Despite the growing use of large language models (LLMs) in clinical dialogue systems, existing evaluations focus on task completion or fluency, offering little insight into the behavioral and risk management requirements essential for safety-critical systems. This paper presents MATRIX (Multi-Agent simulaTion fRamework for safe Interactions and conteXtual clinical conversational evaluation), a structured, extensible framework for safety-oriented evaluation of clinical dialogue agents. MATRIX integrates three components: (1) a safety-aligned taxonomy of clinical scenarios, expected system behaviors and failure modes derived through structured safety engineering methods; (2) BehvJudge, an LLM-based evaluator for detecting safety-relevant dialogue failures, validated against expert clinician annotations; and (3) PatBot, a simulated patient agent capable of producing diverse, scenario-conditioned responses, evaluated for realism and behavioral fidelity with human factors expertise, and a patient-preference study. Across three experiments, we show that MATRIX enables systematic, scalable safety evaluation. BehvJudge with Gemini 2.5-Pro achieves expert-level hazard detection (F1 0.96, sensitivity 0.999), outperforming clinicians in a blinded assessment of 240 dialogues. We also conducted one of the first realism analyses of LLM-based patient simulation, showing that PatBot reliably simulates realistic patient behavior in quantitative and qualitative evaluations. Using MATRIX, we demonstrate its effectiveness in benchmarking five LLM agents across 2,100 simulated dialogues spanning 14 hazard scenarios and 10 clinical domains. MATRIX is the first framework to unify structured safety engineering with scalable, validated conversational AI evaluation, enabling regulator-aligned safety auditing. We release all evaluation tools, prompts, structured scenarios, and datasets.
We present DELIVER (Directed Execution of Language-instructed Item Via Engineered Relay), a fully integrated framework for cooperative multi-robot pickup and delivery driven by natural language commands. DELIVER unifies natural language understanding, spatial decomposition, relay planning, and motion execution to enable scalable, collision-free coordination in real-world settings. Given a spoken or written instruction, a lightweight instance of LLaMA3 interprets the command to extract pickup and delivery locations. The environment is partitioned using a Voronoi tessellation to define robot-specific operating regions. Robots then compute optimal relay points along shared boundaries and coordinate handoffs. A finite-state machine governs each robot's behavior, enabling robust execution. We implement DELIVER on the MultiTRAIL simulation platform and validate it in both ROS2-based Gazebo simulations and real-world hardware using TurtleBot3 robots. Empirical results show that DELIVER maintains consistent mission cost across varying team sizes while reducing per-agent workload by up to 55% compared to a single-agent system. Moreover, the number of active relay agents remains low even as team size increases, demonstrating the system's scalability and efficient agent utilization. These findings underscore DELIVER's modular and extensible architecture for language-guided multi-robot coordination, advancing the frontiers of cyber-physical system integration.
The emergence of advanced reasoning capabilities in Large Language Models (LLMs) marks a transformative development in healthcare applications. Beyond merely expanding functional capabilities, these reasoning mechanisms enhance decision transparency and explainability-critical requirements in medical contexts. This survey examines the transformation of medical LLMs from basic information retrieval tools to sophisticated clinical reasoning systems capable of supporting complex healthcare decisions. We provide a thorough analysis of the enabling technological foundations, with a particular focus on specialized prompting techniques like Chain-of-Thought and recent breakthroughs in Reinforcement Learning exemplified by DeepSeek-R1. Our investigation evaluates purpose-built medical frameworks while also examining emerging paradigms such as multi-agent collaborative systems and innovative prompting architectures. The survey critically assesses current evaluation methodologies for medical validation and addresses persistent challenges in field interpretation limitations, bias mitigation strategies, patient safety frameworks, and integration of multimodal clinical data. Through this survey, we seek to establish a roadmap for developing reliable LLMs that can serve as effective partners in clinical practice and medical research.
Large language models (LLMs) show promise for extracting information from Electronic Health Records (EHR) and supporting clinical decisions. However, deployment in clinical settings faces challenges due to hallucination risks. We propose Hallucination Controlled Accuracy at k% (HCAcc@k%), a novel metric quantifying the accuracy-reliability trade-off at varying confidence thresholds. We introduce TrustEHRAgent, a confidence-aware agent incorporating stepwise confidence estimation for clinical question answering. Experiments on MIMIC-III and eICU datasets show TrustEHRAgent outperforms baselines under strict reliability constraints, achieving improvements of 44.23%p and 25.34%p at HCAcc@70% while baseline methods fail at these thresholds. These results highlight limitations of traditional accuracy metrics in evaluating healthcare AI agents. Our work contributes to developing trustworthy clinical agents that deliver accurate information or transparently express uncertainty when confidence is low.
Large language model (LLM)-based agents have demonstrated remarkable capabilities in decision-making tasks, but struggle significantly with complex, long-horizon planning scenarios. This arises from their lack of macroscopic guidance, causing disorientation and failures in complex tasks, as well as insufficient continuous oversight during execution, rendering them unresponsive to environmental changes and prone to deviations. To tackle these challenges, we introduce HiPlan, a hierarchical planning framework that provides adaptive global-local guidance to boost LLM-based agents'decision-making. HiPlan decomposes complex tasks into milestone action guides for general direction and step-wise hints for detailed actions. During the offline phase, we construct a milestone library from expert demonstrations, enabling structured experience reuse by retrieving semantically similar tasks and milestones. In the execution phase, trajectory segments from past milestones are dynamically adapted to generate step-wise hints that align current observations with the milestone objectives, bridging gaps and correcting deviations. Extensive experiments across two challenging benchmarks demonstrate that HiPlan substantially outperforms strong baselines, and ablation studies validate the complementary benefits of its hierarchical components.
We introduce the Concurrent Modular Agent (CMA), a framework that orchestrates multiple Large-Language-Model (LLM)-based modules that operate fully asynchronously yet maintain a coherent and fault-tolerant behavioral loop. This framework addresses long-standing difficulties in agent architectures by letting intention emerge from language-mediated interactions among autonomous processes. This approach enables flexible, adaptive, and context-dependent behavior through the combination of concurrently executed modules that offload reasoning to an LLM, inter-module communication, and a single shared global state.We consider this approach to be a practical realization of Minsky's Society of Mind theory. We demonstrate the viability of our system through two practical use-case studies. The emergent properties observed in our system suggest that complex cognitive phenomena like self-awareness may indeed arise from the organized interaction of simpler processes, supporting Minsky-Society of Mind concept and opening new avenues for artificial intelligence research. The source code for our work is available at: https://github.com/AlternativeMachine/concurrent-modular-agent.
This paper introduces MovieCORE, a novel video question answering (VQA) dataset designed to probe deeper cognitive understanding of movie content. Unlike existing datasets that focus on surface-level comprehension, MovieCORE emphasizes questions that engage System-2 thinking while remaining specific to the video material. We present an innovative agentic brainstorming approach, utilizing multiple large language models (LLMs) as thought agents to generate and refine high-quality question-answer pairs. To evaluate dataset quality, we develop a set of cognitive tests assessing depth, thought-provocation potential, and syntactic complexity. We also propose a comprehensive evaluation scheme for assessing VQA model performance on deeper cognitive tasks. To address the limitations of existing video-language models (VLMs), we introduce an agentic enhancement module, Agentic Choice Enhancement (ACE), which improves model reasoning capabilities post-training by up to 25%. Our work contributes to advancing movie understanding in AI systems and provides valuable insights into the capabilities and limitations of current VQA models when faced with more challenging, nuanced questions about cinematic content. Our project page, dataset and code can be found at https://joslefaure.github.io/assets/html/moviecore.html.
As AI advances toward general intelligence, the focus is shifting from systems optimized for static tasks to creating open-ended agents that learn continuously. In this paper, we introduce Experience-driven Lifelong Learning (ELL), a framework for building self-evolving agents capable of continuous growth through real-world interaction. The framework is built on four core principles: (1) Experience Exploration: Agents learn through continuous, self-motivated interaction with dynamic environments, navigating interdependent tasks and generating rich experiential trajectories. (2) Long-term Memory: Agents preserve and structure historical knowledge, including personal experiences, domain expertise, and commonsense reasoning, into a persistent memory system. (3) Skill Learning: Agents autonomously improve by abstracting recurring patterns from experience into reusable skills, which are actively refined and validated for application in new tasks. (4) Knowledge Internalization: Agents internalize explicit and discrete experiences into implicit and intuitive capabilities as "second nature". We also introduce StuLife, a benchmark dataset for ELL that simulates a student's holistic college journey, from enrollment to academic and personal development, across three core phases and ten detailed sub-scenarios. StuLife is designed around three key paradigm shifts: From Passive to Proactive, From Context to Memory, and From Imitation to Learning. In this dynamic environment, agents must acquire and distill practical skills and maintain persistent memory to make decisions based on evolving state variables. StuLife provides a comprehensive platform for evaluating lifelong learning capabilities, including memory retention, skill transfer, and self-motivated behavior. Beyond evaluating SOTA LLMs on the StuLife benchmark, we also explore the role of context engineering in advancing AGI.
Beyond scratch coding, exploiting large-scale code repositories (e.g., GitHub) for practical tasks is vital in real-world software development, yet current benchmarks rarely evaluate code agents in such authentic, workflow-driven scenarios. To bridge this gap, we introduce GitTaskBench, a benchmark designed to systematically assess this capability via 54 realistic tasks across 7 modalities and 7 domains. Each task pairs a relevant repository with an automated, human-curated evaluation harness specifying practical success criteria. Beyond measuring execution and task success, we also propose the alpha-value metric to quantify the economic benefit of agent performance, which integrates task success rates, token cost, and average developer salaries. Experiments across three state-of-the-art agent frameworks with multiple advanced LLMs show that leveraging code repositories for complex task solving remains challenging: even the best-performing system, OpenHands+Claude 3.7, solves only 48.15% of tasks. Error analysis attributes over half of failures to seemingly mundane yet critical steps like environment setup and dependency resolution, highlighting the need for more robust workflow management and increased timeout preparedness. By releasing GitTaskBench, we aim to drive progress and attention toward repository-aware code reasoning, execution, and deployment -- moving agents closer to solving complex, end-to-end real-world tasks. The benchmark and code are open-sourced at https://github.com/QuantaAlpha/GitTaskBench.
Standard single-turn, static benchmarks fall short in evaluating the nuanced capabilities of Large Language Models (LLMs) on complex tasks such as software engineering. In this work, we propose a novel interactive evaluation framework that assesses LLMs on multi-requirement programming tasks through structured, feedback-driven dialogue. Each task is modeled as a requirement dependency graph, and an ``interviewer'' LLM, aware of the ground-truth solution, provides minimal, targeted hints to an ``interviewee'' model to help correct errors and fulfill target constraints. This dynamic protocol enables fine-grained diagnostic insights into model behavior, uncovering strengths and systematic weaknesses that static benchmarks fail to measure. We build on DevAI, a benchmark of 55 curated programming tasks, by adding ground-truth solutions and evaluating the relevance and utility of interviewer hints through expert annotation. Our results highlight the importance of dynamic evaluation in advancing the development of collaborative code-generating agents.
Large Language Models (LLMs) are being integrated into professional domains, yet their limitations in high-stakes fields like law remain poorly understood. This paper defines the core capabilities that an AI system must possess to function as a reliable reasoning tool in judicial decision-making. Using the IRAC (Issue-Rule-Application-Conclusion) model as an analytical framework, the study focuses on the most challenging phases of legal adjudication: determining the applicable Rule (R) and performing the Application (A) of that rule to the facts of a case. From a judicial perspective, the analysis deconstructs legal reasoning into a series of core requirements, including the ability to select the correct legal framework across jurisdictions, generate sound arguments based on the doctrine of legal sources, distinguish ratio decidendi from obiter dictum in case law, resolve ambiguity arising from general clauses like "reasonableness", manage conflicting legal provisions, and correctly apply the burden of proof. The paper then maps various AI enhancement mechanisms, such as Retrieval-Augmented Generation (RAG), multi-agent systems, and neuro-symbolic AI, to these requirements, assessing their potential to bridge the gap between the probabilistic nature of LLMs and the rigorous, choice-driven demands of legal interpretation. The findings indicate that while these techniques can address specific challenges, significant challenges remain, particularly in tasks requiring discretion and transparent, justifiable reasoning. Our paper concludes that the most effective current role for AI in law is a dual one: as a high-volume assistant for simple, repetitive cases and as a sophisticated "sparring partner" for human experts in complex matters.
While modern recommender systems are instrumental in navigating information abundance, they remain fundamentally limited by static user modeling and reactive decision-making paradigms. Current large language model (LLM)-based agents inherit these shortcomings through their overreliance on heuristic pattern matching, yielding recommendations prone to shallow correlation bias, limited causal inference, and brittleness in sparse-data scenarios. We introduce STARec, a slow-thinking augmented agent framework that endows recommender systems with autonomous deliberative reasoning capabilities. Each user is modeled as an agent with parallel cognitions: fast response for immediate interactions and slow reasoning that performs chain-of-thought rationales. To cultivate intrinsic slow thinking, we develop anchored reinforcement training - a two-stage paradigm combining structured knowledge distillation from advanced reasoning models with preference-aligned reward shaping. This hybrid approach scaffolds agents in acquiring foundational capabilities (preference summarization, rationale generation) while enabling dynamic policy adaptation through simulated feedback loops. Experiments on MovieLens 1M and Amazon CDs benchmarks demonstrate that STARec achieves substantial performance gains compared with state-of-the-art baselines, despite using only 0.4% of the full training data.
The proliferation of Internet of things (IoT) devices in smart cities, transportation, healthcare, and industrial applications, coupled with the explosive growth of AI-driven services, has increased demands for efficient distributed computing architectures and networks, driving cloud-edge-terminal collaborative intelligence (CETCI) as a fundamental paradigm within the artificial intelligence of things (AIoT) community. With advancements in deep learning, large language models (LLMs), and edge computing, CETCI has made significant progress with emerging AIoT applications, moving beyond isolated layer optimization to deployable collaborative intelligence systems for AIoT (CISAIOT), a practical research focus in AI, distributed computing, and communications. This survey describes foundational architectures, enabling technologies, and scenarios of CETCI paradigms, offering a tutorial-style review for CISAIOT beginners. We systematically analyze architectural components spanning cloud, edge, and terminal layers, examining core technologies including network virtualization, container orchestration, and software-defined networking, while presenting categorizations of collaboration paradigms that cover task offloading, resource allocation, and optimization across heterogeneous infrastructures. Furthermore, we explain intelligent collaboration learning frameworks by reviewing advances in federated learning, distributed deep learning, edge-cloud model evolution, and reinforcement learning-based methods. Finally, we discuss challenges (e.g., scalability, heterogeneity, interoperability) and future trends (e.g., 6G+, agents, quantum computing, digital twin), highlighting how integration of distributed computing and communication can address open issues and guide development of robust, efficient, and secure collaborative AIoT systems.
Minecraft, as an open-world virtual interactive environment, has become a prominent platform for research on agent decision-making and execution. Existing works primarily adopt a single Large Language Model (LLM) agent to complete various in-game tasks. However, for complex tasks requiring lengthy sequences of actions, single-agent approaches often face challenges related to inefficiency and limited fault tolerance. Despite these issues, research on multi-agent collaboration remains scarce. In this paper, we propose CausalMACE, a holistic causality planning framework designed to enhance multi-agent systems, in which we incorporate causality to manage dependencies among subtasks. Technically, our proposed framework introduces two modules: an overarching task graph for global task planning and a causality-based module for dependency management, where inherent rules are adopted to perform causal intervention. Experimental results demonstrate our approach achieves state-of-the-art performance in multi-agent cooperative tasks of Minecraft.
Large Language Models (LLMs) have transformed the field of artificial intelligence by unlocking the era of generative applications. Built on top of generative AI capabilities, Agentic AI represents a major shift toward autonomous, goal-driven systems that can reason, retrieve, and act. However, they also inherit the bias present in both internal and external information sources. This significantly affects the fairness and balance of retrieved information, and hence reduces user trust. To address this critical challenge, we introduce a novel Bias Mitigation Agent, a multi-agent system designed to orchestrate the workflow of bias mitigation through specialized agents that optimize the selection of sources to ensure that the retrieved content is both highly relevant and minimally biased to promote fair and balanced knowledge dissemination. The experimental results demonstrate an 81.82\% reduction in bias compared to a baseline naive retrieval strategy.
Large language models (LLMs) have shown significant promise in embodied decision-making tasks within virtual open-world environments. Nonetheless, their performance is hindered by the absence of domain-specific knowledge. Methods that finetune on large-scale domain-specific data entail prohibitive development costs. This paper introduces VistaWise, a cost-effective agent framework that integrates cross-modal domain knowledge and finetunes a dedicated object detection model for visual analysis. It reduces the requirement for domain-specific training data from millions of samples to a few hundred. VistaWise integrates visual information and textual dependencies into a cross-modal knowledge graph (KG), enabling a comprehensive and accurate understanding of multimodal environments. We also equip the agent with a retrieval-based pooling strategy to extract task-related information from the KG, and a desktop-level skill library to support direct operation of the Minecraft desktop client via mouse and keyboard inputs. Experimental results demonstrate that VistaWise achieves state-of-the-art performance across various open-world tasks, highlighting its effectiveness in reducing development costs while enhancing agent performance.
Large language model (LLM)-based agents have demonstrated remarkable capabilities in addressing complex tasks, thereby enabling more advanced information retrieval and supporting deeper, more sophisticated human information-seeking behaviors. However, most existing agents operate in a purely reactive manner, responding passively to user instructions, which significantly constrains their effectiveness and efficiency as general-purpose platforms for information acquisition. To overcome this limitation, this paper proposes AppAgent-Pro, a proactive GUI agent system that actively integrates multi-domain information based on user instructions. This approach enables the system to proactively anticipate users' underlying needs and conduct in-depth multi-domain information mining, thereby facilitating the acquisition of more comprehensive and intelligent information. AppAgent-Pro has the potential to fundamentally redefine information acquisition in daily life, leading to a profound impact on human society. Our code is available at: https://github.com/LaoKuiZe/AppAgent-Pro. Our code is available at: https://github.com/LaoKuiZe/AppAgent-Pro. The demonstration video could be found at: https://www.dropbox.com/scl/fi/hvzqo5vnusg66srydzixo/AppAgent-Pro-demo-video.mp4?rlkey=o2nlfqgq6ihl125mcqg7bpgqu&st=d29vrzii&dl=0.
Signature-based Intrusion Detection Systems (IDS) detect malicious activities by matching network or host activity against predefined rules. These rules are derived from extensive Cyber Threat Intelligence (CTI), which includes attack signatures and behavioral patterns obtained through automated tools and manual threat analysis, such as sandboxing. The CTI is then transformed into actionable rules for the IDS engine, enabling real-time detection and prevention. However, the constant evolution of cyber threats necessitates frequent rule updates, which delay deployment time and weaken overall security readiness. Recent advancements in agentic systems powered by Large Language Models (LLMs) offer the potential for autonomous IDS rule generation with internal evaluation. We introduce FALCON, an autonomous agentic framework that generates deployable IDS rules from CTI data in real-time and evaluates them using built-in multi-phased validators. To demonstrate versatility, we target both network (Snort) and host-based (YARA) mediums and construct a comprehensive dataset of IDS rules with their corresponding CTIs. Our evaluations indicate FALCON excels in automatic rule generation, with an average of 95% accuracy validated by qualitative evaluation with 84% inter-rater agreement among multiple cybersecurity analysts across all metrics. These results underscore the feasibility and effectiveness of LLM-driven data mining for real-time cyber threat mitigation.
Automated tabular understanding and reasoning are essential tasks for data scientists. Recently, Large language models (LLMs) have become increasingly prevalent in tabular reasoning tasks. Previous work focuses on (1) finetuning LLMs using labeled data or (2) Training-free prompting LLM agents using chain-of-thought (CoT). Finetuning offers dataset-specific learning at the cost of generalizability. Training-free prompting is highly generalizable but does not take full advantage of training data. In this paper, we propose a novel prompting-based reasoning approach, Learn then Retrieve: LRTab, which integrates the benefits of both by retrieving relevant information learned from training data. We first use prompting to obtain CoT responses over the training data. For incorrect CoTs, we prompt the LLM to predict Prompt Conditions to avoid the error, learning insights from the data. We validate the effectiveness of Prompt Conditions using validation data. Finally, at inference time, we retrieve the most relevant Prompt Conditions for additional context for table understanding. We provide comprehensive experiments on WikiTQ and Tabfact, showing that LRTab is interpretable, cost-efficient, and can outperform previous baselines in tabular reasoning.
Automated code generation has long been considered the holy grail of software engineering. The emergence of Large Language Models (LLMs) has catalyzed a revolutionary breakthrough in this area. However, existing methods that only rely on LLMs remain inadequate in the quality of generated code, offering no guarantees of satisfying practical requirements. They lack a systematic strategy for requirements development and modeling. Recently, LLM-based agents typically possess powerful abilities and play an essential role in facilitating the alignment of LLM outputs with user requirements. In this paper, we envision the first multi-agent framework for reliable code generation based on \textsc{re}quirements \textsc{de}velopment and \textsc{fo}rmalization, named \textsc{ReDeFo}. This framework incorporates three agents, highlighting their augmentation with knowledge and techniques of formal methods, into the requirements-to-code generation pipeline to strengthen quality assurance. The core of \textsc{ReDeFo} is the use of formal specifications to bridge the gap between potentially ambiguous natural language requirements and precise executable code. \textsc{ReDeFo} enables rigorous reasoning about correctness, uncovering hidden bugs, and enforcing critical properties throughout the development process. In general, our framework aims to take a promising step toward realizing the long-standing vision of reliable, auto-generated software.
With the recent rapid advancement of Agentic Intelligence, agentic tool use in LLMs has become increasingly important. During multi-turn interactions between agents and users, the dynamic, uncertain, and stochastic nature of user demands poses significant challenges to the agent's tool invocation capabilities. Agents are no longer expected to simply call tools to deliver a result; rather, they must iteratively refine their understanding of user needs through communication while simultaneously invoking tools to resolve user queries. Existing reinforcement learning (RL) approaches for tool use lack the integration of genuinely dynamic users during the RL training process. To bridge this gap, we introduce MUA-RL (Multi-turn User-interacting Agent Reinforcement Learning for agentic tool use), a novel reinforcement learning framework that, for the first time in the field of agentic tool use, integrates LLM-simulated users into the reinforcement learning loop. MUA-RL aims to enable autonomous learning of models to communicate with users efficiently and use various tools to solve practical problems in dynamic multi-turn interactions. Evaluations are done on several multi-turn tool-using benchmarks (see Figure 1). Specifically, MUA-RL-32B achieves 67.3 on TAU2 Retail, 45.4 on TAU2 Airline, 28.3 on TAU2 Telecom, 28.4 on BFCL-V3 Multi Turn, and 82.5 on ACEBench Agent -- outperforming or matching the performance of larger open-source models such as DeepSeek-V3-0324 and Qwen3-235B-A22B in non-thinking settings.
Peer-to-peer (P2P) energy trading is becoming central to modern distribution systems as rooftop PV and home energy management systems become pervasive, yet most existing market and reinforcement learning designs emphasize efficiency or private profit and offer little real-time guidance to ensure equitable outcomes under uncertainty. To address this gap, a fairness-aware multiagent reinforcement learning framework, FairMarket-RL, is proposed in which a large language model (LLM) critic shapes bidding policies within a continuous double auction under partial observability and discrete price-quantity actions. After each trading slot, the LLM returns normalized fairness scores Fairness-to-Grid (FTG), Fairness-Between-Sellers (FBS), and Fairness-of-Pricing (FPP) that are integrated into the reward via ramped coefficients and tunable scaling, so that fairness guidance complements, rather than overwhelms, economic incentives. The environment models realistic residential load and PV profiles and enforce hard constraints on prices, physical feasibility, and policy-update stability. Across a progression of experiments from a small pilot to a larger simulated community and a mixed-asset real-world dataset, the framework shifts exchanges toward local P2P trades, lowers consumer costs relative to grid-only procurement, sustains strong fairness across participants, and preserves utility viability. Sensitivity analyses over solar availability and aggregate demand further indicate robust performance, suggesting a scalable, LLM-guided pathway to decentralized electricity markets that are economically efficient, socially equitable, and technically sound.
Large language models (LLMs) are increasingly used to simulate human decision-making, but their intrinsic biases often diverge from real human behavior--limiting their ability to reflect population-level diversity. We address this challenge with a persona-based approach that leverages individual-level behavioral data from behavioral economics to adjust model biases. Applying this method to the ultimatum game--a standard but difficult benchmark for LLMs--we observe improved alignment between simulated and empirical behavior, particularly on the responder side. While further refinement of trait representations is needed, our results demonstrate the promise of persona-conditioned LLMs for simulating human-like decision patterns at scale.
Large language model (LLM)-powered agents are increasingly used to plan and execute scientific workflows, yet most research cyberinfrastructure (CI) exposes heterogeneous APIs and implements security models that present barriers for use by agents. We report on our experience using the Model Context Protocol (MCP) as a unifying interface that makes research capabilities discoverable, invokable, and composable. Our approach is pragmatic: we implement thin MCP servers over mature services, including Globus Transfer, Compute, and Search; status APIs exposed by computing facilities; Octopus event fabric; and domain-specific tools such as Garden and Galaxy. We use case studies in computational chemistry, bioinformatics, quantum chemistry, and filesystem monitoring to illustrate how this MCP-oriented architecture can be used in practice. We distill lessons learned and outline open challenges in evaluation and trust for agent-led science.
As AI agents become increasingly capable of tool use and long-horizon tasks, they have begun to be deployed in settings where multiple agents can interact. However, whereas prior work has mostly focused on human-AI interactions, there is an increasing need to understand AI-AI interactions. In this paper, we adapt the iterated public goods game, a classic behavioral economics game, to analyze the behavior of four reasoning and non-reasoning models across two conditions: models are either told they are playing against "another AI agent" or told their opponents are themselves. We find that, across different settings, telling LLMs that they are playing against themselves significantly changes their tendency to cooperate. While our study is conducted in a toy environment, our results may provide insights into multi-agent settings where agents "unconsciously" discriminating against each other could inexplicably increase or decrease cooperation.
This paper explores the combination of two intrinsic motivation strategies to improve the efficiency of reinforcement learning (RL) agents in environments with extreme sparse rewards, where traditional learning struggles due to infrequent positive feedback. We propose integrating Variational State as Intrinsic Reward (VSIMR), which uses Variational AutoEncoders (VAEs) to reward state novelty, with an intrinsic reward approach derived from Large Language Models (LLMs). The LLMs leverage their pre-trained knowledge to generate reward signals based on environment and goal descriptions, guiding the agent. We implemented this combined approach with an Actor-Critic (A2C) agent in the MiniGrid DoorKey environment, a benchmark for sparse rewards. Our empirical results show that this combined strategy significantly increases agent performance and sampling efficiency compared to using each strategy individually or a standard A2C agent, which failed to learn. Analysis of learning curves indicates that the combination effectively complements different aspects of the environment and task: VSIMR drives exploration of new states, while the LLM-derived rewards facilitate progressive exploitation towards goals.
One of the enduring challenges in education is how to empower students to take ownership of their learning by setting meaningful goals, tracking their progress, and adapting their strategies when faced with setbacks. Research has shown that this form of leaner-centered learning is best cultivated through structured, supportive environments that promote guided practice, scaffolded inquiry, and collaborative dialogue. In response, educational efforts have increasingly embraced artificial-intelligence (AI)-powered digital learning environments, ranging from educational apps and virtual labs to serious games. Recent advances in large language models (LLMs) and neuro-symbolic systems, meanwhile, offer a transformative opportunity to reimagine how support is delivered in digital learning environments. LLMs are enabling socially interactive learning experiences and scalable, cross-domain learning support that can adapt instructional strategies across varied subjects and contexts. In parallel, neuro-symbolic AI provides new avenues for designing these agents that are not only adaptive but also scalable across domains. Based on these remarks, this paper presents a multi-agent, neuro-symbolic framework designed to resolve the aforementioned challenges. The framework assigns distinct pedagogical roles to specialized agents: an RL-based 'tutor' agent provides authoritative, non-verbal scaffolding, while a proactive, LLM-powered 'peer' agent facilitates the social dimensions of learning. While prior work has explored such agents in isolation, our framework's novelty lies in unifying them through a central educational ontology. Through case studies in both college-level and middle school settings, we demonstrate the framework's adaptability across domains. We conclude by outlining key insights and future directions for advancing AI-driven learning environments.
Large language models (LLMs) have demonstrated exceptional capabilities when trained within executable runtime environments, notably excelling at software engineering tasks through verified feedback loops. Yet, scalable and generalizable execution-grounded environments remain scarce, limiting progress in training more capable ML agents. We introduce CTF-Dojo, the first large-scale executable runtime tailored for training LLMs with verifiable feedback, featuring 658 fully functional Capture-The-Flag (CTF)-style challenges containerized in Docker with guaranteed reproducibility. To enable rapid scaling without manual intervention, we develop CTF-Forge, an automated pipeline that transforms publicly available artifacts into ready-to-use execution environments in minutes, eliminating weeks of expert configuration traditionally required. We trained LLM-based agents on just 486 high-quality, execution-verified trajectories from CTF-Dojo, achieving up to 11.6% absolute gains over strong baselines across three competitive benchmarks: InterCode-CTF, NYU CTF Bench, and Cybench. Our best-performing 32B model reaches 31.9% Pass@1, establishing a new open-weight state-of-the-art that rivals frontier models like DeepSeek-V3-0324 and Gemini-2.5-Flash. By framing CTF-style tasks as a benchmark for executable-agent learning, CTF-Dojo demonstrates that execution-grounded training signals are not only effective but pivotal in advancing high-performance ML agents without dependence on costly proprietary systems.
Large Language Models (LLMs) have demonstrated remarkable capabilities in understanding and generating human-like text, yet they largely operate as reactive agents, responding only when directly prompted. This passivity creates an "awareness gap," limiting their potential as truly collaborative partners in dynamic human discussions. We introduce $\textit{DiscussLLM}$, a framework designed to bridge this gap by training models to proactively decide not just $\textit{what}$ to say, but critically, $\textit{when}$ to speak. Our primary contribution is a scalable two-stage data generation pipeline that synthesizes a large-scale dataset of realistic multi-turn human discussions. Each discussion is annotated with one of five intervention types (e.g., Factual Correction, Concept Definition) and contains an explicit conversational trigger where an AI intervention adds value. By training models to predict a special silent token when no intervention is needed, they learn to remain quiet until a helpful contribution can be made. We explore two architectural baselines: an integrated end-to-end model and a decoupled classifier-generator system optimized for low-latency inference. We evaluate these models on their ability to accurately time interventions and generate helpful responses, paving the way for more situationally aware and proactive conversational AI.
Imagine decision-makers uploading data and, within minutes, receiving clear, actionable insights delivered straight to their fingertips. That is the promise of the AI Data Scientist, an autonomous Agent powered by large language models (LLMs) that closes the gap between evidence and action. Rather than simply writing code or responding to prompts, it reasons through questions, tests ideas, and delivers end-to-end insights at a pace far beyond traditional workflows. Guided by the scientific tenet of the hypothesis, this Agent uncovers explanatory patterns in data, evaluates their statistical significance, and uses them to inform predictive modeling. It then translates these results into recommendations that are both rigorous and accessible. At the core of the AI Data Scientist is a team of specialized LLM Subagents, each responsible for a distinct task such as data cleaning, statistical testing, validation, and plain-language communication. These Subagents write their own code, reason about causality, and identify when additional data is needed to support sound conclusions. Together, they achieve in minutes what might otherwise take days or weeks, enabling a new kind of interaction that makes deep data science both accessible and actionable.