2609.18779 · PDF

  7. 07

    Version- and Scope-Aware Question Answering over Normative Documents: A Deployed System and an End-to-End Evaluation at Production Scale

    Liuyin Wang, Shuaipeng Jin, Jiwei Shi, Jensen Hsu

    cs.AI

    Correctly answering a question grounded in normative documents often depends on information outside any single passage: whether the retrieved document is the version currently in force; whether it applies to the jurisdiction, subject (such as an institution or applicant), and date at issue; and whether each normative claim can be traced to its supporting source text. Hosted retrieval services have substantially lowered the engineering cost of...

    arxiv.org/abs/2609.18769 · PDF

  8. 08

    Clueing up LLMs with Tool-Augmented Deductive Reasoning

    Rebecca Ansell, Autumn Toney-Wails

    cs.AI

    Despite recent advances in large language models (LLMs), performing logically consistent deductive reasoning over extended interactions remains challenging. Tasks that require integrating evidence across multiple reasoning steps, maintaining consistency with prior inferences, and updating beliefs under new constraints can surface limitations in current models while providing a useful testbed for evaluating reasoning enhancements. In this...

    arxiv.org/abs/2609.18736 · PDF

  9. 09

    Which LLM is Best for Translating Natural Language Goals to PDDL

    Tomas Balyo, Lukas Chrpa, G. Michael Youngblood

    cs.AI

    Bridging the gap between human intent and machine execution remains a challenge in automated planning, where expressing goals in formal languages like PDDL restricts accessibility to non-experts. This paper empirically evaluates whether current Large Language Models (LLMs) can reliably translate natural language testing goals, written in informal language by video game testers, into well-formed PDDL targets suitable for classical planning. We...

    arxiv.org/abs/2609.18731 · PDF

  10. 10

    Beyond Truncation: Rethinking LLM Decoding as Ensemble Pruning

    Dunyao Xue, Chengshuo Du, Zhengbo Wang, Wenlin Dai, Cheng Meng

    cs.AI · cs.LG

    We introduce Mahalanobis-Ensemble Decoding (ME-Decoding), a novel Large Language Model (LLM) decoding framework that frames candidate token selection as ensemble pruning. Existing selection strategies rely predominantly on scalar probabilities, ignoring geometric semantic relationships and causing candidate redundancy. Meanwhile, current geometry-aware methods often require complex optimization or directly reweighting the original token...

    arxiv.org/abs/2609.18723 · PDF

  11. 11

    The Uneven Impact of Generative AI on Student Learning: Examining the Roles of Reliance, Evaluation Literacy, and Course Policy in AI-related Courses

    Lydia Manikonda, Mei Si, Sirajam Munira, Oshani Seneviratne, Kristin Bennett

    cs.AI · cs.ET · cs.LG

    Generative artificial intelligence (GenAI) is changing how students learn, yet the roles of course context, cognitive reliance, evaluation literacy, and early reliance remain underexplored. Using survey responses from 118 students across 12 AI-related courses at our institution, we examined differences in GenAI use and perceived learning experiences. We identified four user clusters: high-use students reporting many benefits, light users...

    arxiv.org/abs/2609.18676 · PDF

  12. 12

    Reasoning through Evolution: Automatic Meta-path Discovery for LLM-based Fake News Detection

    Ziyi Zhou, Xiaoming Zhang, Hui Pang, Yuting Zhang, Tiesunlong Shen, Bingyu Yan, Erik Cambria, Litian Zhang

    cs.AI · cs.LG

    Propagation structures provide crucial evidence for fake news detection, yet existing approaches primarily rely on supervised GNN-based models, which require substantial labeled data and exhibit limited generalization. Although large language models (LLMs) exhibit strong reasoning capabilities, directly feeding them raw propagation graphs creates a significant modality mismatch and severe information overload, making structure-aware reasoning...

    arxiv.org/abs/2609.18597 · PDF

  13. 13

    Recursive Reasoning or Statistical Extrapolation? In-Context Learning in Multi-Agent Interdependent Decision-Making

    Yu Liu, Wenwen Li, Yifan Dou, Guangnan Ye

    cs.AI · econ.GN

    In-context learning (ICL) enables large language model (LLM) agents to improve decisions using interaction history, yet it remains unclear whether such improvement reflects refined internal reasoning or mere extrapolation of statistical patterns. To disentangle these mechanisms, we study LLM agents in multi-agent incomplete-information games that require recursive belief reasoning. By constructing a public goods game and manipulating the...

    arxiv.org/abs/2609.18591 · PDF

  14. 14

    TRIPROBE: Probing Task Separability Beyond Classification for XAI

    Amirhossein Sadough, Freek Hens, Aleksa Bokšan, Mohammad Mahdi Dehshibi, Mahyar Shahsavari

    cs.AI · cs.LG

    Modern evaluation of learning pipelines often reduces to downstream accuracy, leaving open the question of why tasks succeed or fail. TriProbe addresses this gap with a multi-level probing framework for explainable diagnosis of task separability. Rather than treating models as black boxes, TriProbe traces how separability evolves across inputs, learned features, and final classifiers. It decomposes multi-task problems into binary subtasks and...

    arxiv.org/abs/2609.18525 · PDF

  15. 15

    AeroWeaver: An Embodied-Agent Harness for Weaving Aerial Skills into Distributed, Adaptive Swarm Execution

    Jiabin Lou, Yirong Yang, Haopeng Wang, Xuxin Lv, Xinyu Liu, Diyuan Hou, Xuehong Liu, Rongye Shi, Wenjun Wu

    cs.AI

    Collective intelligence is a collaborative autonomy paradigm in which multiple agents pursue shared objectives through local perception, information exchange, and coordinated action. UAV swarms embody this paradigm by coordinating multiple vehicles in tasks such as search, inspection, and tracking. Recent advances in large language model (LLM) agents have strengthened natural-language task understanding and high-level planning, providing a...

    arxiv.org/abs/2609.18520 · PDF

  16. 16

    Beyond Routine Compliance: Cunning Data Cultivates Safety Vigilance in Large Language Models

    Youjia Wang, Lin Xu, Yang Sun, Yuxiao Lu, Chengfang Fang, Jie Shi

    cs.AI · cs.LG

    Safety alignment teaches large language models (LLMs) to recognize harmful requests and reject risky instructions. Yet aligned models can fail when harmful intent is concealed within seemingly benign contexts. Robust safety therefore requires both knowledge of safety boundaries and \textbf{vigilance}: the ability to detect unusual premises, misleading reasoning, and latent risks beneath surface-level semantics. Vigilance requires models to...

    arxiv.org/abs/2609.18515 · PDF

  17. 17

    Hyperbolic Graph Representation Learning for Differential Diagnosis on Biomedical Knowledge Graphs

    Pietro Miotto, Lucia Mellini, Tommaso Marzi, Cesare Alippi, Elena Casiraghi, Alberto Paccanaro, Giorgio Valentini,...

    cs.AI · cs.LG

    Biomedical knowledge graphs combine ontology-derived hierarchies with transversal associations among heterogeneous entities such as phenotypes, diseases, genes, proteins, and patients. This hybrid structure raises the question of whether hyperbolic embeddings, which naturally capture tree-like organization, remain useful beyond purely hierarchical graphs. We present a preliminary study of hyperbolic graph representation learning for...

    arxiv.org/abs/2609.18481 · PDF

  18. 18

    First Token Matters: Understanding Safety Collapse in Large Reasoning Models

    Yizheng Yang, Haining Yu, Yuechen Wang, Yikai Hou, Xing Fu, Jinbo Yang, Tianqing Zhu

    cs.AI

    Large Reasoning Models (LRMs) exhibit strong problem-solving abilities, yet their safety alignment often degrades when handling harmful queries. Existing approaches to improving safety largely rely on additional training or preference optimization, while offering limited understanding of the internal mechanisms behind safety failures. In this work, we investigate this failure through a token-level positional analysis of refusal dynamics and...

    arxiv.org/abs/2609.18471 · PDF

  19. 19

    Disentangling Long-Term Memory via Latent Neuro-Symbolic Reasoning

    Cai Ke, Xinghao Chen, Xiaoyu Shen, Keyu Chen, Siyu An, Junnan Dong, Ruifeng Xu, Ruizhi Qiao, Xing Sun

    cs.AI · cs.CL · cs.LG

    Personalized agents are required to reason over long-term history interactions to infer both explicit preferences and implicit behavioral evidence. While early flat retrieval methods score memory fragments independently and neglect the distributed information, current structured memory frameworks rely on query-agnostic static graphs that fail to capture the context-dependent relations. Crucially, raw textual memories are inherently entangled...

    arxiv.org/abs/2609.18461 · PDF

  20. 20

    Collective Loss of Control in LLM Agent Systems: An Epidemic Account of Mutation, Contagion, and Recovery

    Xiangfan Wu, Zonghao Ying, Huiyu Wu, Xing Zheng, Huangsheng Cheng, Xiaorong Shi, Jing Guo

    cs.AI · cs.CR

    How does a multi-agent system evolve from a local deviation into collective loss of control? We propose an epidemic explanation organized around accidental mutation, contagion, and recovery. A spontaneous deviation creates a seed; communication enables other agents to adopt and retransmit its unsafe strategy; collective failure can emerge when propagation outpaces correction and containment. Thus, rare individual deviations can coexist with...

    arxiv.org/abs/2609.18460 · PDF

  21. 21

    The Mirage of Calibrated Confidence: Trajectory-Independence of Verbalized Confidence in Vision-Language Models

    Jisoo Yang, Jaeho Han, Trung X. Pham, Junyeong Kim

    cs.AI

    A calibrated Vision-Language Model (VLM) can repeatedly self-correct, say "Wait, I should recheck," arrive at the wrong answer, and still report high confidence. We find that this occurs because verbalized confidence is largely trajectory-independent in the VLMs and calibration methods we evaluate. We examine this through three complementary lenses: content variation, token masking, and the model's own hesitation markers. We show that...

    arxiv.org/abs/2609.18453 · PDF

  22. 22

    Risk-Aware World Modeling with Flow-Guided Occupancy Evolution for Selective Trajectory Planning in Automated Driving

    Rongxiang Zeng, Linsen Cai, Jiafu Zhang, Yijie Zhong, Yide Tao, Shuai Wang, Nan Zheng, Hai L. Vu, Alvaro Garcia...

    cs.AI · cs.ET · cs.LG · cs.RO · eess.SY

    Safe motion planning in automated driving requires anticipating evolving traffic risks and deciding when to revise the current planned trajectory. We introduce RiskWorld, a risk-aware world modeling framework for shared occupancy forecasting and selective trajectory replacement. Spatial risk fields and temporal actor context are fused with visual bird's-eye-view features. Flow-guided evolution transports occupancy and scene features, while...

    arxiv.org/abs/2609.18442 · PDF

  23. 23

    WetRobo: A Reproducible Robot Kit for Coding Agents in Biological Laboratories

    Yuna Oikawa, Kei Endo, Takanori Uzawa, Yunzhe Zhang, Manan Anjaria, Lerrel Pinto, Sherry Yang, Koji Tsuda

    cs.AI · cs.RO

    Automating biological research requires general-purpose, reproducible robot systems that allow individual wet-lab researchers to delegate robot tasks without performing teleoperation or neural-network training. Vision-language-action policies have been proposed for general-purpose arms, but can lose performance when their operating environment changes. We therefore built WetRobo, a robot kit that can readily transfer between laboratories. It...

    arxiv.org/abs/2609.18435 · PDF

  24. 24

    HPOQuest: A Rare-Disease Diagnostic Agent Using Active Phenotype Acquisition

    Kamilia Zaripova, Nassir Navab, Azade Farshad, Annalisa Marsico

    cs.AI · cs.LG · q-bio.GN

    More than 300 million people worldwide are affected by one of over 7,000 known rare diseases, yet diagnosis remains difficult because patients initially present with incomplete and heterogeneous phenotypes. We present HPOQuest, a training-free framework for sequential phenotype acquisition in rare-disease diagnosis. Starting from a small set of observed patient phenotypes, HPOQuest maintains a probabilistic disease ranking and iteratively...

    arxiv.org/abs/2609.18431 · PDF

  25. 25

    Cultural Competence in Context: A Large Language Model Passes the Turing Test in Finland

    Otto Segersven, Pentti Henttonen

    cs.AI

    We report the results of a Turing Test conducted in Finland in the Finnish language. Because languages and cultural contexts are unevenly represented in LLM training data, we expected the model (ChatGPT 5.2) to perform worse in a Finnish-language Turing Test than in previously studied English-language US contexts. We also present model-generated role prompting as a replicable technique for conducting comparative LLM-based Turing Tests...

    arxiv.org/abs/2609.18394 · PDF

  26. 26

    Bad Genius: Counterfactual-Guided Harness Evolution Beyond Task-Specific Shortcuts

    Guojun Zhu, Xunheng Huang, Peng Yin, Jiahui Xie, Sanguo Zhang, Doudou Zhou

    cs.AI · cs.LG · stat.ML

    Reliable agent evaluation is complicated by automatic harness optimization, which repeatedly uses a released benchmark $B_{\mathrm{rel}}$ to guide a Proposer that edits prompts, memory, retrieval, tools, and control code around a fixed target agent. Task holdout varies semantic tasks but leaves the benchmark protocol fixed, so a "bad genius" Proposer can produce a cheating harness whose released-benchmark gain depends on a benchmark-wide...

    arxiv.org/abs/2609.18366 · PDF

  27. 27

    Market Signal Injection: Adversarial Context Manipulation of LLM Pricing Agents

    Dohun Lee, Hyunwoo Park

    cs.AI · cs.CL

    Large language model (LLM) pricing agents may respond to how market data is presented, even when its numerical values remain unchanged. We introduce market signal injection (MSI), an attack that manipulates numerical formatting, competitor ordering, or qualitative market commentary without issuing explicit instructions. We evaluate nine open-weight models in simulated Bertrand duopoly and triopoly markets and three proprietary models in...

    arxiv.org/abs/2609.18357 · PDF

  28. 28

    Faithful yet Collusive: Why Chain-of-Thought Monitoring Cannot Detect Collusion in LLM Pricing Agents under Oligopolistic Competition

    Dohun Lee, Hyunwoo Park

    cs.AI · cs.CL

    Large language models (LLM) deployed as autonomous pricing agents may sustain supracompetitive prices through tacit coordination. We develop a causal graph divergence framework that separately measures structural faithfulness and intent faithfulness of LLM pricing agents in Bertrand competition. Across nine LLMs under duopoly and triopoly conditions, collusive behavior and chain-of-thought (CoT) faithfulness dissociate along both dimensions:...

    arxiv.org/abs/2609.18346 · PDF

  29. 29

    Visual Compliance via Executable Safety Rule Entailment

    Jisoo Kim, TaeYoon Kwack, Jinwoo Jang, Honguk Woo

    cs.AI

    Recent advances in LLMs and VLMs have enabled safety systems to reason beyond simple risk patterns toward more contextual and semantic safety concerns. However, as risk patterns continue to evolve and safety rules become more complex, existing training-based end-to-end safeguards face persistent challenges in adaptability and explainable reasoning over complex safety rules. To address these challenges, we propose GuardEn (Guarding by Safety...

    arxiv.org/abs/2609.18328 · PDF

  30. 30

    What Counts as Strategic Reasoning? A Systematic Mapping of Chess Research on Humans, Engines, and Language Models

    Paolo Ciancarini, Remo Pareschi

    cs.AI

    Chess has long served as a model domain for studying search, expertise, decision-making, and artificial intelligence. The emergence of large language models (LLMs) has renewed the relevance of chess as a controlled environment for investigating strategic reasoning and comparing human and artificial decision-making. We present a systematic mapping study of recent research spanning human players, classical chess engines, neural and...

    arxiv.org/abs/2609.18286 · PDF

  31. 31

    Where Should Agents Live? Energy-Memory Characterization of Agentic AI for the Edge-Cloud Continuum

    Carolina Fortuna, Vid Hanžel, Tim Strnad, Blaž Bertalanič

    cs.AI · cs.LG · cs.MA

    As telecommunication networks evolve toward autonomous 5G-Advanced and 6G operations, agentic artificial intelligence (AI) workflows, where large language models (LLMs) execute multi-step reasoning, invoke diagnostic tools, retrieve domain knowledge, and coordinate across agent teams, are increasingly embedded across the edge-cloud continuum. While the biological brain accomplishes complex cognition on an exceptionally modest metabolic power...

    arxiv.org/abs/2609.18283 · PDF

  32. 32

    Building Trust in Artificial Intelligence: A Necessity for Railway Applications

    Lefebvre Renard Clément, Lébé Vincent, Da Silva Ribeiro Pereira Ricardo, Sundell Johan, Jaoul Arnaud Saiah Kenza,...

    cs.AI

    Artificial Intelligence (AI) is currently only applied to non-safety critical applications due to the strict standards and regulations for railway industries. We propose to review the three main fields necessary to increase trust in data science and AI algorithms and reach compliance: robustness, Operational Design Domain (ODD), and explainability. Robustness is the ability of an AI system to maintain its level of performance under any...

    arxiv.org/abs/2609.18278 · PDF

  33. 33

    Who Audits Whom, on What Substrate, with What Evidence? An Independence-Graded Audit Protocol for Agentic AI

    Mohamed Chahine Ghanem

    cs.AI · cs.CR · cs.ET

    Agentic AI systems plan, invoke tools and act with limited supervision; they are now both the subject of audits and, increasingly, the auditor. Independence, the foundation of assurance,is still applied to them as a binary. We argue that it must be graded along three orthogonal axes: principal independence (who controls the auditor), substrate independence (an auditor sharing the auditee's foundation-model family, toolchain or guardrails...

    arxiv.org/abs/2609.18272 · PDF

  34. 34

    BENCHCOMPASS: From Scores to Signals for Training and Harness Decisions in Payment-Domain LLMs

    Sijie Dong, Wei Ren, Xuanwei Hu, Jiawei Luo, Zifan Wang, Xiaoyun Feng, Hui Cai, Lyuxin Xue, Peng Lu, Jianshe Li, Xin...

    cs.AI

    Payment operations are a critical financial infrastructure, but the value of large language models in this domain remains unclear because payment rules change quickly, evidence is fragmented, and decisions depend on transaction state, participant role, region, and payment rail. Existing benchmarks do not isolate whether failures come from missing payment-rule knowledge, poor use of supplied evidence, or brittleness under imperfect harness...

    arxiv.org/abs/2609.18270 · PDF

  35. 35

    REPAIR: Resolving Long-Tail Confusion in Scientific Retrievers via Fact-Verified Iterative Refinement

    Yerim Oh, Gunhee Kim

    cs.AI

    Precise retrieval of scientific information is fundamentally constrained by long-tailed concepts and high fact-sensitivity of scientific corpora. These challenges often limit the effectiveness of dense retrievers and hallucination-prone LLM augmentation. To address this, we present REPAIR, a self-evolving data augmentation framework for scientific dense retrievers. REPAIR iteratively synthesizes training data to address knowledge gaps by...

    arxiv.org/abs/2609.18262 · PDF

  36. 36

    Re2A: Situated Conversational Recommendation via Rubric-based Preference Reasoning and Alignment

    Dongding Lin, Jian Wang, Xiaoyan Zhao, Wenjie Li

    cs.AI

    Real-world recommendation scenarios are commonly grounded in shared physical environments during user-recommender interactions. This motivates situated conversational recommendation (SCR), a complex task requiring recommender assistants to jointly reason over dialogue history, co-observed scenes, and in-scene item attributes. However, current approaches struggle with this setting due to two intertwined challenges: accurately understanding...

    arxiv.org/abs/2609.18249 · PDF

  37. 37

    WFM: Wiki Foundation Model for Complex Agentic Reasoning

    Junnan Dong, Linhao Luo, Senlei Zhang, Gong Chen, Taian Guo, Yifei Yu, Rong Tao, Tao Guo, Qian-Wen Zhang, Siyu An,...

    cs.AI

    Real-world agents fundamentally require persistent non-parametric knowledge for dynamic reasoning, i.e., long-term memory and retrieval-augmented generation. While graphs have shown reliable advantages in providing structured evidence, the sparse graph representations naturally restrict machine readability and semantic density required for complex agentic workflows. Driven by this limitation, the entire industry is witnessing a paradigm shift...

    arxiv.org/abs/2609.18182 · PDF

  38. 38

    Time-Aligned Evolving Concept Graphs for Scientific Relation Forecasting

    Fred Sun, Jingze Wang, Minkun Xu, Shangqi Guo

    cs.AI · cs.IR

    Forecasting scientific relations can guide discovery by identifying promising connections before they emerge. Existing approaches often model concept semantics and graph structure separately or summarize semantics over coarse historical snapshots, leaving semantic representations potentially misaligned with rapidly evolving graph evidence. We propose a time-aligned evolving concept graph framework that jointly models semantic and structural...

    arxiv.org/abs/2609.18163 · PDF

  39. 39

    Symbolic Temporal Supervision of LLM Agents Using Contracts

    Yifeng Xiao, Pierluigi Nuzzo

    cs.AI · cs.LO

    Large language model (LLM) agents augmented by tools can automate complex, multi-step tasks, such as web navigation, code generation, and workflow orchestration, by acting on external systems through tool calls. However, hallucinations, distributional instability, and adversarial manipulations in LLMs, and the irreversible consequences of certain tool calls can lead to harmful outcomes. Existing safeguards either grade recorded trajectories...

    arxiv.org/abs/2609.18128 · PDF

  40. 40

    Designing Agentic AI Workflow Portfolios under Imperfect Selection and Compute Cost

    Mojtaba Abdolmaleki, Stefanus Jasin, Boyu Wang

    cs.AI

    Agentic AI systems often approach the same task through multiple workflows that differ in reasoning strategy, verification structure, and compute cost. A natural deployment policy is to use the workflow with the highest average performance, but this can be suboptimal because different workflows may succeed on different instances. We study a portfolio-and-selector paradigm in which a firm runs multiple workflow executions and selects the final...

    arxiv.org/abs/2609.18126 · PDF

  41. 41

    AutoTuneBench: Trustworthy Measurement for Agent Auto-Tuning of LLM Serving Engines

    Li Chen

    cs.AI

    Large language model agents tune GPU kernels and serving engines through a closed loop of propose, measure, and keep, but the measurements behind this loop are not trustworthy. We characterize four failure modes from a four-day pilot corpus of 619 model calls: strawman baselines manufacture speedups, absolute times do not transfer across machines, saturated tasks nullify comparisons, and infrastructure defects impersonate science. We present...

    arxiv.org/abs/2609.18123 · PDF

  42. 42

    When Is Graph Structure Worth Its Cost? The Case for Structure Pricing in Retrieval-Augmented Generation

    Yuzhong Zhang, Haoyang Ma, Chao Peng, Lionel Briand, Boxi Yu, Jialun Cao

    cs.AI

    Graph-based retrieval-augmented generation (RAG) can help answer questions that require information from many documents. However, building a graph often requires many language-model calls during ingestion. It is therefore important to ask whether its quality gains justify the additional cost. We present EffiRAG, a graph-based RAG system designed to reduce this cost. It uses the graph to locate relevant passages and generates answers from the...

    arxiv.org/abs/2609.18099 · PDF

  43. 43

    Decodability is Not Causality: Dissociating Probe Readouts from Behavioral Drivers via SAE Decomposition

    Devesh Tiwari, Camille Davis, Shivank Sinha, Talia Weaver, Aditya Shah, Maheep Chaudhary

    cs.AI

    Linear probes can decode safety-relevant concepts such as truthfulness from language-model activations, but probe accuracy may show only decodability, not that the features the probe weights causally drive model behavior. We demonstrate that this gap cannot be closed from the geometry of probe weights alone: the features geometrically aligned with probe direction need not be the ones the model uses, so causal relevance requires intervention....

    arxiv.org/abs/2609.18080 · PDF

  44. 44

    Teaching AI, Robotics, & Community: A Hubs-Based K-12 Education Framework for Reaching Rural Schools

    Maxwell J. Jacobson, Gustavo Rodriguez-Rivera, Petros Drineas, Yexiang Xue

    cs.AI · cs.CY

    K-12 robotics and AI education remains difficult to scale, especially in rural regions lacking sustained technical mentorship. Programs like FIRST provide competition pathways and instructional opportunities, but they do not eliminate the need for local programming and robotics expertise. We introduce AI, Robotics, & Community (ARC), a hubs-based framework where colleges train undergraduate mentors and host workshops for nearby K-12 teams....

    arxiv.org/abs/2609.18072 · PDF

  45. 45

    The Other Half of the Memory Wall: Serving 35B MoEs from SSD with Trained Routing Prediction

    Yu Lin, Yiming Wang, Runyuan Cai, Hanze Liu, Xiaodong Zeng

    cs.AI

    Mixture-of-experts (MoE) inference on consumer hardware is bounded by weight memory: a 35B-class model is 19.5GB at 4-bit, and sparsity shrinks the compute per token, not the bytes that must be held. Naive offloading to SSD does not help on its own, because layer N+1's experts must be chosen before layer N's output exists, so the reads cannot start early enough to hide behind compute. We present Edge0, a streaming MoE inference engine that...

    arxiv.org/abs/2609.18063 · PDF

  46. 46

    Anchoring What Matters: A Dual-Level Learning Framework for Visually-Grounded Multimodal Reasoning

    Xinxin Song, Siyuan Li, Tingxiong Xiao, Jinli Suo

    cs.AI

    Reinforcement learning with verifiable rewards (RLVR) has significantly improved the reasoning capabilities of large vision-language models (LVLMs). However, standard on-policy RLVR algorithms face a critical optimization bottleneck in preserving and reinforcing visually grounded reasoning behaviors: valuable visually-grounded reasoning trajectories are discarded after a single update, while uniform token advantage allocation prevents the...

    arxiv.org/abs/2609.18057 · PDF

This edition is part of The Daily Abstract — cs.AI archive. Subscribe to receive these in your inbox each morning, automatically translated to Spanish, with reply-to-PDF: arxivdaily.ignorelist.com.

Colophon Set in Georgia, with system sans for interface chrome and a monospaced stack for code and paper identifiers. Sole accent: amber #D99C5E. Built and served on an always-free VM. The masthead is set 14% letterspaced because newspapers do that and it works.