Artificial intelligence has experienced a significant boom with the emergence of agentic AI,where autonomous agents are increasingly replacing human intervention,enabling systems to perceive,reason,and act independent...Artificial intelligence has experienced a significant boom with the emergence of agentic AI,where autonomous agents are increasingly replacing human intervention,enabling systems to perceive,reason,and act independently to achieve specific goals.Despite its transformative potential,comprehensive information on agentic AI remains scarce in the literature.This paper provides the first comprehensive review of agentic AI,focusing on its evolution and three core aspects:patterns,types,and environments.The evolution of agentic AI is traced through five phases to the current era of multi-modal and collaborative agents,driven by advancements in reinforcement learning,neural networks,and large language models(LLMs).Five key patterns:tool use,reflection,ReAct,planning,and multi-agent collaboration(MAC)define how agentic AI systems interact and process tasks.These systems are categorized into seven categories,each tailored for specific operational styles and autonomy in decision making.The environments in which these agents operate are classified as static,dynamic,fully observable,partially observable,deterministic,stochastic,single-agent,and multiagent,emphasizing the impact of environmental complexity on agent behavior.Agentic AI has revolutionized systems through autonomous decision making and resource optimization,yet challenges persist in aligning AI with human values,ensuring adaptability,and addressing ethical constraints.Future research focuses on multidomain agents,human–AI collaboration,and self-improving systems.This work provides researchers,practitioners,and policymakers with a structured approach to understanding and advancing the rapidly evolving landscape of agentic AI systems.展开更多
Large LanguageModels(LLMs)are increasingly utilized for semantic understanding and reasoning,yet their use in sensitive settings is limited by privacy concerns.This paper presents In-Mig,a mobile-agent architecture th...Large LanguageModels(LLMs)are increasingly utilized for semantic understanding and reasoning,yet their use in sensitive settings is limited by privacy concerns.This paper presents In-Mig,a mobile-agent architecture that integrates LLM reasoning within agents that can migrate across organizational venues.Unlike centralized approaches,In-Mig performs reasoning in situ,ensuring that raw data remains within institutional boundaries while allowing for cross-venue synthesis.The architecture features a policy-scoped memory model,utility-driven route planning,and cryptographic trust enforcement.Aprototype using JADE for mobility and quantizedMistral-7B demonstrates practical feasibility.Evaluation across various scenarios shows that In-Mig achieves 92%similarity to centralized baselines,confirming its utility and strong privacy guarantees.These results suggest that migrating,privacy-preserving LLM agents can effectively support decentralized reasoning in trust-sensitive domains.展开更多
While the complexity of fifth-generation wireless networks is being widely commented upon,there is great anticipation for the arrival of the sixth generation(6G),with its enriched capabilities and features.It can easi...While the complexity of fifth-generation wireless networks is being widely commented upon,there is great anticipation for the arrival of the sixth generation(6G),with its enriched capabilities and features.It can easily be imagined that,without proper design,the enrichment of 6G will further increase system complexity.To address this issue,we propose the Agentic-AI Core(A-Core),an artificial intelligence(AI)-empowered,mission-oriented core network architecture for next-generation mobile telecommunications.In A-Core,network capabilities can be added and updated on the fly and further programmed into missions for enabling and offering diverse services to customers.These missions are created and executed by autonomous network agents according to the customer's intent,which may be expressed in natural language.The agents resolve intents from customers into workflows of network capabilities by leveraging a large-scale network AI model and follow the workflows to execute the mission.As an open,agile system architecture,A-Core holds promise for accelerating innovation and greatly reducing standard release times.The advantages of A-Core are demonstrated through two use cases.展开更多
Artificial Intelligence(AI)is transforming the healthcare landscape,yet many current applications remain narrowly task-specific,constrained by data complexity and inherent biases.This paper explores the emergence of n...Artificial Intelligence(AI)is transforming the healthcare landscape,yet many current applications remain narrowly task-specific,constrained by data complexity and inherent biases.This paper explores the emergence of next generation"agentic AI"systems,characterized by advanced autonomy,adaptability,scalability,and prob-abilistic reasoning,which address critical challenges in medical management.These systems enhance various aspects of healthcare,including diagnostics,clinical decision support,treatment planning,patient monitoring,administrative operations,drug discovery,and robotic-assisted surgery.Powered by multimodal AI,agentic systems integrate diverse data sources,iteratively refine outputs,and leverage vast knowledge bases to deliver context-aware,patient-centric care with heightened precision and reduced error rates.These advancements promise to enhance patient outcomes,optimize clinical workflows,and expand the reach of AI-driven solutions.However,their deployment introduces ethical,privacy,and regulatory challenges,emphasizing the need for robust governance frameworks and interdisciplinary collaboration.Agentic AI has the potential to redefine healthcare,driving personalized,efficient,and scalable services while extending its impact beyond clinical settings to global public health initiatives.By addressing disparities and enhancing care delivery in resourcelimited environments,this technology could significantly advance equitable healthcare.Realizing the full po-tential of agentic AI will require sustained research,innovation,and cross-disciplinary partnerships to ensure its responsible and transformative integration into healthcare systems worldwide.展开更多
Governance scholarship overwhelmingly treats artificial intelligence as a tool—an object to be regulated,audited,or aligned with human values.This article argues that when organizations treat the outputs of AI agents...Governance scholarship overwhelmingly treats artificial intelligence as a tool—an object to be regulated,audited,or aligned with human values.This article argues that when organizations treat the outputs of AI agents as institutionally binding,these systems cross a threshold from instruments into institutional actors.Drawing on Goffman’s interaction order and Meyer and Rowan’s institutional isomorphism,we develop a diagnostic framework that specifies when and how AI agents acquire practical actorhood within organizations.We formalize a 2×2 Actorhood Matrix along two axes—discretion granted and institutional embedding—yielding four system types:Tool,Infrastructure,Shadow Actor,and Institutional Actor.The Actorhood Matrix is proposed as a reusable diagnostic method for identifying when AI systems cross the institutional threshold from tools to role occupants.Applying this framework to five empirical cases(Klarna’s AI customer service,GitHub Copilot Workspace,NHS AI triage,Harvey AI legal assistant,and generic FAQ chatbots)demonstrates that institutional actorhood is not a property of technical sophistication but of organizational role assignment.We identify a critical transition zone where override rates fall below ten percent and propose four testable propositions linking discretion,embedding,and temporal persistence to governance outcomes.The article concludes that governance frameworks designed for“tools”are structurally inadequate for systems that have become practical role occupants within institutional settings.展开更多
Agentic AI represents a significant advancement in artificial intelligence,enabling proactive agents that can set goals,make decisions,and adapt to changing situations.However,the performance of these systems is heavi...Agentic AI represents a significant advancement in artificial intelligence,enabling proactive agents that can set goals,make decisions,and adapt to changing situations.However,the performance of these systems is heavily dependent on the quality and relevance of the data they process.This research highlights the critical risk posed by faulty,insecure,or contextually inappropriate input data in modern Agentic AI systems.To address this challenge,this study proposes the Autonomous Data Integrity Layer(ADIL).This flexible architecture integrates best practices from security engineering and data science to ensure that Agentic AI systems operate with clean,validated,and contextually relevant data.By focusing on data integrity,ADIL enhances the reliability,accountability,and effectiveness of Agentic AI systems,leading to more trustworthy and robust intelligent agents.展开更多
Advancements in large language models(LLMs)have markedly improved the adaptability of artificial intelligence(AI)agents in dynamic and open environments.However,with the growing number and diversity of agents,ensuring...Advancements in large language models(LLMs)have markedly improved the adaptability of artificial intelligence(AI)agents in dynamic and open environments.However,with the growing number and diversity of agents,ensuring secure,reliable,and autonomous collaboration among them has become an urgent and critical challenge.To this end,this letter proposes agent reinforced generation(ARG)to establish a multi-agent system with audit trail functionality,privacy compliance,and autonomous coordination.ARG integrates the model context protocol(MCP)and agent-to-agent(A2A)protocol to define the rules and logic governing agent-to-agent communications as well as agent-to-tool/data engagements.Decentralized autonomous organizations and operations(DAOs)are employed to enable agents to coordinate and execute tasks in a transparent and tamper-resistant manner.Additionally,the operational process of ARG is elaborated from task issuance to completion to validate the auditability and immutability of task coordination and execution.Finally,we highlight five key features of ARG,including parallelism and throughput,scalability across domains and load,fault tolerance and graceful failure,resource efficiency through delegation,as well as data security and privacy protection,positioning it as a promising paradigm for the realization of agentic intelligence.展开更多
Artificial intelligence(AI)is reshaping financial systems and services,as intelligent AI agents increasingly form the foundation of autonomous,goal-driven systems capable of reasoning,learning,and action.This review s...Artificial intelligence(AI)is reshaping financial systems and services,as intelligent AI agents increasingly form the foundation of autonomous,goal-driven systems capable of reasoning,learning,and action.This review synthesizes recent research and developments in the application of AI agents across core financial domains.Specifically,it covers the deployment of agent-based AI in algorithmic trading,fraud detection,credit risk assessment,roboadvisory,and regulatory compliance(RegTech).The review focuses on advanced agent-based methodologies,including reinforcement learning,multi-agent systems,and autonomous decision-making frameworks,particularly those leveraging large language models(LLMs),contrasting these with traditional AI or purely statistical models.Our primary goals are to consolidate current knowledge,identify significant trends and architectural approaches,review the practical efficiency and impact of current applications,and delineate key challenges and promising future research directions.The increasing sophistication of AI agents offers unprecedented opportunities for innovation in finance,yet presents complex technical,ethical,and regulatory challenges that demand careful consideration and proactive strategies.This review aims to provide a comprehensive understanding of this rapidly evolving landscape,highlighting the role of agent-based AI in the ongoing transformation of the financial industry,and is intended to serve financial institutions,regulators,investors,analysts,researchers,and other key stakeholders in the financial ecosystem.展开更多
In recent years,researchers have leveraged single-agent reinforcement learning to boost educational outcomes and deliver personalized interventions;yet this paradigm provides no capacity for inter-agent interaction.Mu...In recent years,researchers have leveraged single-agent reinforcement learning to boost educational outcomes and deliver personalized interventions;yet this paradigm provides no capacity for inter-agent interaction.Multi-agent reinforcement learning(MARL)overcomes this limitation by allowing several agents to learn simultaneously within a shared environment,each choosing actions that maximize its own or the group's rewards.By explicitly modeling and exploiting agent-to-agent dynamics,MARL can align those interactions with pedagogical goals such as peer tutoring,collaborative problem-solving,or gamified competition,thus opening richer avenues for adaptive and socially informed learning experiences.This survey investigates the impact of MARL on educational outcomes by examining evidence of its effectiveness in enhancing learner performance,engagement,equity,and reducing teacher workload compared to single agent or traditional approaches.It explores the educational domains and pedagogical problems addressed by MARL,identifies the algorithmic families used,and analyzes their influence on learning.The review also assesses experimental settings and evaluation metrics to determine ecological validity,and outlines current challenges and future research directions in applying MARL to education.展开更多
文摘Artificial intelligence has experienced a significant boom with the emergence of agentic AI,where autonomous agents are increasingly replacing human intervention,enabling systems to perceive,reason,and act independently to achieve specific goals.Despite its transformative potential,comprehensive information on agentic AI remains scarce in the literature.This paper provides the first comprehensive review of agentic AI,focusing on its evolution and three core aspects:patterns,types,and environments.The evolution of agentic AI is traced through five phases to the current era of multi-modal and collaborative agents,driven by advancements in reinforcement learning,neural networks,and large language models(LLMs).Five key patterns:tool use,reflection,ReAct,planning,and multi-agent collaboration(MAC)define how agentic AI systems interact and process tasks.These systems are categorized into seven categories,each tailored for specific operational styles and autonomy in decision making.The environments in which these agents operate are classified as static,dynamic,fully observable,partially observable,deterministic,stochastic,single-agent,and multiagent,emphasizing the impact of environmental complexity on agent behavior.Agentic AI has revolutionized systems through autonomous decision making and resource optimization,yet challenges persist in aligning AI with human values,ensuring adaptability,and addressing ethical constraints.Future research focuses on multidomain agents,human–AI collaboration,and self-improving systems.This work provides researchers,practitioners,and policymakers with a structured approach to understanding and advancing the rapidly evolving landscape of agentic AI systems.
文摘Large LanguageModels(LLMs)are increasingly utilized for semantic understanding and reasoning,yet their use in sensitive settings is limited by privacy concerns.This paper presents In-Mig,a mobile-agent architecture that integrates LLM reasoning within agents that can migrate across organizational venues.Unlike centralized approaches,In-Mig performs reasoning in situ,ensuring that raw data remains within institutional boundaries while allowing for cross-venue synthesis.The architecture features a policy-scoped memory model,utility-driven route planning,and cryptographic trust enforcement.Aprototype using JADE for mobility and quantizedMistral-7B demonstrates practical feasibility.Evaluation across various scenarios shows that In-Mig achieves 92%similarity to centralized baselines,confirming its utility and strong privacy guarantees.These results suggest that migrating,privacy-preserving LLM agents can effectively support decentralized reasoning in trust-sensitive domains.
文摘While the complexity of fifth-generation wireless networks is being widely commented upon,there is great anticipation for the arrival of the sixth generation(6G),with its enriched capabilities and features.It can easily be imagined that,without proper design,the enrichment of 6G will further increase system complexity.To address this issue,we propose the Agentic-AI Core(A-Core),an artificial intelligence(AI)-empowered,mission-oriented core network architecture for next-generation mobile telecommunications.In A-Core,network capabilities can be added and updated on the fly and further programmed into missions for enabling and offering diverse services to customers.These missions are created and executed by autonomous network agents according to the customer's intent,which may be expressed in natural language.The agents resolve intents from customers into workflows of network capabilities by leveraging a large-scale network AI model and follow the workflows to execute the mission.As an open,agile system architecture,A-Core holds promise for accelerating innovation and greatly reducing standard release times.The advantages of A-Core are demonstrated through two use cases.
基金funded in part through the NIH/NCI Cancer Center Support Grant P30 CA008748.
文摘Artificial Intelligence(AI)is transforming the healthcare landscape,yet many current applications remain narrowly task-specific,constrained by data complexity and inherent biases.This paper explores the emergence of next generation"agentic AI"systems,characterized by advanced autonomy,adaptability,scalability,and prob-abilistic reasoning,which address critical challenges in medical management.These systems enhance various aspects of healthcare,including diagnostics,clinical decision support,treatment planning,patient monitoring,administrative operations,drug discovery,and robotic-assisted surgery.Powered by multimodal AI,agentic systems integrate diverse data sources,iteratively refine outputs,and leverage vast knowledge bases to deliver context-aware,patient-centric care with heightened precision and reduced error rates.These advancements promise to enhance patient outcomes,optimize clinical workflows,and expand the reach of AI-driven solutions.However,their deployment introduces ethical,privacy,and regulatory challenges,emphasizing the need for robust governance frameworks and interdisciplinary collaboration.Agentic AI has the potential to redefine healthcare,driving personalized,efficient,and scalable services while extending its impact beyond clinical settings to global public health initiatives.By addressing disparities and enhancing care delivery in resourcelimited environments,this technology could significantly advance equitable healthcare.Realizing the full po-tential of agentic AI will require sustained research,innovation,and cross-disciplinary partnerships to ensure its responsible and transformative integration into healthcare systems worldwide.
文摘Governance scholarship overwhelmingly treats artificial intelligence as a tool—an object to be regulated,audited,or aligned with human values.This article argues that when organizations treat the outputs of AI agents as institutionally binding,these systems cross a threshold from instruments into institutional actors.Drawing on Goffman’s interaction order and Meyer and Rowan’s institutional isomorphism,we develop a diagnostic framework that specifies when and how AI agents acquire practical actorhood within organizations.We formalize a 2×2 Actorhood Matrix along two axes—discretion granted and institutional embedding—yielding four system types:Tool,Infrastructure,Shadow Actor,and Institutional Actor.The Actorhood Matrix is proposed as a reusable diagnostic method for identifying when AI systems cross the institutional threshold from tools to role occupants.Applying this framework to five empirical cases(Klarna’s AI customer service,GitHub Copilot Workspace,NHS AI triage,Harvey AI legal assistant,and generic FAQ chatbots)demonstrates that institutional actorhood is not a property of technical sophistication but of organizational role assignment.We identify a critical transition zone where override rates fall below ten percent and propose four testable propositions linking discretion,embedding,and temporal persistence to governance outcomes.The article concludes that governance frameworks designed for“tools”are structurally inadequate for systems that have become practical role occupants within institutional settings.
文摘Agentic AI represents a significant advancement in artificial intelligence,enabling proactive agents that can set goals,make decisions,and adapt to changing situations.However,the performance of these systems is heavily dependent on the quality and relevance of the data they process.This research highlights the critical risk posed by faulty,insecure,or contextually inappropriate input data in modern Agentic AI systems.To address this challenge,this study proposes the Autonomous Data Integrity Layer(ADIL).This flexible architecture integrates best practices from security engineering and data science to ensure that Agentic AI systems operate with clean,validated,and contextually relevant data.By focusing on data integrity,ADIL enhances the reliability,accountability,and effectiveness of Agentic AI systems,leading to more trustworthy and robust intelligent agents.
基金supported by the Science and Technology Development Fund,Macao SAR(Nos.0093/2023/RIA2 and 0145/2023/RIA3).
文摘Advancements in large language models(LLMs)have markedly improved the adaptability of artificial intelligence(AI)agents in dynamic and open environments.However,with the growing number and diversity of agents,ensuring secure,reliable,and autonomous collaboration among them has become an urgent and critical challenge.To this end,this letter proposes agent reinforced generation(ARG)to establish a multi-agent system with audit trail functionality,privacy compliance,and autonomous coordination.ARG integrates the model context protocol(MCP)and agent-to-agent(A2A)protocol to define the rules and logic governing agent-to-agent communications as well as agent-to-tool/data engagements.Decentralized autonomous organizations and operations(DAOs)are employed to enable agents to coordinate and execute tasks in a transparent and tamper-resistant manner.Additionally,the operational process of ARG is elaborated from task issuance to completion to validate the auditability and immutability of task coordination and execution.Finally,we highlight five key features of ARG,including parallelism and throughput,scalability across domains and load,fault tolerance and graceful failure,resource efficiency through delegation,as well as data security and privacy protection,positioning it as a promising paradigm for the realization of agentic intelligence.
基金supported by the Ministry of Education and Science of the Republic of North Macedonia through the project“Utilizing AI and National Large Language Models to Advance Macedonian Language Capabilties”。
文摘Artificial intelligence(AI)is reshaping financial systems and services,as intelligent AI agents increasingly form the foundation of autonomous,goal-driven systems capable of reasoning,learning,and action.This review synthesizes recent research and developments in the application of AI agents across core financial domains.Specifically,it covers the deployment of agent-based AI in algorithmic trading,fraud detection,credit risk assessment,roboadvisory,and regulatory compliance(RegTech).The review focuses on advanced agent-based methodologies,including reinforcement learning,multi-agent systems,and autonomous decision-making frameworks,particularly those leveraging large language models(LLMs),contrasting these with traditional AI or purely statistical models.Our primary goals are to consolidate current knowledge,identify significant trends and architectural approaches,review the practical efficiency and impact of current applications,and delineate key challenges and promising future research directions.The increasing sophistication of AI agents offers unprecedented opportunities for innovation in finance,yet presents complex technical,ethical,and regulatory challenges that demand careful consideration and proactive strategies.This review aims to provide a comprehensive understanding of this rapidly evolving landscape,highlighting the role of agent-based AI in the ongoing transformation of the financial industry,and is intended to serve financial institutions,regulators,investors,analysts,researchers,and other key stakeholders in the financial ecosystem.
文摘In recent years,researchers have leveraged single-agent reinforcement learning to boost educational outcomes and deliver personalized interventions;yet this paradigm provides no capacity for inter-agent interaction.Multi-agent reinforcement learning(MARL)overcomes this limitation by allowing several agents to learn simultaneously within a shared environment,each choosing actions that maximize its own or the group's rewards.By explicitly modeling and exploiting agent-to-agent dynamics,MARL can align those interactions with pedagogical goals such as peer tutoring,collaborative problem-solving,or gamified competition,thus opening richer avenues for adaptive and socially informed learning experiences.This survey investigates the impact of MARL on educational outcomes by examining evidence of its effectiveness in enhancing learner performance,engagement,equity,and reducing teacher workload compared to single agent or traditional approaches.It explores the educational domains and pedagogical problems addressed by MARL,identifies the algorithmic families used,and analyzes their influence on learning.The review also assesses experimental settings and evaluation metrics to determine ecological validity,and outlines current challenges and future research directions in applying MARL to education.
