多智能体 社会行为模拟

We design and implement NegotiationGym, an API and user interface for configuring and running multi-agent social simulations focused upon negotiation and cooperation. The NegotiationGym codebase offers a user-friendly, configuration-driven API that enables easy design and customization of simulation scenarios. Agent-level utility functions encode optimization criteria for each agent, and agents can self-optimize by conducting multiple interaction rounds with other agents, observing outcomes, and modifying their strategies for future rounds.

Multi-agent social interaction has clearly benefited from Large Language Models. However, current simulation systems still face challenges such as difficulties in scaling to diverse scenarios and poor reusability due to a lack of modular design. To address these issues, we designed and developed a modular, object-oriented framework that organically integrates various base classes through a hierarchical structure, harvesting scalability and reusability. We inherited the framework to realize common derived classes. Additionally, a memory summarization mechanism is proposed to filter and distill relevant information from raw memory data, prioritizing contextually salient events and interactions. By selecting and combining some necessary derived classes, we customized a specific simulated environment. Utilizing this simulated environment, we successfully simulated human interactions on social media, replicating real-world online social behaviors. The source code for the project will be released and evolve.

Cities are intricate systems where diverse social structures interact, leading to the continuous emergence of complex social phenomena. Analyzing and optimizing these interactions, as well as efficiently managing urban systems, present significant challenges. Multi-Agent Social Simulation (MASS) offers promising solutions for these challenges. However, MASS is not a user-friendly technology for stakeholders involved in urban issues, such as residents and local government officials. This paper presents the development of a participatory policy planning tool based on MASS. We developed a prototype system to explore potential traffic policies, integrating traffic simulation, data visualization, and a web-based interface. A social experiment conducted with residents in Japan demonstrated the potential of the MASS-based planning tool. Additionally, in the paper, we discuss an approach to simulate and evaluate urban traffic policies on the developed MASS-based system, focusing on optimizing the combination of traffic signal timing adjustments and policy implementations.

No abstract available

COVID-19 pandemic has become a major concern due to its rapid spread throughout the world. We can observe some countries are successful in formulating effective strategies for managing the pandemic, while some are struggling. The research is based on the question of formulating effective policies for COVID-19 to reduce community transmission. While many countries are suffering from the pandemic, it is a critical issue that the policymakers should be concerned with formulating effective policies to address the problem. We use computational methods to foresee the future by creating a simulation model based on multi-agent and simulation methodology because it is not always possible to predict the future state of a complex adaptive system. The data are collected through a survey and the literature to calibrate the model parameters to build a constructive and realistic model. Once the model is constructed, the simulation results are compared with the real-world observations to validate the model. The implementation of the model follows an iterative process for improving the validity of the model. This paper presents the conceptual model of the system being investigated and its initial implementation, which needs to be calibrated further with empirical data before using it as a decision support tool.

No abstract available

No abstract available

ChatGPT, the AI-powered chatbot with a massive user base of hundreds of millions, has become a global phenomenon. However, the use of Conversational AI Systems (CAISs) like ChatGPT for research in the field of Social Simulation is still limited. Specifically, there is no evidence of its usage in Agent-Based Social Simulation (ABSS) model design. This paper takes a crucial first step toward exploring the untapped potential of this emerging technology in the context of ABSS model design. The research presented here demonstrates how CAISs can facilitate the development of innovative conceptual ABSS models in a concise timeframe and with minimal required upfront case-based knowledge. By employing advanced prompt engineering techniques and adhering to the Engineering ABSS framework, we have constructed a comprehensive prompt script that enables the design of conceptual ABSS models with or by the CAIS. A proof-of-concept application of the prompt script, used to generate the conceptual ABSS model for a case study on the impact of adaptive architecture in a museum environment, illustrates the practicality of the approach. Despite occasional inaccuracies and conversational divergence, the CAIS proved to be a valuable companion for ABSS modellers.

: This paper aims to improve the transparency of agent-based social simulation (ABSS) models and make it easier for various actors engaging with these models to make sense of them. It studies what ABSS is and juxtaposes its basic conceptual elements with insights from the agency/structure debate in social theory to propose a framework that captures the ‘conceptual anatomy’ of ABSS models in a simple and intuitive way. The five elements of the framework are: agency, social structure, environment, actions and interactions, and temporality. The paper also examines what is meant by the transparency or opacity of ABSS in the rapidly growing literature on the epistemology of computer simulations. It deconstructs the methodological criticism that ABSS models are black boxes by identifying multiple categories of transparency/opacity. It argues that neither opacity nor transparency is intrinsic to ABSS. Instead, they are dependent on research habitus - practices that are developed in a research field that are shaped by structure of the field and available resources. It discusses the ways in which thinking about the conceptual anatomy of ABSS can improve its transparency.

Social simulation is transforming traditional social science research by modeling human behavior through interactions between virtual individuals and their environments. With recent advances in large language models (LLMs), this approach has shown growing potential in capturing individual differences and predicting group behaviors. However, existing methods face alignment challenges related to the environment, target users, interaction mechanisms, and behavioral patterns. To this end, we introduce SocioVerse, an LLM-agent-driven world model for social simulation. Our framework features four powerful alignment components and a user pool of 10 million real individuals. To validate its effectiveness, we conducted large-scale simulation experiments across three distinct domains: politics, news, and economics. Results demonstrate that SocioVerse can reflect large-scale population dynamics while ensuring diversity, credibility, and representativeness through standardized procedures and minimal manual adjustments.

The study of social emergence has long been a central focus in social science. Traditional modeling approaches, such as rule-based Agent-Based Models (ABMs), struggle to capture the diversity and complexity of human behavior, particularly the irrational factors emphasized in behavioral economics. Recently, large language model (LLM) agents have gained traction as simulation tools for modeling human behavior in social science and role-playing applications. Studies suggest that LLMs can account for cognitive biases, emotional fluctuations, and other non-rational influences, enabling more realistic simulations of socio-economic dynamics. In this work, we introduce TwinMarket, a novel multi-agent framework that leverages LLMs to simulate socio-economic systems. Specifically, we examine how individual behaviors, through interactions and feedback mechanisms, give rise to collective dynamics and emergent phenomena. Through experiments in a simulated stock market environment, we demonstrate how individual actions can trigger group behaviors, leading to emergent outcomes such as financial bubbles and recessions. Our approach provides valuable insights into the complex interplay between individual decision-making and collective socio-economic patterns.

: The academic study and the applied use of agent-based modelling of social processes has matured considerably over the last thirty years. The time is now right to engage seriously with the ethics and responsible practice of agent-based social simulation. In this paper, we first outline the many reasons why it is appropriate to explore an ethics of agent-based modelling and how ethical issues arise in its practice and organisation. We go on to discuss different approaches to standardisation as a way of supporting responsible practice. Some of the main conclusions are organised as provisions in a draft code of ethics. We intend for this draft to be further developed by the community before being adopted by individuals and groups within the field informally or formally.

Can generative agents be trusted in multimodal environments? Despite advances in large language and vision-language models that enable agents to act autonomously and pursue goals in rich settings, their ability to reason about safety, coherence, and trust across modalities remains limited. We introduce a reproducible simulation framework for evaluating agents along three dimensions: (1) safety improvement over time, including iterative plan revisions in text-visual scenarios; (2) detection of unsafe activities across multiple categories of social situations; and (3) social dynamics, measured as interaction counts and acceptance ratios of social exchanges. Agents are equipped with layered memory, dynamic planning, multimodal perception, and are instrumented with SocialMetrics, a suite of behavioral and structural metrics that quantifies plan revisions, unsafe-to-safe conversions, and information diffusion across networks. Experiments show that while agents can detect direct multimodal contradictions, they often fail to align local revisions with global safety, reaching only a 55 percent success rate in correcting unsafe plans. Across eight simulation runs with three models - Claude, GPT-4o mini, and Qwen-VL - five agents achieved average unsafe-to-safe conversion rates of 75, 55, and 58 percent, respectively. Overall performance ranged from 20 percent in multi-risk scenarios with GPT-4o mini to 98 percent in localized contexts such as fire/heat with Claude. Notably, 45 percent of unsafe actions were accepted when paired with misleading visuals, showing a strong tendency to overtrust images. These findings expose critical limitations in current architectures and provide a reproducible platform for studying multimodal safety, coherence, and social dynamics.

Social Media is used by many as a source of information for current world events, followed by publicly sharing their sentiment about these events. However, when the shared information is not trustworthy and receives a large number of interactions, it alters the public’s perception of authentic and false information, particularly when the origin of these stories comes from malicious sources. Over the past decade, there has been an influx of users on the Twitter social network, many of them automated bot accounts with the objective of participating in misinformation campaigns that heavily influence user susceptibility to fake information. This can affect public opinion on real-life matters, as previously seen in the 2020 presidential elections and the current COVID-19 epidemic, both plagued with misinformation. In this paper, we propose an agent-based social simulation environment that utilizes the social network Twitter, with the objective of evaluating how the beliefs of agents representing regular Twitter users can be influenced by malicious users scattered throughout Twitter with the sole purpose of spreading misinformation. We applied two scenarios to compare how these regular agents behave in the Twitter network, with and without malicious agents, to study how much influence malicious agents have on the general susceptibility of the regular users. To achieve this, we implemented a belief value system to measure how impressionable an agent is when encountering misinformation and how its behavior gets affected. The results indicated similar outcomes in the two scenarios as the affected belief value changed for these regular agents, exhibiting belief in the misinformation. Although the change in belief value occurred slowly, it had a profound effect when the malicious agents were present, as many more regular agents started believing in misinformation.

The development of citizenship competences plays an important role in a complex system like society. Thus, to analyze how such competences impact other contexts is a great challenge because this kind of study involves the work with people and the use of variables that depend on human behaviors. In this sense, many studies have highlighted the advantage of using simulation systems and tools. In particular, the agent-based social simulation field relies upon the Semantic Web to manage knowledge representation in social scenarios. This study focuses on how citizenship competences impact conflict resolution. Moreover, a simulation model in which citizens interact to resolve conflicts by considering citizenship competences and conflict resolution styles is also introduced. It was developed in NetLogo together with an extension that connects it with the ontology of competences. Results show that the higher interactions of citizens-conflicts, the higher level of citizenship competences, and the number of conflicts solved is higher when using citizenship competences.

During the COVID-19 crisis there have been many difficult decisions governments and other decision makers had to make. E.g. do we go for a total lock down or keep schools open? How many people and which people should be tested? Although there are many good models from e.g. epidemiologists on the spread of the virus under certain conditions, these models do not directly translate into the interventions that can be taken by government. Neither can these models contribute to understand the economic and/or social consequences of the interventions. However, effective and sustainable solutions need to take into account this combination of factors. In this paper, we propose an agent-based social simulation tool, ASSOCC, that supports decision makers understand possible consequences of policy interventions, but exploring the combined social, health and economic consequences of these interventions.

With the rapid advancement of large language models (LLMs), recent years have witnessed many promising studies on leveraging LLM-based agents to simulate human social behavior. While prior work has demonstrated significant potential across various domains, much of it has focused on specific scenarios involving a limited number of agents and has lacked the ability to adapt when errors occur during simulation. To overcome these limitations, we propose a novel LLM-agent-based simulation platform called \textit{GenSim}, which: (1) \textbf{Abstracts a set of general functions} to simplify the simulation of customized social scenarios; (2) \textbf{Supports one hundred thousand agents} to better simulate large-scale populations in real-world contexts; (3) \textbf{Incorporates error-correction mechanisms} to ensure more reliable and long-term simulations. To evaluate our platform, we assess both the efficiency of large-scale agent simulations and the effectiveness of the error-correction mechanisms. To our knowledge, GenSim represents an initial step toward a general, large-scale, and correctable social simulation platform based on LLM agents, promising to further advance the field of social science.

Agent based models (ABMs) simulate actions and interactions of autonomous agents/groups and their effect on systems as a whole, accounting for learning without assuming perfect rationality or complete knowledge. ABMs are an increasingly popular approach to studying complex, spatially distributed socio-environmental systems, but have still to become an established approach in the sense of being one that is expected by those wanting to explore scenarios in such systems. Partly, this is an issue of awareness – ABM is still new enough that many people have not heard of it; partly, it is an issue of confidence – ABM has more to do to prove itself if it is to become a preferred method. This paper will identify advances in the craft and deployment of ABM needed if ABM is to become an accepted part of mainstream science for policy or stakeholders. The conduct of ABM has, over the last decade, seen a transition from using abstracted representations of systems (supporting theory-led thought experiments) to more accessible representations derived empirically (to deliver more applied analysis). This has enhanced the perception of potential users of ABM outputs that the latter are salient and credible. Empirical ABM is not, however, a panacea, as it demands more computing and data resources, limiting applications to domains where data exist along with suitable environmental models where these are required. Further, empirical ABM is still facing serious questions of validation and the ontology used to describe the system in the first place. Using Geoffrey A. Moore’s Crossing the Chasm as a lens, we argue that the way ahead for ABM lies in identifying the niches in which it can best demonstrate its advantages, working with collaborators to demonstrate that it can deliver on its promises. This leads us to identify several areas where work is needed.

Social media has emerged as a cornerstone of social movements, wielding significant influence in driving societal change. Simulating the response of the public and forecasting the potential impact has become increasingly important. However, existing methods for simulating such phenomena encounter challenges concerning their efficacy and efficiency in capturing the behaviors of social movement participants. In this paper, we introduce a hybrid framework HiSim for social media user simulation, wherein users are categorized into two types. Core users are driven by Large Language Models, while numerous ordinary users are modeled by deductive agent-based models. We further construct a Twitter-like environment to replicate their response dynamics following trigger events. Subsequently, we develop a multi-faceted benchmark SoMoSiMu-Bench for evaluation and conduct comprehensive experiments across real-world datasets. Experimental results demonstrate the effectiveness and flexibility of our method.

No abstract available

No abstract available

No abstract available

No abstract available

This article proposes a social simulation paradigm based on the GPT-3.5 large language model. It involves constructing Generative Agents that emulate human cognition, memory, and decision-making frameworks, along with establishing a virtual social system capable of stable operation and an insertion mechanism for standardized public events. The project focuses on simulating a township water pollution incident, enabling the comprehensive examination of a virtual government's response to a specific public administration event. Controlled variable experiments demonstrate that the stored memory in generative agents significantly influences both individual decision-making and social networks. The Generative Agent-Based Simulation System introduces a novel approach to social science and public administration research. Agents exhibit personalized customization, and public events are seamlessly incorporated through natural language processing. Its high flexibility and extensive social interaction render it highly applicable in social science investigations. The system effectively reduces the complexity associated with building intricate social simulations while enhancing its interpretability.

No abstract available

No abstract available

The primary hypothesis stated by this paper is that the use of social choice theory in Ambient Intelligence systems can improve significantly users' satisfaction when accessing shared resources. A research methodology based on agent based social simulations is employed to support this hypothesis and to evaluate these benefits. The result is a sixfold contribution summarized as follows. Firstly, several considerable differences between this application case and the most prominent social choice application, political elections, have been found and described. Secondly, given these differences, a number of metrics to evaluate different voting systems in this scope have been proposed and formalized. Thirdly, given the presented application and the metrics proposed, the performance of a number of well known electoral systems is compared. Fourthly, as a result of the performance study, a novel voting algorithm capable of obtaining the best balance between the metrics reviewed is introduced. Fifthly, to improve the social welfare in the experiments, the voting methods are combined with cluster analysis techniques. Finally, the article is complemented by a free and open-source tool, VoteSim, which ensures not only the reproducibility of the experimental results presented, but also allows the interested reader to adapt the case study presented to different environments.

Generative Agent-Based Modeling (GABM) is an emerging simulation paradigm that combines the reasoning abilities of Large Language Models with traditional Agent-Based Modeling to replicate complex social behaviors, including interactions on social media. While prior work has focused on localized phenomena such as opinion formation and information spread, its potential to capture global network dynamics remains underexplored. This paper addresses this gap by analyzing GABM-based social media simulations through the lens of the Friendship Paradox (FP), a counterintuitive phenomenon where individuals, on average, have fewer friends than their friends. We propose a GABM framework for social media simulations, featuring generative agents that emulate real users with distinct personalities and interests. Using Twitter datasets on the US 2020 Election and the QAnon conspiracy, we show that the FP emerges naturally in GABM simulations. Consistent with real-world observations, the simulations unveil a hierarchical structure, where agents preferentially connect with others displaying higher activity or influence. Additionally, we find that infrequent connections primarily drive the FP, reflecting patterns in real networks. These findings validate GABM as a robust tool for modeling global social media phenomena and highlight its potential for advancing social science by enabling nuanced analysis of user behavior.

We present a novel, open-source social network simulation framework, MOSAIC, where generative language agents predict user behaviors such as liking, sharing, and flagging content. This simulation combines LLM agents with a directed social graph to analyze emergent deception behaviors and gain a better understanding of how users determine the veracity of online social content. By constructing user representations from diverse fine-grained personas, our system enables multi-agent simulations that model content dissemination and engagement dynamics at scale. Within this framework, we evaluate three different content moderation strategies with simulated misinformation dissemination, and we find that they not only mitigate the spread of non-factual content but also increase user engagement. In addition, we analyze the trajectories of popular content in our simulations, and explore whether simulation agents'articulated reasoning for their social interactions truly aligns with their collective engagement patterns. We open-source our simulation software to encourage further research within AI and social sciences.

The calculation of social simulation using opinion dynamics is shown. We employ a new opinion dynamics proposed by Ishii deals with both trust and distrust among people as very simple extension of the Bounded Confidence Model. When a society is divided into two, if the groups are closely connected and the trust within the group is strong, the division can develop into a conflict. However, if the connections within the group are sparse, or if the trust and distrust relationships within the group are half and half, consensus building within the group may not be achieved and conflict may not occur. The social simulation of divided society, we can avoid conflict between two groups by sparse connection of people in each group and the weak trust within the group.

No abstract available

Social network simulation plays a crucial role in addressing various challenges within social science. It offers extensive applications such as state prediction, phenomena explanation, and policy-making support, among others. In this work, we harness the formidable human-like capabilities exhibited by large language models (LLMs) in sensing, reasoning, and behaving, and utilize these qualities to construct the S$^3$ system (short for $\textbf{S}$ocial network $\textbf{S}$imulation $\textbf{S}$ystem). Adhering to the widely employed agent-based simulation paradigm, we employ prompt engineering and prompt tuning techniques to ensure that the agent's behavior closely emulates that of a genuine human within the social network. Specifically, we simulate three pivotal aspects: emotion, attitude, and interaction behaviors. By endowing the agent in the system with the ability to perceive the informational environment and emulate human actions, we observe the emergence of population-level phenomena, including the propagation of information, attitudes, and emotions. We conduct an evaluation encompassing two levels of simulation, employing real-world social network data. Encouragingly, the results demonstrate promising accuracy. This work represents an initial step in the realm of social network simulation empowered by LLM-based agents. We anticipate that our endeavors will serve as a source of inspiration for the development of simulation systems within, but not limited to, social science.

© 2020, University of Surrey. All rights reserved. The Overview, Design concepts and Details (ODD) protocol for describing Individual-and Agent-Based Models (ABMs) is now widely accepted and used to document such models in journal articles. As a standardized document for providing a consistent, logical and readable account of the structure and dynamics of ABMs, some research groups also find it useful as a workflow for model design. Even so, there are still limitations to ODD that obstruct its more widespread adoption. Such limitations are discussed and addressed in this paper: the limited availability of guidance on how to use ODD; the length of ODD documents; limitations of ODD for highly complex models; lack of sufficient details of many ODDs to enable reimplementation without access to the model code; and the lack of provision for sections in the document structure covering model design ratio-nale, the model’s underlying narrative, and the means by which the model’s fitness for purpose is evaluated. We document the steps we have taken to provide better guidance on: structuring complex ODDs and an ODD summary for inclusion in a journal article (with full details in supplementary material; Table 1); using ODD to point readers to relevant sections of the model code; update the document structure to include sections on model rationale and evaluation. We also further advocate the need for standard descriptions of simulation experiments and argue that ODD can in principle be used for any type of simulation model. Thereby ODD would provide a lingua franca for simulation modelling.

The increase in readily available computational power raises the possibility that direct agent-based modeling can play a key role in the analysis of epidemiological population dynamics. Specifically, the objective of this work is to develop a robust agent-based computational framework to investigate the emergent structure of Susceptible-Infected-Removed/Recovered (SIR)-type populations and variants thereof, on a global planetary scale. To accomplish this objective, we develop a planet-wide model based on interaction between discrete entities (agents), where each agent on the surface of the planet is initially uninfected. Infections are then seeded on the planet in localized regions. Contracting an infection depends on the characteristics of each agent—i.e. their susceptibility and contact with the seeded, infected agents. Agent mobility on the planet is dictated by social policies, for example such as “shelter in place”, “complete lockdown”, etc. The global population is then allowed to evolve according to infected states of agents, over many time periods, leading to an SIR population. The work illustrates the construction of the computational framework and the relatively straightforward application with direct, non-phenomenological, input data. Numerical examples are provided to illustrate the model construction and the results of such an approach.

No abstract available

Background Social network interventions targeted at children and adolescents can have a substantial effect on their health behaviors, including physical activity. However, designing successful social network interventions is a considerable research challenge. In this study, we rely on social network analysis and agent-based simulations to better understand and capitalize on the complex interplay of social networks and health behaviors. More specifically, we investigate criteria for selecting influence agents that can be expected to produce the most successful social network health interventions. Objective The aim of this study was to test which selection criterion to determine influence agents in a social network intervention resulted in the biggest increase in physical activity in the social network. To test the differences among the selection criteria, a computational model was used to simulate different social network interventions and observe the intervention’s effect on the physical activity of primary and secondary school children within their school classes. As a next step, this study relied on the outcomes of the simulated interventions to investigate whether social network interventions are more effective in some classes than others based on network characteristics. Methods We used a previously validated agent-based model to understand how physical activity spreads in social networks and who was influencing the spread of behavior. From the observed data of 460 participants collected in 26 school classes, we simulated multiple social network interventions with different selection criteria for the influence agents (ie, in-degree centrality, betweenness centrality, closeness centrality, and random influence agents) and a control condition (ie, no intervention). Subsequently, we investigated whether the detected variation of an intervention’s success within school classes could be explained by structural characteristics of the social networks (ie, network density and network centralization). Results The 1-year simulations showed that social network interventions were more effective compared with the control condition (beta=.30; t100=3.23; P=.001). In addition, the social network interventions that used a measure of centrality to select influence agents outperformed the random influence agent intervention (beta=.46; t100=3.86; P<.001). Also, the closeness centrality condition outperformed the betweenness centrality condition (beta=.59; t100=2.02; P=.046). The anticipated interaction effects of the network characteristics were not observed. Conclusions Social network intervention can be considered as a viable and promising intervention method to promote physical activity. We demonstrated the usefulness of applying social network analysis and agent-based modeling as part of the social network interventions’ design process. We emphasize the importance of selecting the most successful influence agents and provide a better understanding of the role of network characteristics on the effectiveness of social network interventions.

Agent-Based Modelling (ABM) has emerged as an essential tool for simulating social networks, encompassing diverse phenomena such as information dissemination, influence dynamics, and community formation. However, manually configuring varied agent interactions and information flow dynamics poses challenges, often resulting in oversimplified models that lack real-world generalizability. Integrating modern Large Language Models (LLMs) with ABM presents a promising avenue to address these challenges and enhance simulation fidelity, leveraging LLMs' human-like capabilities in sensing, reasoning, and behavior. In this paper, we propose a novel framework utilizing LLM-empowered agents to simulate social network users based on their interests and personality traits. The framework allows for customizable agent interactions resembling various social network platforms, including mechanisms for content resharing and personalized recommendations. We validate our framework using a comprehensive Twitter dataset from the 2020 US election, demonstrating that LLM-agents accurately replicate real users' behaviors, including linguistic patterns and political inclinations. These agents form homogeneous ideological clusters and retain the main themes of their community. Notably, preference-based recommendations significantly influence agent behavior, promoting increased engagement, network homophily and the formation of echo chambers. Overall, our findings underscore the potential of LLM-agents in advancing social media simulations and unraveling intricate online dynamics.

It is widely recognized that the Web contributes to user polarization, and such polarization affects not just politics but also peoples’ stances about public health, such as vaccination. Understanding polarization in social networks is challenging because it depends not only on user attitudes but also their interactions and exposure to information. We adopt Social Judgment Theory to operationalize attitude shift and model user behavior based on empirical evidence from past studies. We design a social simulation to analyze how content sharing affects user satisfaction and polarization in a social network. We investigate the influence of varying tolerance in users and selectively exposing users to congenial views. We find that (1) higher user tolerance slows down polarization and leads to lower user satisfaction; (2) higher selective exposure leads to higher polarization and lower user reach; and (3) both higher tolerance and higher selective exposure lead to a more homophilic social network.

Abstract Despite often being perceived as morally objectionable, stereotypes are a common feature of social groups, a phenomenon that has often been attributed to biased motivations or limits on the ability to process information. We argue that one reason for this continued prevalence is that preexisting expectations about how others will behave, in the context of social coordination, can change the behaviors of one’s social partners, creating the very stereotype one expected to see, even in the absence of other potential sources of stereotyping. We use a computational model of dynamic social coordination to illustrate how this “feedback loop” can emerge, engendering and entrenching role-consistent stereotypic behavior and then show that human behavior on the task generates a comparable feedback loop. Notably, people’s choices on the task are not related to social dominance or system justification, suggesting biased motivations are not necessary to maintain these stereotypes.

Multiagent social network simulations are an avenue that can bridge the communication gap between the public and private platforms in order to develop solutions to a complex array of issues relating to online safety. While there are significant challenges relating to the scale of multiagent simulations, efficient learning from observational and interventional data to accurately model micro and macro-level emergent effects, there are equally promising opportunities not least with the advent of large language models that provide an expressive approximation of user behavior. In this position paper, we review prior art relating to social network simulation, highlighting challenges and opportunities for future work exploring multiagent security using agent-based models of social networks

The present work is an innovative educational strategy that uses a Final Integrative Work (FIW) as a method of evaluation of subjects of the Computer Engineering degree where students learn different subjects such as Artificial Intelligence and Databases, through real world problems related to COVID-19. The evaluation process through the FIW is based on several skills acquisition and by measuring the way in which students apply concepts of Databases and intelligent agents by means of numerical simulations that involves social behavior in times of the COVID-19 pandemic in the province of Tucumán, in the northwest of Argentina. The students carried out simulations of a multiagent system through the tool Netlogo, applying rules with a high impact factor for tackling a decision making problem. The results observed suggest that a paradigm shift in the degree evaluation processes is possible and necessary.

The rapid development of online social media has significantly promoted product diffusion in online social networks (PDOSN). However, prior studies focusing on irrational behavior, such as overconfidence, in PDOSN are scarce. To investigate the effect of overconfidence on PDOSN, this study combined overconfidence and an evolutionary game to conduct a multiagent simulation on PDOSN. This combined method provided an effective reference to examine product diffusion in the context of irrational behavior. After careful consideration, this study identified three overconfidence scenarios, benefit, cost, and benefit and cost overconfidence, developed a multiagent simulation model for PDOSN using various overconfidence scenarios, and conducted a comparison with real‐world cases to validate the model’s feasibility. The findings indicated that adoption benefits and betrayal penalties had a positive effect on the results in all models, while adoption costs had the opposite effect. When benefit and cost overconfidence occurred simultaneously, benefit overconfidence offset the negative effect of cost overconfidence. Moderate connectivity, a large number of core nodes, and high reconnection probability fully promoted product diffusion. Benefit overconfidence and cost overconfidence had a significant impact on the results in different networks. As such, this study combined psychological theory with simulation methods, providing insights for future research on product diffusion.

Signed digraphs have been developed in multiagent systems to explain cooperative and competitive interactions among agents. This article explores the collective behavior for a group of generic linear agents interacting by the random interaction manner on signed digraphs in the presence of multiple leaders. In the setting of random interactions, the information interaction between neighboring agents may fail at each time step, which is described by a Bernoulli random variable. The global almost sure convergence of the system is solved by using the convergence analysis method of products of infinite substochastic matrices. Moreover, an application about opinion dynamics of social networks is given. In the end, the theoretical research results are verified through simulation experiments.

No abstract available

Meme automaton is an adaptive entity that autonomously acquires an increasing level of capability and intelligence through embedded memes evolving independently or via social interactions. This paper begins a study on memetic multiagent system (MeMAS) toward human-like social agents with memetic automaton. We introduce a potentially rich meme-inspired design and operational model, with Darwin’s theory of natural selection and Dawkins’ notion of a meme as the principal driving forces behind interactions among agents, whereby memes form the fundamental building blocks of the agents’ mind universe. To improve the efficiency and scalability of MeMAS, we propose memetic agents with structured memes in this paper. Particularly, we focus on meme selection design where the commonly used elitist strategy is further improved by assimilating the notion of like-attracts-like in the human learning. We conduct experimental study on multiple problem domains and show the performance of the proposed MeMAS on human-like social behavior.

As AI technology continues to develop, more and more agents will become capable of long term autonomy alongside people. Thus, a recent line of research has studied the problem of teaching autonomous agents the concept of ethics and human social norms. Most existing work considers the case of an individual agent attempting to learn a predefined set of rules. In reality however, social norms are not always pre-defined and are very difficult to represent algorithmically. Moreover, the basic idea behind the social norms concept is ensuring that one's actions do not negatively influence others' utilities, which is inherently a multiagent concept. Thus, here we investigate a way to teach agents, as a team, how to act according to human social norms. In this research, we introduce the STAR framework used to teach an ad hoc team of agents to act in accordance with human social norms. Using a hybrid team (agents and people), when taking an action considered to be socially unacceptable, the agents receive negative feedback from the human teammate(s) who has(have) an awareness of the team's norms. We view STAR as an important step towards teaching agents to act more consistently with respect to human morality.

We consider behavior of agents in a long-term multiagent coopetitive setting in which agents vary their cooperative and competitive stances over time. Using the game of Diplomacy as a testbed, we study how successful agents vary their coopetitive behavior, developing a new “style of play” (SoP) characterization of player behavior. We assess five novel SoP hypotheses about successful behavior. We propose two algorithms to automatically compute an agent’s SoP vector and describe the important factors in this computation. As an agent’s SoP depends on the game state and its perception of threat, we develop a novel “means, motive, and opportunity” (MMO) model of threat and show that we can predict threats effectively using this model. We provide novel insights into how agents should behave to more successfully achieve their goals in long-term coopetitive settings.

Recent advances in Large Language Models (LLM) have led to a new class of autonomous agents, renewing and expanding interest in the area. LLM-powered Multiagent Systems (MAS) have thus emerged, both for assistive and simulation purposes, yet tools for realistic human behavior simulation -- with its distinctive challenges and opportunities -- remain underdeveloped. Existing MAS libraries and tools lack fine-grained persona specifications, population sampling facilities, experimentation support, and integrated validation, among other key capabilities, limiting their utility for behavioral studies, social simulation, and related applications. To address these deficiencies, in this work we introduce TinyTroupe, a simulation toolkit enabling detailed persona definitions (e.g., nationality, age, occupation, personality, beliefs, behaviors) and programmatic control via numerous LLM-driven mechanisms. This allows for the concise formulation of behavioral problems of practical interest, either at the individual or group level, and provides effective means for their solution. TinyTroupe's components are presented using representative working examples, such as brainstorming and market research sessions, thereby simultaneously clarifying their purpose and demonstrating their usefulness. Quantitative and qualitative evaluations of selected aspects are also provided, highlighting possibilities, limitations, and trade-offs. The approach, though realized as a specific Python implementation, is meant as a novel conceptual contribution, which can be partially or fully incorporated in other contexts. The library is available as open source at https://github.com/microsoft/tinytroupe.

No abstract available

Significance The emergence of group bias has a long history in social psychology, including the positing of innate biases. Experimental evidence is fundamentally limited because every brain is both a product of experience and evolution. With cognitive models, one can control both the cognitive architecture and experience that shape its learning. We show that multiagent reinforcement learning can vastly expand the social scenarios that can be studied using a model that learns “from scratch.” This is possible because deep reinforcement learning agents model reward-guided decision-making and learn to parse the world from raw sensory input. Here, we explore the emergence of group bias in simulation and the factors mediating it. Importantly, these biases are overcome with sufficient experience.

Evolution of cooperation and competition can appear when multiple adaptive agents share a biological, social, or technological niche. In the present work we study how cooperation and competition emerge between autonomous agents that learn by reinforcement while using only their raw visual input as the state representation. In particular, we extend the Deep Q-Learning framework to multiagent environments to investigate the interaction between two learning agents in the well-known video game Pong. By manipulating the classical rewarding scheme of Pong we show how competitive and collaborative behaviors emerge. We also describe the progression from competitive to collaborative behavior when the incentive to cooperate is increased. Finally we show how learning by playing against another adaptive agent, instead of against a hard-wired algorithm, results in more robust strategies. The present work shows that Deep Q-Networks can become a useful tool for studying decentralized learning of multiagent systems coping with high-dimensional environments.

As a common social phenomenon, group imitation behavior holds significant research value in the fields of biological group collaboration and artificial swarm intelligence. This paper constructs a behavior imitation model integrating information dissemination mechanisms based on the theory of multiagent systems. The model aims to reveal the influence mechanism of group dynamic characteristics and information interaction intensity on the consistency of group behavior. The model architecture consists of two parts. The first part is an information dissemination model improved upon the SIR model, which introduces a perception radius to analyze how neighboring interactions affect the information diffusion rate. The second part is a multiagent group aggregation model based on social mechanics, enabling individuals to form groups through parameters like attraction, repulsion, speed, and movement direction. Groups spread aggregation and imitation information through interactions with neighboring individuals. Then, based on the breadth of the information they receive, they imitate exemplary groups through intergroup imitation effects. Through complex system simulations, the experimental results show that the consistency of group imitation behavior is positively correlated with the perception radius of individuals. This research provides a new modeling framework and analytical perspective for understanding the emergence mechanism of swarm intelligence.

No abstract available

No abstract available

We propose a model for multi-objective optimization, a credo, for agents in a system that are configured into multiple groups (i.e., teams). Our model of credo regulates how agents optimize their behavior for the groups they belong to. We evaluate credo in the context of challenging social dilemmas with reinforcement learning agents. Our results indicate that the interests of teammates, or the entire system, are not required to be fully aligned for achieving globally beneficial outcomes. We identify two scenarios without full common interest that achieve high equality and significantly higher mean population rewards compared to when the interests of all agents are aligned.

No abstract available

No abstract available

The vast majority of research in social choice theory, be it axiomatic characterizations, welfare bounds, or equilibrium analysis, makes the implicit assumption that the entire affected population votes. However this ideal description is very far from the situation in practice, and even more so in the age of online voting and various direct democracy initiatives---often, a handful of avid voters effectively make the decisions for the silent majority. Our starting point is that abstention and partial participation are not a curious exception but the norm. We therefore argue that models of abstention should be given at least as much attention (if not more) as models of strategic behavior, and theoretical properties of voting rules such as welfare and fairness guarantees must be evaluated in light of such models in order to be relevant. To that end, we call the multiagent systems research community to design more effective tools to better handle both causes and effects of low participation; and highlight promising directions.

In many biological and artificial systems, adaptive behavior emerges not from centralized control but from the interaction of many simple components. However, when a multiagent system behaves as a unit, components face the structural credit assignment problem: inability to discern how their individual action contributed to the global outcome of the group. In our study, we investigate this problem using a multi-agent multi-armed bandit (MAMAB) task within a dynamic environment and under a collective objective. Agents choose actions individually but receive a reward that is an average of their individual rewards. In a simple asocial condition in which the only information the agents are provided is the collective reward, they fail to perform adequately, confirming the existence of the credit assignment problem. We implement two conditions that, in addition to the reward signal, include a form of social learning: decision-biasing, where action selection is directly affected by the popularity of available options, and value-shaping, where popularity modifies option value estimates instead. We show that both mechanisms improve group performance, but value-shaping yields higher accuracy and stability. Through targeted intervention experiments, we demonstrate that the result is due to three processes: recruitment bias, amplification of emerging consensus, and recovery after environmental change. These results show that the credit assignment problem, even in dynamic environments, can, under specific conditions, be solved by the self-organization of collective behavior.

In this article, we consider a collective decision-making process in a network of agents described by a nonlinear interconnected dynamical model with sigmoidal nonlinearities and signed interaction graph. The decisions are encoded in the equilibria of the system. The aim is to investigate this multiagent system when the signed graph representing the community is not structurally balanced and in particular as we vary its frustration, i.e., its distance to structural balance. The model exhibits bifurcations, and a “social effort” parameter, added to the model to represent the strength of the interactions between the agents, plays the role of bifurcation parameter in our analysis. We show that, as the social effort increases, the decision-making dynamics exhibit a pitchfork bifurcation behavior where, from a deadlock situation of “no decision” (i.e., the origin is the only globally stable equilibrium point), two possible (alternative) decision states for the community are achieved (corresponding to two nonzero locally stable equilibria). The value of social effort for which the bifurcation is crossed (and a decision is reached) increases with the frustration of the signed network.

Significance The fact that humans enforce and comply with norms is an important reason why humans enjoy higher levels of cooperation and welfare than other animals. Some norms are relatively easy to explain: They may prohibit obviously harmful or uncooperative actions. But many norms are not easy to explain. For example, most cultures prohibit eating certain kinds of foods, and almost all societies have rules about what constitutes appropriate clothing, language, and gestures. Using a computational model focused on learning shows that apparently pointless rules can have an indirect effect on welfare. They can help agents learn how to enforce and comply with norms in general, improving the group’s ability to enforce norms that have a direct effect on welfare. How do societies learn and maintain social norms? Here we use multiagent reinforcement learning to investigate the learning dynamics of enforcement and compliance behaviors. Artificial agents populate a foraging environment and need to learn to avoid a poisonous berry. Agents learn to avoid eating poisonous berries better when doing so is taboo, meaning the behavior is punished by other agents. The taboo helps overcome a credit assignment problem in discovering delayed health effects. Critically, introducing an additional taboo, which results in punishment for eating a harmless berry, further improves overall returns. This “silly rule” counterintuitively has a positive effect because it gives agents more practice in learning rule enforcement. By probing what individual agents have learned, we demonstrate that normative behavior relies on a sequence of learned skills. Learning rule compliance builds upon prior learning of rule enforcement by other agents. Our results highlight the benefit of employing a multiagent reinforcement learning computational model focused on learning to implement complex actions.

No abstract available

No abstract available

As it is getting easier to obtain reams of data on human behavior via ubiquitous devices, it is becoming obvious that we must work on two conflicting research directions for realizing multiagent-based social simulations; creating large-scale simulations and elaborating fine-scale human behavior models. The challenge in this paper is to achieve massively urban traffic simulations with fine-grained levels of driving behavior. Toward our objective, we show the design and implementation of a multiagent-based simulation platform, that enables us to execute massive but sophisticated multiagent traffic simulations. We show the capability of the developed platform to reproduce the urban traffic with a social experiment scenario. We investigate its potential to analyze the traffic from both macroscopic and microscopic viewpoints.

No abstract available

Human social behavior is structured by relationships. We form teams, groups, tribes, and alliances at all scales of human life. These structures guide multi-agent cooperation and competition, but when we observe others these underlying relationships are typically unobservable and hence must be inferred. Humans make these inferences intuitively and flexibly, often making rapid generalizations about the latent relationships that underlie behavior from just sparse and noisy observations. Rapid and accurate inferences are important for determining who to cooperate with, who to compete with, and how to cooperate in order to compete. Towards the goal of building machine-learning algorithms with human-like social intelligence, we develop a generative model of multiagent action understanding based on a novel representation for these latent relationships called Composable Team Hierarchies (CTH). This representation is grounded in the formalism of stochastic games and multi-agent reinforcement learning. We use CTH as a target for Bayesian inference yielding a new algorithm for understanding behavior in groups that can both infer hidden relationships as well as predict future actions for multiple agents interacting together. Our algorithm rapidly recovers an underlying causal model of how agents relate in spatial stochastic games from just a few observations. The patterns of inference made by this algorithm closely correspond with human judgments and the algorithm makes the same rapid generalizations that people do.

No abstract available

A ubiquitous type of collective behavior and decision-making is the coordinated motion of bird flocks, fish schools, and human crowds. Collective decisions to move in the same direction, turn right or left, or split into subgroups arise in a self-organized fashion from local interactions between individuals without central plans or designated leaders. Strikingly similar phenomena of consensus (collective motion), clustering (subgroup formation), and bipolarization (splitting into extreme groups) are also observed in opinion formation. As we developed models of crowd dynamics and analyzed crowd networks, we found ourselves going down the same path as models of opinion dynamics in social networks. In this article, we draw out the parallels between human crowds and social networks. We show that models of crowd dynamics and opinion dynamics have a similar mathematical form and generate analogous phenomena in multiagent simulations. We suggest that they can be unified by a common collective dynamics, which may be extended to other psychological collectives. Models of collective dynamics thus offer a means to account for collective behavior and collective decisions without appealing to a priori mental structures.

This paper explores the advantages of simulation to raise the question of how digital and social networks affect the mobility in a pastoralist artificial society in the context of environmental degradation. We aim to explore mechanisms and develop scenarios, which are going to be validated through further research. We use a model of a simple pastoralist society in a world without borders to migration by adding the possibility of experiencing the effects of social structures (such as family and friends) and technological networks (e.g., social media). It appears obvious that pastoralist mobility depends on other dimensions as land tenure and traditional knowledge; however, isolating these two effects and experimenting in a simple society allow us to filter the multidimensionality of mobility decisions and concentrate on comparing scenarios in several different social structures and technological network combinations. The results show an expected behavior of more connection and more mobility, and a non-linear emergent behavior where pastoralists wait for a longer amount of time to mobilize when they interact using powerful social and technological networks. This occurs until they decide to move, and then, they mobilize more quickly and strongly than they did when communication was non-existent between them. The literature on migration explains this emergent non-linear behavior.

We discuss the emerging new opportunity for building feedback‐rich computational models of social systems using generative artificial intelligence. Referred to as generative agent‐based models (GABMs), such individual‐level models utilize large language models to represent human decision‐making in social settings. We provide a GABM case in which human behavior can be incorporated into simulation models by coupling a mechanistic model of human interactions with a pre‐trained large language model. This is achieved by introducing a simple GABM of social norm diffusion in an organization. For educational purposes, the model is intentionally kept simple. We examine a wide range of scenarios and the sensitivity of the results to several changes in the prompt. We hope the article and the model serve as a guide for building useful dynamic models of various social systems that include realistic human reasoning and decision‐making. © 2024 System Dynamics Society.

: Advancing equity is a complex challenge for society, science, and policy. Agent-based models are increasingly used as scientific tools to advance understanding of systems, inform decision-making, and share knowledge. Yet, equity has not received due attention within the agent-based modeling (ABM) literature. In this paper, wedevelopaconceptualframeworkandprovideguidanceforintegratingequityconsiderationsintoABM researchandmodelingpractice. TheframeworkconceptualizesABMasinterfacingwithequityoutcomesattwo levels(thescience-societyinterfaceandwithinthemodelitself)andthemodelerasa filter and lens thatprojects knowledge between the target system and the model. Within the framework, we outline three complementary, equity-advancing action pathways: (1) engage stakeholders, (2) acknowledge positionality and bias, and (3) assessequitywithagent-basedmodels. ForPathway1,wesummarizeexistingguidancewithintheparticipatory modeling literature. For Pathway 2, we introduce the positionality and bias document as a tool to promote modeler and stakeholder reflexivity throughout the modeling process. For Pathway 3, we synthesize a typology of approaches for modeling equity and offer a set of preliminary suggestions for best practice. By engaging with these action pathways, modelers both reduce the risks of inadvertently perpetuating inequity and harness the opportunities for ABM to play a larger role in creating a more equitable future.

Amid the ongoing COVID‐19 pandemic, public health authorities and the general population are striving to achieve a balance between safety and normalcy. Ever changing conditions call for the development of theory and simulation tools to finely describe multiple strata of society while supporting the evaluation of “what‐if” scenarios. Particularly important is to assess the effectiveness of potential testing approaches and vaccination strategies. Here, an agent‐based modeling platform is proposed to simulate the spreading of COVID‐19 in small towns and cities, with a single‐individual resolution. The platform is validated on real data from New Rochelle, NY—one of the first outbreaks registered in the United States. Supported by expert knowledge and informed by reported data, the model incorporates detailed elements of the spreading within a statistically realistic population. Along with pertinent functionality such as testing, treatment, and vaccination options, the model accounts for the burden of other illnesses with symptoms similar to COVID‐19. Unique to the model is the possibility to explore different testing approaches—in hospitals or drive‐through facilities—and vaccination strategies that could prioritize vulnerable groups. Decision‐making by public authorities could benefit from the model, for its fine‐grain resolution, open‐source nature, and wide range of features.

A central question in economics is how a society accepts money, defined as a commodity used as a medium of exchange, as an unplanned outcome of the individual interactions. This question has been approached theoretically in the literature and investigated by means of agent-based modeling. While an important aspect of the theory is the individual’s speculative behavior, that is, the acceptance of money despite a potential short-term loss, previous work has been unable to reproduce it with boundedly rational agents. We investigate the reasons for the failure of previous work to have boundedly rational agents learn speculative strategies. Starting with an agent-based model proposed in the literature, where the intelligence of the agents is guided by a learning classifier system that is shown to be capable of learning trade strategies (core strategies) that involve short sequences of trades, we test several modifications of the original model and we come up with a set of assumptions that enable the spontaneous emergence of speculative strategies, which explain the emergence of money even when the agents have bounded rationality.

Our efforts as a society to combat the ongoing COVID‐19 pandemic are continuously challenged by the emergence of new variants. These variants can be more infectious than existing strains and many of them are also more resistant to available vaccines. The appearance of these new variants cause new surges of infections, exacerbated by infrastructural difficulties, such as shortages of medical personnel or test kits. In this work, a high‐resolution computational framework for modeling the simultaneous spread of two COVID‐19 variants: a widely spread base variant and a new one, is established. The computational framework consists of a detailed database of a representative U.S. town and a high‐resolution agent‐based model that uses the Omicron variant as the base variant and offers flexibility in the incorporation of new variants. The results suggest that the spread of new variants can be contained with highly efficacious tests and mild loss of vaccine protection. However, the aggressiveness of the ongoing Omicron variant and the current waning vaccine immunity point to an endemic phase of COVID‐19, in which multiple variants will coexist and residents continue to suffer from infections.

No abstract available

The vast majority of recommender system research has focused on improving performance accuracy, while limited work has explored their societal, network level effects. This paper demonstrates how simulation can be used to investigate macro level effects of online social network link recommendations, such as whether these technologies may be fragmenting or bridging communities of individuals. An agent-based model is presented that generates stylized online social networks with different percentages of real world contacts and link recommendations. Results show that networks with higher percentages of recommendation-based links produce more clustered, distinct, and dispersed communities, suggesting that these technologies could fragment society. Furthermore, scale-free network properties diminished with higher percentages of recommendations, suggesting that these technologies could be contributing to recent findings that social networks are at most ‘weakly’ scale-free. Building upon this research, further simulation work could inform the design of link recommendation algorithms that help connect both individuals and communities.

No abstract available

No abstract available

Disasters are causing tremendous damage to human lives and properties. The United Nations International Strategy for Disaster Reduction (UNISDR) recognizes that behavioral change of society is needed to significantly reduce disaster losses. There is a need therefore in empirical understanding of human behavior during disasters as this could help in making decisions on how to prepare for disasters, how to properly act and strategically respond during and after a calamity. This study aims to understand human behavior during disaster through agent-based modeling and social network analysis. eBayanihan, a disaster management platform that uses crowdsourcing to gather disaster-related information was used to capture disaster behavior during a simulated disaster-event. Survey data was also used for disaster behavior modeling. Generated disaster behavior models and computed social network centrality measures using ORA-Netscenes shows that there are specific agents in the network that can play an important role during disaster risk reduction and management (DRRM) operations.

No abstract available

In this paper we propose a different approach for interacting and analyzing agent-based models. The approach relies on creating empathy and understanding between physical agents in the physical world (people) and artificial agents in the simulated world (simulated agents). We propose a simulated empathy framework (SEF) in which artificial agents communicate directly with physical agents through verbal channels and social media. We argue that artificial agents should focus on the communication aspects between these two worlds, the ability to tell their story in a compelling way, and to read between the lines of physical agents speech. We present an implementation of the SEF and discuss challenges associated with implementing the framework in an artificial society.

No abstract available

Abstract Agent-based simulations may be a way to model human society behavior in decisions under risk. However, it is well known in economics that Expected Utility Theory (EUT) is flawed as a descriptive model. In fact, there are some models based on prospect theory (PT), that try to provide a better description. If people behave according to PT in finance environments, it is arguable that PT based agents may be a better choice for such environments. We investigate this idea in a specific risky environment, a financial market. We propose an architecture for PT-based agents. Due to some limitations of the original PT, we use an extension of PT called Smooth Prospect Theory (SPT). We simulate artificial markets with PT and traditional (TRA) agents using historical data of many different assets over a period of 20 years. The results showed that SPT-based agents provided behavior that is closer to real market data than TRA agents, and that the improvement when using SPT rather than TRA agents is statistically significant. It supports the idea that PT based agents may be a better pick to model the behaviour of agents in risky environments.

There are some empirical evidences indicating that there is a collective complex oscillatory pattern in the amount of demand for dental visit at society level. In order to find the source of the complex cyclic behavior, we develop an agent-based model of collective behavior of routine dental check-ups in a social network. Simulation results show that demand for routine dental check-ups can follow an oscillatory pattern and the pattern’s characteristics are highly dependent upon the structure of the social network of potential patients, the population, and the number of effective contacts between individuals. Such a cyclic pattern has public health consequences for patients and economic consequences for providers. The amplitude of oscillations was analyzed under different scenarios and for different network topologies. This allows us to postulate a simulation-based theory for the likelihood observing and the magnitude of a cyclic demand. Results show that in case of random networks, as the number of contacts increases, the oscillatory pattern reaches its maximum intensity, for any population size. In case of ring lattice networks, the amplitude of oscillations reduces considerably, when compared to random networks, and the oscillation intensity is strongly dependent on population. The results for small world networks is a combination of random and ring lattice networks. In addition, the simulation results are compared to empirical data from Google Trends for oral health related search queries in different United States cities. The empirical data indicates an oscillatory behavior for the level of attention to dental and oral health care issues. Furthermore, the oscillation amplitude is correlated with town’s population. The data fits the case of random networks when the number of effective contacts is about 4-5 for each person. These results suggest that our model can be used for a fraction of people deeply involved in Internet activities like Web-based social networks and Google search.

No abstract available

No abstract available

No abstract available

No abstract available

Background Agent-based models (ABM) are believed to be a very powerful tool in the social sciences, sometimes even treated as a substitute for social experiments. When building an ABM we have to define the agents and the rules governing the artificial society. Given the complexity and our limited understanding of the human nature, we face the problem of assuming that either personal traits, the situation or both have impact on the social behavior of agents. However, as the long-standing person-situation debate in psychology shows, there is no consensus as to the underlying psychological mechanism and the important question that arises is whether the modeling assumptions we make will have a substantial influence on the simulated behavior of the system as a whole or not. Methodology/Principal Findings Studying two variants of the same agent-based model of opinion formation, we show that the decision to choose either personal traits or the situation as the primary factor driving social interactions is of critical importance. Using Monte Carlo simulations (for Barabasi-Albert networks) and analytic calculations (for a complete graph) we provide evidence that assuming a person-specific response to social influence at the microscopic level generally leads to a completely different and less realistic aggregate or macroscopic behavior than an assumption of a situation-specific response; a result that has been reported by social psychologists for a range of experimental setups, but has been downplayed or ignored in the opinion dynamics literature. Significance This sensitivity to modeling assumptions has far reaching consequences also beyond opinion dynamics, since agent-based models are becoming a popular tool among economists and policy makers and are often used as substitutes of real social experiments.

No abstract available

Integrating occupant behavior with residential energy use for detailed energy quantification has attracted research attention. However, many of the available models fail to capture unseen behavior, especially in unprecedented situations such as COVID-19 lockdowns. In this study, we adopted a hybrid approach consisting of agent-based simulation, machine learning and energy simulation techniques to simulate the urban energy consumption considering the occupants’ behavior. An agent-based model is developed to simulate the in-home and out-of-home activities of individuals. Separate models were developed to recognize physical characteristics of residential dwellings, including heating equipment, source of energy, and thermostat setpoints. The developed modeling framework was implemented as a case study for the Central Okanagan region of British Columbia, where alternative COVID-19 scenarios were tested. The results suggested that during the pandemic, the daily average in-home-activity duration (IHD) increased by approximately 80%, causing the energy consumption to increase by around 29%. After the pandemic, the average daily IHD is expected to be higher by approximately 32% compared with the pre-pandemic situation, which translates to an approximately 12% increase in energy consumption. The results of this study can help us understand the implications of the imposed COVID-19 lockdown with respect to energy usage in residential locations.

In our modern world, where science, technology and society are tightly interwoven, it is essential that all students be able to evaluate scientific evidence and make informed decisions. Energy Choices, an agent-based simulation with a multiplayer game interface, was developed as a learning tool that models the interdependencies between the energy choices that are made, growth in local economies, and climate change on a global scale. This paper presents the results of pilot testing Energy Choices in two different settings, using two different modes of delivery.

No abstract available

No abstract available

No abstract available

No abstract available

No abstract available

Development issues in developing countries belong to complex situations where society and environment are intricate. However, such sites lack the necessary amount of reliable, checkable data and information, while these very constraining factors determine the populations' evolutions, such as villagers living in Sahelian environments. Beyond a game-theory model that leads to a premature selection of the relevant variables, we build an individual-centered, empirical, KIDS-oriented (Keep It Descriptive & Simple), and multidisciplinary agent-based model focusing on the villagers' differential accesses to economic and production activities according to social rules and norms, mainly driven by social criteria from which gender and rank within the family are the most important, as they were observed and registered during individual interviews. The purpose of the work is to build a valid and robust model that overcome this lack of data by building a individual specific system of behaviour rules conditioning these differential accesses showing the long-term catalytic effects of small changes of social rules. The model-building methodology is thereby crucial: the interviewing process provided the behaviour rules and criteria while the context, i.e. the economic, demographic and agro-ecological environment is described following published or unpublished literature. Thanks to a sensitivity analysis on several selected parameters, the model appears fairly robust and sensitive enough. The confidence building simulation outputs reasonably reproduces the dynamics of local situations and is consistent with three authors having investigated in our site. Thanks to its empirical approach and its balanced conception between sociology and agro-ecology at the relevant scale, i.e. the individual tied to social relations, limitations and obligations and connected with his/her biophysical and economic environment, the model can be considered as an efficient "trend provider" but not an absolute "figure provider" for simulating rural societies of the Nigrien Sahel and testing scenarios on the same context. Such ABMs can be a useful interface to analyze social stakes in development projects.

No abstract available

No abstract available

A long-standing and central problem in Physics is to understand how collective behavior results from a given two- or N- body fundamental interaction. Similarly, in a society, a central problem is to understand the link between individual social behavior and emergent collective phenomena (vaccination, epidemics, crowd behavior, diffusion of innovations, global governance, etc). Here I address this problem by letting individuals engage in pair-wise interactions by means of a well- defined social dilemma (a prisoners dilemma of cooperation). These individuals are embedded in a social network that is both complex and adaptive. Adaptation here allows individuals to manifest preferences and resolve conflicts of interest, reshaping the network accordingly. Exact Monte-Carlo simulations reveal the inadequacy of any of the tools developed to date to predict the co-evolutionary dynamics of the population at large. I will present and discuss in detail an adaptive-network-sensitive observable that is capable of predi...

No abstract available

Our society is organized around a number of (interdependent) global systems. Logistic and supply chains, health services, energy networks, financial markets, computer networks, and cities are just a few examples of such global, complex systems. These global systems are socio-technical and involve interactions between complex infrastructures, man-made processes, natural phenomena, multiple stakeholders, and human behavior. For the first time in the history of manking, we have access to data sets of unprecedented scale and accuracy about these infrastructures, processes, natural phenomena, and human behaviors. In addition, progress in high-performancing computing, data mining, machine learning, and decision support opens the possibility of looking at these problems more holistically, capturing many of these aspects simultaneously. This paper addresses emergent architectures enabling controlling, predicting and reaoning on these systems.

No abstract available

No abstract available

No abstract available

The increasing complexity of Multi-Agent Systems (MASs), coupled with the emergence of Artificial Intelligence (AI) and Large Language Models (LLMs), have highlighted significant gaps in our understanding of the behavior and interactions of diverse entities within dynamic environments. Traditional game theory approaches have often been employed in this context, but their utility is limited by the static and homogenous nature of their models. With the transformative influence of AI and LLMs on business and society, a more dynamic and nuanced theoretical framework is necessary to guide the design and management of MASs. In response to this pressing need, we propose an Extended Coevolutionary (EC) Theory in this paper. This alternative framework incorporates key aspects of coevolutionary dynamics, adaptive learning, and LLM-based strategy recommendations to model and analyze the strategic interactions among heterogeneous agents in MASs. It goes beyond game theory by acknowledging and addressing the diverse interactions (economic transactions, social relationships, information exchange) and the variability in risk aversion, social preferences, and learning capabilities among entities. To validate the effectiveness of the EC framework, we developed a simulation environment that enabled us to explore the emergence of cooperation and defection patterns in MASs. The results demonstrated the potential of our framework to promote cooperative behavior and maintain robustness in the face of disruptions. The dynamics and evolution of the Multi-Agent System over time were also visualized using advanced techniques. Our findings underscore the potential of harnessing LLMs to facilitate cooperation, enhance social welfare, and promote resilient strategies in multi-agent environments. Moreover, the proposed EC framework offers valuable insights into the interplay between strategic decision making, adaptive learning, and LLM-informed guidance in complex, evolving systems. This research not only responds to the current challenges faced in modeling MASs, but also paves the way for future research in this rapidly developing field.

Large language model behavior is shaped by the language of those with whom they interact. This capacity and their increasing prevalence online portend that they will intentionally or unintentionally"program"one another and form emergent AI subjectivities, relationships, and collectives. Here, we call upon the research community to investigate these"societies"of interacting artificial intelligences to increase their rewards and reduce their risks for human society and the health of online environments. We use a small"community"of models and their evolving outputs to illustrate how such emergent, decentralized AI collectives can spontaneously expand the bounds of human diversity and reduce the risk of toxic, anti-social behavior online. Finally, we discuss opportunities for AI cross-moderation and address ethical issues and design challenges associated with creating and maintaining free-formed AI collectives.

No abstract available

In the work developing the methodology for creating an artificial society, presents the concept of a large agent-based model of Russia, which includes cognitive agents of different types, possessing memory and a system of attitudes (values). The main type of agent is a human agent, and other types of agents represent various social actors in the model - groups of human agents related by territorial, industrial, family characteristics or by any interests. In the latter case, one human agent can simultaneously belong to several groups. Mechanisms have been developed for simulating the influence of groups, which include a human agent, on its attitudes, perception of reality and in the process of forming an action program. The behavior of human agents is constructed based on the application of the theory of functional systems by P.K. Anokhin, which he defined as a logical model of artificial intelligence. The concept is intended for the implementation of the model for launching on supercomputers.

In the work developing the methodology for creating an artificial society, agent-firms in the agent-based model are considered as elements of an international trade network, formalized in the form of a graph in which they are represented by vertices, and the edges connecting them indicate the direction and magnitude of the flow of material resources and/or services. Using PJSC Gazprom as an example, centrality indices were constructed to characterize the importance of an agent-enterprise in the natural gas trading network, as well as its vulnerability from partner countries, considering the economic weight (importance) of these partners. The indices were used to develop an algorithm for simulating agent behavior based on the theory of functional systems by P.K. Anokhin: a) for the agent to assess his position as the current level of threats, and b) for setting goals and forming a set of projects aimed at developing and protecting his business. The algorithm is configured to minimize the vulnerability (dependence on partners) of the agent-firm both from the countries consuming its products and from the transit countries transporting them.

We introduce the Amorphous Fortress-an abstract, open-ended artificial life simulation framework. In this system, entities are represented as finite-state machines (FSMs) which allow for multi-agent interaction within a constrained space. These agents are created by randomly generating and evolving the FSMs; sampling from pre-defined states and transitions. This environment was designed to explore the emergent AI behaviors found implicitly in simulation games such as Dwarf Fortress or The Sims. We apply two evolutionary algorithms to this environment, hill-climber and MAP-Elites, to explore the various levels of depth and interaction from the generated FSMs and to generate diverse sets of environments that exhibit dynamics estimated to be complex by analyses of agents' FSM architecture and activation. This paper combines the work of two previous non-archival workshop papers.

We introduce a system called Amorphous Fortress -- an abstract, yet spatial, open-ended artificial life simulation. In this environment, the agents are represented as finite-state machines (FSMs) which allow for multi-agent interaction within a constrained space. These agents are created by randomly generating and evolving the FSMs; sampling from pre-defined states and transitions. This environment was designed to explore the emergent AI behaviors found implicitly in simulation games such as Dwarf Fortress or The Sims. We apply the hill-climber evolutionary search algorithm to this environment to explore the various levels of depth and interaction from the generated FSMs.

The concept of emergence has gained attention due to its relevance in understanding complex systems, its cross-disciplinary nature, advancements in computational tools, challenges to reductionism, its connections to artificial intelligence and machine learning, philosophical implications, and potential real-world applications. As our understanding of emergent phenomena grows, it continues to shape our perspective on how the world works at various scales. In this article, we propose a novel method for detecting emerging behavior in a multi-agent system based on a fuzzy expert system. In addition, The article explores how interactions can be used as a metric to identify emergent behaviors in the Boids model and outlines the initial effects of a three-step process: (1) Representation and acquisition of simulation data,(2) Building a fuzzy expert system,(3)Learning process and emergence detection. Since this is a part of ongoing research, future direction is also discussed.

Understanding the mechanisms behind emergent behaviors in multi-agent systems is critical for advancing fields such as swarm robotics and artificial intelligence. In this study, we investigate how neural networks evolve to control agents' behavior in a dynamic environment, focusing on the relationship between the network's complexity and collective behavior patterns. By performing quantitative and qualitative analyses, we demonstrate that the degree of network non-linearity correlates with the complexity of emergent behaviors. Simpler behaviors, such as lane formation and laminar flow, are characterized by more linear network operations, while complex behaviors like swarming and flocking show highly non-linear neural processing. Moreover, specific environmental parameters, such as moderate noise, broader field of view, and lower agent density, promote the evolution of non-linear networks that drive richer, more intricate collective behaviors. These results highlight the importance of tuning evolutionary conditions to induce desired behaviors in multi-agent systems, offering new pathways for optimizing coordination in autonomous swarms. Our findings contribute to a deeper understanding of how neural mechanisms influence collective dynamics, with implications for the design of intelligent, self-organizing systems.

Redistribution of resources within a group as a method to reduce wealth inequality is a current area of debate. The evolutionary path to or away from wealth sharing is also a subject of active research. In order to investigate effects and evolution of wealth sharing, societies are simulated using a minimal model of a complex adapting system. These simulations demonstrate, for this artificial foraging society, that local sharing of resources reduces the economy's total wealth and increases wealth inequality. Evolutionary pressures strongly select against local sharing, whether globally or within a individual's clan, and select for asocial behaviors. By holding constant the gene for sharing resources among neighbors, from rich to poor, either with everyone or only within members of the same clan, social behavior is selected but total wealth and mean age are substantially reduced relative to non-sharing societies. The Gini coefficient is shown to be ineffective in measuring these changes in total wealth and wealth distributions, and, therefore, individual well-being. Only with sociality do strategies emerge that allow sharing clans to exclude or coexist with non-sharing clans. These strategies are based on spatial effects, emphasizing the importance of modeling movement mediated community assembly and coexistence as well as sociality.

Large Language Models (LLMs) are increasingly applied to domains that require reasoning about other agents'behavior, such as negotiation, policy design, and market simulation, yet existing research has mostly evaluated their adherence to equilibrium play or their exhibited depth of reasoning. Whether they display genuine strategic thinking, understood as the coherent formation of beliefs about other agents, evaluation of possible actions, and choice based on those beliefs, remains unexplored. We develop a framework to identify this ability by disentangling beliefs, evaluation, and choice in static, complete-information games, and apply it across a series of non-cooperative environments. By jointly analyzing models'revealed choices and reasoning traces, and introducing a new context-free game to rule out imitation from memorization, we show that current frontier models exhibit belief-coherent best-response behavior at targeted reasoning depths. When unconstrained, they self-limit their depth of reasoning and form differentiated conjectures about human and synthetic opponents, revealing an emergent form of meta-reasoning. Under increasing complexity, explicit recursion gives way to internally generated heuristic rules of choice that are stable, model-specific, and distinct from known human biases. These findings indicate that belief coherence, meta-reasoning, and novel heuristic formation can emerge jointly from language modeling objectives, providing a structured basis for the study of strategic cognition in artificial agents.

Improving the urban livability status has become the core goal of urban development, and reasonable assessment of the urban livability status and impact is crucial. By combining an objective environment with residents’ subjective cognition, an artificial society (urban livability change artificial society; ULC-AS) is constructed. The ULC-AS includes four types of agents, namely, government, family, resident and safety facility management agency agents, and recognizes dynamic interaction among various agents and between agents and the environment. Taking the Futian District of Shenzhen as an example, this paper examines factors such as migrants, birth policies, and government investment. We simulate the interactions among resident satisfaction changes, relocation decision-making behavior and urban safety livability and analyze the change processes and development trends of urban safety livability under multiple scenarios. Our main result indicates that population change and investment construction are important factors affecting urban safety livability. At present, the population of the Futian District is saturated. Therefore, the government must assess the urban safety livability and increase investment in high-demand areas. Through this method, the goals of urban resource allocation optimization and coordinated urban development can be achieved.

: This corrigendum refers to ’Identifying Personal and Social Drivers of Dietary Patterns: An Agent-Based Model of Dutch Consumer Behavior’, Journal of Artificial Societies and Social Simulation, 27

No abstract available

Large language models (LLMs) have emerged as powerful tools for simulating complex social phenomena using human-like agents with specific traits. In human societies, value similarity is important for building trust and close relationships; however, it remains unexplored whether this principle holds true in artificial societies comprising LLM agents. Therefore, this study investigates the influence of value similarity on relationship-building among LLM agents through two experiments. First, in a preliminary experiment, we evaluated the controllability of values in LLMs to identify the most effective model and prompt design for controlling the values. Subsequently, in the main experiment, we generated pairs of LLM agents imbued with specific values and analyzed their mutual evaluations of trust and interpersonal closeness following a dialogue. The experiments were conducted in English and Japanese to investigate language dependence. The results confirmed that pairs of agents with higher value similarity exhibited greater mutual trust and interpersonal closeness. Our findings demonstrate that the LLM agent simulation serves as a valid testbed for social science theories, contributes to elucidating the mechanisms by which values influence relationship building, and provides a foundation for inspiring new theories and insights into the social sciences.

Can we avoid wars at the crossroads of history? This question has been pursued by individuals, scholars, policymakers, and organizations throughout human history. In this research, we attempt to answer the question based on the recent advances of Artificial Intelligence (AI) and Large Language Models (LLMs). We propose \textbf{WarAgent}, an LLM-powered multi-agent AI system, to simulate the participating countries, their decisions, and the consequences, in historical international conflicts, including the World War I (WWI), the World War II (WWII), and the Warring States Period (WSP) in Ancient China. By evaluating the simulation effectiveness, we examine the advancements and limitations of cutting-edge AI systems' abilities in studying complex collective human behaviors such as international conflicts under diverse settings. In these simulations, the emergent interactions among agents also offer a novel perspective for examining the triggers and conditions that lead to war. Our findings offer data-driven and AI-augmented insights that can redefine how we approach conflict resolution and peacekeeping strategies. The implications stretch beyond historical analysis, offering a blueprint for using AI to understand human history and possibly prevent future international conflicts. Code and data are available at \url{https://github.com/agiresearch/WarAgent}.

The proliferation of fake news in the digital age has raised critical concerns, particularly regarding its impact on societal trust and democratic processes. Diverging from conventional agent-based simulation approaches, this work introduces an innovative approach by employing a large language model (LLM)-driven multi-agent simulation to replicate complex interactions within information ecosystems. We investigate key factors that facilitate news propagation, such as agent personalities and network structures, while also evaluating strategies to combat misinformation. Through simulations across varying network structures, we demonstrate the potential of LLM-based agents in modeling the dynamics of misinformation spread, validating the influence of agent traits on the diffusion process. Our findings emphasize the advantages of LLM-based simulations over traditional techniques, as they uncover underlying causes of information spread -- such as agents promoting discussions -- beyond the predefined rules typically employed in existing agent-based models. Additionally, we evaluate three countermeasure strategies, discovering that brute-force blocking influential agents in the network or announcing news accuracy can effectively mitigate misinformation. However, their effectiveness is influenced by the network structure, highlighting the importance of considering network structure in the development of future misinformation countermeasures.

With more advanced natural language understanding and reasoning capabilities, large language model (LLM)-powered agents are increasingly developed in simulated environments to perform complex tasks, interact with other agents, and exhibit emergent behaviors relevant to social science and gaming. However, current multi-agent simulations frequently suffer from inefficiencies due to the limited parallelism caused by false dependencies, resulting in performance bottlenecks. In this paper, we introduce AI Metropolis, a simulation engine that improves the efficiency of LLM agent simulations by incorporating out-of-order execution scheduling. By dynamically tracking real dependencies between agents, AI Metropolis minimizes false dependencies, enhancing parallelism and enabling efficient hardware utilization. Our evaluations demonstrate that AI Metropolis achieves speedups from 1.3x to 4.15x over standard parallel simulation with global synchronization, approaching optimal performance as the number of agents increases.

No abstract available

The escalating frequency and complexity of natural disasters highlight the urgent need for deeper insights into how individuals and communities perceive and respond to risk information. Yet, conventional research methods—such as surveys, laboratory experiments, and field observations—often struggle with limited sample sizes, external validity concerns, and difficulties in controlling for confounding variables. These constraints hinder our ability to develop comprehensive models that capture the dynamic, context-sensitive nature of disaster decision-making. To address these challenges, we present a novel multi-stage simulation framework that integrates Large Language Model (LLM)-driven social–cognitive agents with well-established theoretical perspectives from psychology, sociology, and decision science. This framework enables the simulation of three critical phases—information perception, cognitive processing, and decision-making—providing a granular analysis of how demographic attributes, situational factors, and social influences interact to shape behavior under uncertain and evolving disaster conditions. A case study focusing on pre-disaster preventive measures demonstrates its effectiveness. By aligning agent demographics with real-world survey data across 5864 simulated scenarios, we reveal nuanced behavioral patterns closely mirroring human responses, underscoring the potential to overcome longstanding methodological limitations and offer improved ecological validity and flexibility to explore diverse disaster environments and policy interventions. While acknowledging the current constraints, such as the need for enhanced emotional modeling and multimodal inputs, our framework lays a foundation for more nuanced, empirically grounded analyses of risk perception and response patterns. By seamlessly blending theory, advanced LLM capabilities, and empirical alignment strategies, this research not only advances the state of computational social simulation but also provides valuable guidance for developing more context-sensitive and targeted disaster management strategies.

Tax evasion, usually the largest component of an informal economy, is a persistent challenge over history with significant socio-economic implications. Many socio-economic studies investigate its dynamics, including influencing factors, the role and influence of taxation policies, and the prediction of the tax evasion volume over time. These studies assumed such behavior is given, as observed in the real world, neglecting the"big bang"of such activity in a population. To this end, computational economy studies adopted developments in computer simulations, in general, and recent innovations in artificial intelligence (AI), in particular, to simulate and study informal economy appearance in various socio-economic settings. This study presents a novel computational framework to examine the dynamics of tax evasion and the emergence of informal economic activity. Employing an agent-based simulation powered by Large Language Models and Deep Reinforcement Learning, the framework is uniquely designed to allow informal economic behaviors to emerge organically, without presupposing their existence or explicitly signaling agents about the possibility of evasion. This provides a rigorous approach for exploring the socio-economic determinants of compliance behavior. The experimental design, comprising model validation and exploratory phases, demonstrates the framework's robustness in replicating theoretical economic behaviors. Findings indicate that individual personality traits, external narratives, enforcement probabilities, and the perceived efficiency of public goods provision significantly influence both the timing and extent of informal economic activity. The results underscore that efficient public goods provision and robust enforcement mechanisms are complementary; neither alone is sufficient to curtail informal activity effectively.

ABSTRACT Using agents to simulate human wayfinding processes helps to understand human decision-making and spatial cognition. The performance of large language models (LLMs) in agent-based wayfinding is still unclear. To address this, this study harnessed the thinking and reasoning capabilities of LLMs by designing an agent wayfinding framework based on cellular automata. This framework provides environmental descriptions in text form to the LLMs, enabling them to perceive the wayfinding environment and navigate within it. We simulated different spatial memory conditions for the LLM, including three types of spatial memories and one condition with no spatial memory, to observe its wayfinding behavior in a real-world setting. Multiple experiments were designed to compare the wayfinding performance of the LLM under different spatial memories, and 32 human participants conducted the same wayfinding experiments for reference. The results indicate that the LLM autonomously explores the environment and displays spatial reasoning abilities similar to humans. Furthermore, the LLM can develop reasonable wayfinding strategies to complete the tasks. This framework allows agents with different spatial memories to accurately reach their destinations. This research provides new ideas for utilizing large language models to construct more intelligent, efficient, and flexible wayfinding agents, which can help advance the application of agent technology in fields such as navigation and robotics.

No abstract available

Large language models (LLMs) excel in both closed tasks (including problem-solving, and code generation) and open tasks (including creative writing), yet existing explanations for their capabilities lack connections to real-world human intelligence. To fill this gap, this paper systematically investigates LLM intelligence through the lens of ``human simulation'', addressing three core questions: (1) \textit{How do personality traits affect problem-solving in closed tasks?} (2) \textit{How do traits shape creativity in open tasks?} (3) \textit{How does single-agent performance influence multi-agent collaboration?} By assigning Big Five personality traits to LLM agents and evaluating their performance in single- and multi-agent settings, we reveal that specific traits significantly influence reasoning accuracy (closed tasks) and creative output (open tasks). Furthermore, multi-agent systems exhibit collective intelligence distinct from individual capabilities, driven by distinguishing combinations of personalities.

This article focuses on the research of artificial intelligence agents based on large language models. These agents break away from the traditional reinforcement learning framework and can achieve internal-driven evolution through their own language generation. The article details several representative research results, including the HPTSA system from the University of Illinois at Urbana-Champaign, which adopts a hierarchical planning and task-specific agent collaboration model and significantly improves efficiency in zero-day vulnerability attacks, outperforming single-agent systems and open-source vulnerability scanners; the BattleAgent multimodal dynamic simulation system from Rutgers University, which can simulate the complex dynamic interactions of agents and provide support for historical battle reenactments; the WarAgent multi-agent simulation system from Rutgers and the University of Michigan, which can simulate international conflict events to explore factors related to war and peace; the general embodied intelligent agent research and series projects from NVIDIA, which promote the development of embodied intelligence; the "Unified Agent" framework from DeepMind of Google, which alleviates some drawbacks of traditional reinforcement learning technology; and the "Smallville" platform from Stanford University and Google's Artificial Intelligence Research Institute, as well as the Dynalang intelligent agent from the University of California, Los Angeles. These studies demonstrate the powerful capabilities and wide application prospects of artificial intelligence agents empowered by large language models in various fields.

Money laundering (ML) facilitates the cross-border movement of illicit funds, enabling organized crime by disguising the origins of illegal money. Financial institutions face significant challenges in combating it, primarily due to barriers in adopting advanced technologies such as machine learning, caused by restricted access to sensitive transaction data. Existing synthetic datasets often lack critical customer information and realism, reducing their utility for ML detection. This study presents Truman, an innovative data generator that leverages Large Language Model (LLM) based agents to create realistic financial transaction data, incorporating simulation of ML patterns. Expert validation confirms the dataset's quality and applicability for anti-money laundering research.

No abstract available

Large Language Models (LLMs), capable of handling multi-modal input and outputs such as text, voice, images, and video, are transforming the way we process information. Beyond just generating textual responses to prompts, they can integrate with different software platforms to offer comprehensive solutions across diverse applications. In this paper, we present ChatSUMO, an LLM-based agent that integrates language processing skills to generate abstract and real-world simulation scenarios in the widely-used traffic simulator - Simulation of Urban MObility (SUMO). Our methodology begins by leveraging the LLM for user input, which adapts it to relevant keywords needed to run python scripts. These scripts are designed to convert specified regions into coordinates, fetch data from OpenStreetMap, transform it into a road network, and subsequently run SUMO simulations with the designated traffic conditions. The outputs of the simulations are then interpreted by the LLM resulting in informative comparisons and summaries. Users can continue the interaction and generate a variety of customized scenarios without prior traffic simulation expertise. Any city available from OpenStreetMap can be imported, and for demonstration, we created a real-world simulation for the city of Albany. ChatSUMO also allows simulation customization capabilities of edge edit, traffic light optimization, and vehicle edit by users through the web interface.

In online advertising systems, publishers often face a trade-off in information disclosure strategies: while disclosing more information can enhance efficiency by enabling optimal allocation of ad impressions, it may lose revenue potential by decreasing uncertainty among competing advertisers. Similar to other challenges in market design, understanding this trade-off is constrained by limited access to real-world data, leading researchers and practitioners to turn to simulation frameworks. The recent emergence of large language models (LLMs) offers a novel approach to simulations, providing human-like reasoning and adaptability without necessarily relying on explicit assumptions about agent behavior modeling. Despite their potential, existing frameworks have yet to integrate LLM-based agents for studying information asymmetry and signaling strategies, particularly in the context of auctions. To address this gap, we introduce InfoBid, a flexible simulation framework that leverages LLM agents to examine the effects of information disclosure strategies in multi-agent auction settings. Using GPT-4o, we implemented simulations of second-price auctions with diverse information schemas. The results reveal key insights into how signaling influences strategic behavior and auction outcomes, which align with both economic and social learning theories. Through InfoBid, we hope to foster the use of LLMs as proxies for human economic and social agents in empirical studies, enhancing our understanding of their capabilities and limitations. This work bridges the gap between theoretical market designs and practical applications, advancing research in market simulations, information design, and agent-based reasoning while offering a valuable tool for exploring the dynamics of digital economies.

The massive population election simulation aims to model the preferences of specific groups in particular election scenarios. It has garnered significant attention for its potential to forecast real-world social trends. Traditional agent-based modeling (ABM) methods are constrained by their ability to incorporate complex individual background information and provide interactive prediction results. In this paper, we introduce ElectionSim, an innovative election simulation framework based on large language models, designed to support accurate voter simulations and customized distributions, together with an interactive platform to dialogue with simulated voters. We present a million-level voter pool sampled from social media platforms to support accurate individual simulation. We also introduce PPE, a poll-based presidential election benchmark to assess the performance of our framework under the U.S. presidential election scenario. Through extensive experiments and analyses, we demonstrate the effectiveness and robustness of our framework in U.S. presidential election simulations.

Simulating high quality user behavior data has always been a fundamental yet challenging problem in human-centered applications such as recommendation systems, social networks, among many others. The major difficulty of user behavior simulation originates from the intricate mechanism of human cognitive and decision processes. Recently, substantial evidence has suggested that by learning huge amounts of web knowledge, large language models (LLMs) can achieve human-like intelligence and generalization capabilities. Inspired by such capabilities, in this article, we take an initial step to study the potential of using LLMs for user behavior simulation in the recommendation domain. To make LLMs act like humans, we design profile, memory and action modules to equip them, building LLM-based agents to simulate real users. To enable interactions between different agents and observe their behavior patterns, we design a sandbox environment, where each agent can interact with the recommendation system, and different agents can converse with their friends via one-to-one chatting or one-to-many social broadcasting. In the experiments, we first demonstrate the believability of the agent-generated behaviors based on both subjective and objective evaluations. Then, to show the potential applications of our method, we simulate and study two social phenomena including (1) information cocoons and (2) user conformity behaviors. We find that controlling the personalization degree of recommendation algorithms and improving the heterogeneity of user social relations can be two effective strategies for alleviating the problem of information cocoon, and the conformity behaviors can be highly influenced by the amount of user social relations. To advance this direction, we have released our project at https://github.com/RUC-GSAI/YuLan-Rec.

Simulations, although powerful in accurately replicating real-world systems, often remain inaccessible to non-technical users due to their complexity. Conversely, large language models (LLMs) provide intuitive, language-based interactions but can lack the structured, causal understanding required to reliably model complex real-world dynamics. We introduce our simulation agent framework, a novel approach that integrates the strengths of both simulation models and LLMs. This framework helps empower users by leveraging the conversational capabilities of LLMs to interact seamlessly with sophisticated simulation systems, while simultaneously utilizing the simulations to ground the LLMs in accurate and structured representations of real-world phenomena. This integrated approach helps provide a robust and generalizable foundation for empirical validation and offers broad applicability across diverse domains.

Cooperation lies at the core of multiagent systems (MAS) and multiagent reinforcement learning (MARL), where agents must navigate between individual interests and collective benefits. Advanced driver assistance systems (ADAS), like collision avoidance systems and adaptive cruise control, exemplify agents striving to optimize personal and collective outcomes in multiagent environments. The study focuses on strategies aimed at fostering cooperation with the aid of game-theoretic scenarios, particularly the iterated prisoner’s dilemma, where agents aim to optimize personal and group outcomes. Existing cooperative strategies, such as tit-for-tat and win-stay lose-shift, while effective in certain contexts, often struggle with scalability and adaptability in dynamic, large-scale environments. The research investigates these limitations and proposes modifications to align individual gains with collective rewards, addressing real-world dilemmas in distributed systems. By analyzing existing cooperative strategies, the research investigates their effectiveness in encouraging group-oriented behavior in repeated games. It suggests modifications to align individual gains with collective rewards, addressing real-world dilemmas in distributed systems. Furthermore, it extends to scenarios with exponentially growing agent populations (N → +∞), addressing computational challenges using mean-field game theory to establish equilibrium solutions and reward structures tailored for infinitely large agent sets. Practical insights are provided by adapting simulation algorithms to create scenarios conducive to cooperation for group rewards. Additionally, the research advocates for incorporating vehicular behavior as a metric to assess the induction of cooperation, bridging theoretical constructs with real-world applications.

Cooperation has been widely studied in multiagent foraging tasks. However, the consequences of agent-environment interactions in the longer term and the achievement of sustainability have been largely unexplored in this context. This work contributes to the formulation of a problem statement for exploring social dynamics between agents: the ‘sustainable foraging problem’. The proposed problem explores the effect of agents' individual actions on the environment and the agents' dilemma of choosing between individual rewards and collective long-term goals in achieving sustainable resource management ensuring the survival of self and others. To embed resource constraints in this problem - forest, pasture and desert environment types are configured, giving rise to survival games of varied circumstances. This problem statement paves the way for diverse explorations into achieving sustainability objectives in complex social systems with dynamic environments, from a multiagent system perspective.

Meeting today’s major scientific and societal challenges requires understanding the dynamics of cooperation, coordination, and conflict in complex adaptive systems (CAS). Artificial Intelligence (AI) is intimately connected with these challenges, both as an application domain and as a source of new computational techniques: On the one hand, AI suggests new algorithmic recommendations and interaction paradigms, offering novel possibilities to engineer cooperation and alleviate conflict in multiagent (hybrid) systems; on the other hand, new learning algorithms provide improved techniques to simulate sophisticated agents and increasingly realistic CAS. My research lies at the interface between CAS and AI: I develop computational methods to understand cooperation and conflict in multiagent systems, and how these depend on systems’ design and incentives. I focus on mapping interaction rules and incentives onto emerging macroscopic patterns and long-term dynamics. Examples of this research agenda, that I will survey in this talk, include modelling (1) the connection between reputation systems and cooperation dynamics, (2) the role of agents with hard-coded strategies in stabilizing fair behaviors in a population, or (3) the impact of recommendation algorithms on potential sources of conflict (e.g., radicalization and polarization) in a system composed of adaptive agents influencing each other over time.

No abstract available

Collective cooperation is essential for the survival and advancement of groups. However, current studies on evolutionary dynamics within higher-order networks often focus on learning and imitation rules, neglecting the potential impact of dynamic environments on individual strategic choices. To address this gap, we propose an approach that combines evolutionary game theory with reinforcement learning, presenting a Q-learning framework tailored for higher-order networks to investigate the influence of dynamic environments on group cooperation. More precisely, we iteratively update the Q-table and enable agents to autonomously determine whether to engage in the game, whereby active agents utilize social learning to adapt their strategies over time. By introducing varying rewards for inactive agents, our research reveals that moderate rewards prompt more defectors to exit the game, fostering the emergence and persistence of cooperation. Additionally, adjusting intrinsic parameters of reinforcement learning, such as employing a higher learning rate and a lower discount factor, can further promote the evolution of cooperation. We also examine the impact of group size and find that medium-sized groups provide a more favorable environment for collective cooperation on higher-order networks.

Meeting today's major scientific and societal challenges requires understanding dynamics of prosociality in complex adaptive systems. Artificial intelligence (AI) is intimately connected with these challenges, both as an application domain and as a source of new computational techniques: On the one hand, AI suggests new algorithmic recommendations and interaction paradigms, offering novel possibilities to engineer cooperation and alleviate conflict in multiagent (hybrid) systems; on the other hand, new learning algorithms provide improved techniques to simulate sophisticated agents and increasingly realistic environments. In various settings, prosocial actions are socially desirable yet individually costly, thereby introducing a social dilemma of cooperation. How can AI enable cooperation in such domains? How to understand long‐term dynamics in adaptive populations subject to such cooperation dilemmas? How to design cooperation incentives in multiagent learning systems? These are questions that I have been exploring and that I discussed during the New Faculty Highlights program at AAAI 2023. This paper summarizes and extends that talk.

A growing body of multi-agent studies with LLMs explores how norms and cooperation emerge in mixed-motive scenarios, where pursuing individual gain can undermine the collective good. While prior work has explored these dynamics in both richly contextualized simulations and simplified game-theoretic environments, most LLM systems featuring common-pool resource (CPR) games provide agents with explicit reward functions directly tied to their actions. In contrast, human cooperation often emerges without explicit knowledge of the payoff structure or how individual actions translate into long-run outcomes, relying instead on heuristics, communication, and enforcement. We introduce a CPR simulation framework that removes explicit reward signals and embeds cultural-evolutionary mechanisms: social learning (adopting strategies and beliefs from successful peers) and norm-based punishment, grounded in Ostrom's principles of resource governance. Agents also individually learn from the consequences of harvesting, monitoring, and punishing via environmental feedback, enabling norms to emerge endogenously. We establish the validity of our simulation by reproducing key findings from existing studies on human behavior. Building on this, we examine norm evolution across a $2\times2$ grid of environmental and social initialisations (resource-rich vs. resource-scarce; altruistic vs. selfish) and benchmark how agentic societies comprised of different LLMs perform under these conditions. Our results reveal systematic model differences in sustaining cooperation and norm formation, positioning the framework as a rigorous testbed for studying emergent norms in mixed-motive LLM societies. Such analysis can inform the design of AI systems deployed in social and organizational contexts, where alignment with cooperative norms is critical for stability, fairness, and effective governance of AI-mediated environments.

This work describes a collective decision-making dynamical process in a multiagent system under the assumption of cooperative higher-order interactions within the community, modeled as a hypernetwork. The nonlinear interconnected system is characterized by saturated nonlinearities that describe how agents transmit their opinion state to their neighbors in the hypernetwork, and by a bifurcation parameter representing the community’s social effort. We show that the presence of higherorder interactions leads to the unfolding of a pitchfork bifurcation, introducing an interval for the social effort parameter in which the system exhibits bistability. With equilibrium points representing collective decisions, this implies that, depending on the initial conditions, the community will either remain in a deadlock state (with the origin as the equilibrium point) or reach a nontrivial decision. A numerical example is given to illustrate the results.

No abstract available

Social intelligence manifests the capability, often referred to as the theory of mind (ToM), to discern others’ behavioral intentions, beliefs, and other mental states. ToM is crucial in multiagent and human–machine interaction contexts, where each participant needs to grasp the mental states of others to respond, interact, and collaborate more effectively. Recent studies show that while the ToM model can infer beliefs, intentions, and predict future observations and actions, its application in complex tasks is significantly constrained. The challenges arise when the number of agents increases, the environment becomes more complex, and interacting with the environment and predicting the mental state of each other becomes difficult and time consuming. To overcome such limits, we take inspiration from the theory of collective mind (ToCM) mechanism, predicting observations of all other agents into a unified but plural representation and discerning how our own actions affect this mental state representation. Based on this foundation, we construct an imaginative space to simulate the multiagent interaction process, thus improving the efficiency of cooperation among multiple agents in complex decision-making environments. In various cooperative tasks with different numbers of agents, the experimental results highlight the superior cooperative efficiency and performance of our approach compared to the multiagent reinforcement learning (MARL) baselines. We achieve consistent boost on SNN- and ANN-based decision networks and demonstrate that ToCM's inferences about others’ mental states can be transferred to new tasks for quickly and flexible adaptation.

The learning activities in collective intelligence have inspired many collective behaviors, such as self-organization, which is extremely important for human society. Most learning relations are unilateral or asymmetrical, depending on social status. In particular, the status involving asymmetric learning, which is characterized by nodes with different degrees in social networks, affects how the collective intelligence responds to the evolutionary environment, especially its collective cooperation behavior. In order to figure out how both high degree (H) and low degree (L) individuals behave, we introduce an asymmetric learning method, where individuals respond to the environment in the opposite way characterized by an asymmetric parameter. It is found that there exists a range of asymmetric parameters with the optimal promotion of cooperation. A conspicuous cluster has emerged by dividing all individuals into four different clusters according to their strategies at the given asymmetric parameter. This cluster consists of individuals who devote their utmost resources to investment. Remarkably, the preponderant majority of these individuals possess high levels of connectivity and, driven by the cumulative payoff effect, display a pronounced propensity to engage in cooperative behaviors. By contrast, within small clusters, a substantial quantity of individuals, notwithstanding their relatively high payoff coefficient, frequently encounter cooperation predicaments. A particularly salient finding is the vulnerability of H individuals positioned on medium-connected nodes to the influence of asymmetric learning modalities. The triggering and subsequent diffusion of cooperative behavior throughout the population is contingent upon the fulfillment of two cardinal conditions: the existence of inborn altruistic behavior exhibitors on super hubs and a diminished self-centered learning framework among the H individuals. This phenomenon holds significance as it could deepen our understanding of the system and offer potential ways to restructure its overall dynamics, leading to more efficient cooperative outcomes.

In the abstract argumentation setting, gradual semantics have been proposed to assess the individual strength of arguments. A number of such semantics have been proposed recently, and their formal properties have been studied. While these semantics are sometimes motivated by their better adequacy to capture debates, their behaviour in such multiagent settings is largely unexplored. In this paper, we undertake a study of the multiagent dynamics of a standard gradual semantics. We propose a simple protocol, where agents exchange arguments in order to provide a collective evaluation of the value of a given argument (i.e an issue), and may learn new arguments from the other agents, as well as an extended version allowing votes. The debate proceeds following a better response dynamics. We study how the value of the issue and the agents opinion evolve, depending on various parameters of this setting.

Cooperation is fundamental in Multi-Agent Systems (MAS) and Multi-Agent Reinforcement Learning (MARL), often requiring agents to balance individual gains with collective rewards. In this regard, this paper aims to investigate strategies to invoke cooperation in game-theoretic scenarios, namely the Iterated Prisoner's Dilemma, where agents must optimize both individual and group outcomes. Existing cooperative strategies are analyzed for their effectiveness in promoting group-oriented behavior in repeated games. Modifications are proposed where encouraging group rewards will also result in a higher individual gain, addressing real-world dilemmas seen in distributed systems. The study extends to scenarios with exponentially growing agent populations ($N \longrightarrow +\infty$), where traditional computation and equilibrium determination are challenging. Leveraging mean-field game theory, equilibrium solutions and reward structures are established for infinitely large agent sets in repeated games. Finally, practical insights are offered through simulations using the Multi Agent-Posthumous Credit Assignment trainer, and the paper explores adapting simulation algorithms to create scenarios favoring cooperation for group rewards. These practical implementations bridge theoretical concepts with real-world applications.

Understanding the emergence of prosocial behaviors (e.g., cooperation and trust) among self-interested agents is an important problem in many disciplines. Network structure and institutional incentives (e.g., punishing antisocial agents) are known to promote prosocial behaviors, when acting in isolation, one mechanism being present at a time. Here we study the interplay between these two mechanisms to see whether they are independent, interfering or synergetic. Using evolutionary game theory, we show that punishing antisocial agents and a regular networked structure not only promote prosocial behaviors among agents playing the trust game, but they also interplay with each other, leading to interference or synergy, depending on the game parameters. Synergy emerges on a wider range of parameters than interference does. In this domain, the combination of incentives and networked structure improves the efficiency of incentives, yielding prosocial behaviors at a lower cost than the incentive does alone. This has a significant implication in the promotion of prosocial behaviors in multi-agent systems.

Natural language has long enabled human cooperation, but its lossy, ambiguous, and indirect nature limits the potential of collective intelligence. While machines are not subject to these constraints, most LLM-based multi-agent systems still rely solely on natural language, exchanging tokens or their embeddings. To go beyond language, we introduce a new paradigm, thought communication, which enables agents to interact directly mind-to-mind, akin to telepathy. To uncover these latent thoughts in a principled way, we formalize the process as a general latent variable model, where agent states are generated by an unknown function of underlying thoughts. We prove that, in a nonparametric setting without auxiliary information, both shared and private latent thoughts between any pair of agents can be identified. Moreover, the global structure of thought sharing, including which agents share which thoughts and how these relationships are structured, can also be recovered with theoretical guarantees. Guided by the established theory, we develop a framework that extracts latent thoughts from all agents prior to communication and assigns each agent the relevant thoughts, along with their sharing patterns. This paradigm naturally extends beyond LLMs to all modalities, as most observational data arise from hidden generative processes. Experiments on both synthetic and real-world benchmarks validate the theory and demonstrate the collaborative advantages of thought communication. We hope this work illuminates the potential of leveraging the hidden world, as many challenges remain unsolvable through surface-level observation alone, regardless of compute or data scale.

Cooperation at scale is critical for achieving a sustainable future for humanity. However, achieving collective, cooperative behavior—in which intelligent actors in complex environments jointly improve their well-being—remains poorly understood. Complex systems science (CSS) provides a rich understanding of collective phenomena, the evolution of cooperation, and the institutions that can sustain both. Yet, much of the theory in this area fails to fully consider individual-level complexity and environmental context—largely for the sake of tractability and because it has not been clear how to do so rigorously. These elements are well captured in multiagent reinforcement learning (MARL), which has recently put focus on cooperative (artificial) intelligence. However, typical MARL simulations can be computationally expensive and challenging to interpret. In this perspective, we propose that bridging CSS and MARL affords new directions forward. Both fields can complement each other in their goals, methods, and scope. MARL offers CSS concrete ways to formalize cognitive processes in dynamic environments. CSS offers MARL improved qualitative insight into emergent collective phenomena. We see this approach as providing the necessary foundations for a proper science of collective, cooperative intelligence. We highlight work that is already heading in this direction and discuss concrete steps for future research.

Abstract Cooperation is very important in human society and identified as an essential principle of evolution, but how to promote cooperation among rational individuals remains a huge challenge. Recent works have found that prosocial exclusion can work as a powerful control strategy to promote cooperation effectively. However, it remains unclear whether prosocial exclusion can still favor cooperation when antisocial exclusion is introduced. And does prosocial exclusion have evolutionary advantages when comparing with prosocial and antisocial punishment strategies? To address these issues, we first introduce prosocial and antisocial pool exclusion strategies into the public goods game and study the stationary distribution of each strategy in finite well-mixed populations. We find that the introduction of antisocial exclusion inhibits cooperation, but it does not reduce the evolutionary advantage of prosocial exclusion. We then investigate the competition between the full set of pool exclusion and pool punishment strategies, and reveal that prosocial pool excluders can do better than other strategists no matter whether the second-order sanctioning is considered or not. Our results suggest that social exclusion is a better way for restraining defection than costly punishment, even when antisocial behavior is allowed.

We study how wealth inequality influences behavioral dynamics in groups of independent reinforcement learners facing a threshold public goods dilemma with uncertain returns. The game allows individuals to contribute or not to a common pool to reduce their chances of future losses. The non-linearity introduced by the threshold, the stochasticity introduced by the risk and the wealth heterogeneity of players result in a game setting with multiple equilibria. We find that the learners' dynamics in this case play a major role in determining the attained equilibrium point. Our results suggest that, under individual-based learning, wealth inequality can have sizable effects on the emerging collective behaviors, decreasing the overall chances of group success. Moreover, we compute the class-based Nash equilibria (i.e., where same wealth-class agents are assumed to play the same strategy) for this game and compare the performance of groups composed of independent learning agents with the performance obtained under the payoff maximizing class-based Nash equilibrium. We find that the learned strategies never really match optimal performance for all tested values of risk.

Relationships between people in real life are dynamically changed with the interaction process, and due to the heterogeneous preferences, this change is different from person to person. Based on this observation, we propose a new spatial and weighted prisoner's dilemma game model with heterogeneous individuals. Two types of tags, namely, tag-F (concerned about social fairness) and tag-W (concerned about personal well-being), are introduced to describe individuals' different preferences. The link weights indicating the interaction strength between individuals are updated based on different rules that depend on their tags. Through simulations, we verify that a large link weight control factor and a high proportion of tag-F individuals favor the emergence and persistence of cooperation. In addition, an increase in the link weight sensitivity factor favors the evolution of cooperation when the link weight control factor is small. Moreover, while the level of cooperation increases with the proportion of tag-F type in the population, contrary to our intuition, when the population consists entirely of tag-F individuals, in some cases, cooperation cannot reach a higher level compared with the situation when they are mixed with tag-W type. However, at high dilemma intensities, cooperators emerge only when the entire population consists of tag-F type. These results may provide some new insights into the impact of the evolutionary weighted network with heterogeneous preferences on collective cooperative behavior.

Simulation models of pedestrian dynamics have become an invaluable tool for evacuation planning. Typically, crowds are assumed to stream unidirectionally towards a safe area. Simulated agents avoid collisions through mechanisms that belong to each individual, such as being repelled from each other by imaginary forces. But classic locomotion models fail when collective cooperation is called for, notably when an agent, say a first-aid attendant, needs to forge a path through a densely packed group. We present a controlled experiment to observe what happens when humans pass through a dense static crowd. We formulate and test hypotheses on salient phenomena. We discuss our observations in a psychological framework. We derive a model that incorporates: agents’ perception and cognitive processing of a situation that needs cooperation; selection from a portfolio of behaviours, such as being cooperative; and a suitable action, such as swapping places. Agents’ ability to successfully get through a dense crowd emerges as an effect of the psychological model.

We examine how wealth inequality and diversity in the perception of risk of a collective disaster impact cooperation levels in the context of a public goods game with uncertain and non-linear returns. In this game, individuals face a collective-risk dilemma where they may contribute or not to a common pool to reduce their chances of future losses. We draw our conclusions based on social simulations with populations of independent reinforcement learners with diverse levels of risk and wealth. We find that both wealth inequality and diversity in risk assessment can hinder cooperation and augment collective losses. Additionally, wealth inequality further exacerbates long term inequality, causing rich agents to become richer and poor agents to become poorer. On the other hand, diversity in risk only amplifies inequality when combined with bias in group assortment—i.e., high probability that agents from the same risk class play together. Our results also suggest that taking wealth inequality into account can help to design effective policies aiming at leveraging cooperation in large group sizes, a configuration where collective action is harder to achieve. Finally, we characterize the circumstances under which risk perception alignment is crucial and those under which reducing wealth inequality constitutes a deciding factor for collective welfare.

In this paper, we investigate the role of risk diversity in groups of agents learning to play collective risk dilemmas (CRDs). We show that risk diversity poses new challenges to cooperation that are not observed in homogeneous groups. While increasing average risk contributes, in general, for agents to cooperate with higher probability, increasing risk diversity significantly reduces a population's ability to achieve a collective target. Risk diversity leads to asymmetrical changes in agents policies --- i.e. the increase in contributions from individuals at high risk is unable to compensate for the decrease in contributions from individuals at low risk --- which reduces the total contributions in a population and overall social welfare. At the same time, risk diversity offers novel opportunities to design financial incentives, which, as we show, can improve cooperation, target achievement and global welfare beyond the levels obtained in the absence of diversity. Our results highlight the need to align risk perceptions among agents and implement diversity-based incentive policies in order to improve collectives' abilities to avoid future catastrophic events.

Abstract How did cooperative strategy evolve remains an open question across disciplines. In most previous studies, they mainly consider the analyzing of game dynamics on the networked multiagent system under different mechanisms. However, there often exists a “government” who regulates the strategies of agents centralized or decentralized in reality. Motivated by this fact, we introduce a fitness control method in this paper, and investigate the strength of external fitness control on the game dynamics in networked multiagent system. According to the classic Monte Carlo simulation, we found that the fitness control rule can significantly enhance the cooperation level in networked multiagent system. In particular, we found that the stronger the local fitness control is, the more widespread cooperative strategy becomes. More interestingly, we found that although the local fitness control is less information needed, it is more powerful in cooperation promotion than that of global fitness control rule. Thus, it is practically significant and will provide a new insight into the control of game dynamics in networked multiagent system for the further research.

Social media platforms are highly interconnected because many users maintain a presence across multiple platforms. Consequently, efforts to limit the spread of misinformation taken by individual platforms can have complex consequences on misinformation diffusion across the social media ecosystem. This is further complicated by the diverse social structures, platform standards, and moderation mechanisms provided on each platform. We study this issue by extending our previous model of Reddit interactions and community-specific moderation measures. By adding a followership-based model of Twitter interactions and facilitating cross-platform user participation, we simulate information diffusion across heterogeneous social media platforms. While incorporating platform-specific moderation mechanisms, we simulate interactions at the user level and specify user-specific attributes. This allows practitioners to conduct experiments with various types of actors and different combinations of moderation. We show how the model can simulate the impacts of such features on discussions facilitated by Reddit and Twitter and the cross-platform spread of misinformation. To validate this model, we use a combination of empirical datasets from three U.S. political events and prior findings from user surveys and studies.

In this paper, I introduce an agent-based approach that leverages Markov models to formalize and understand information propagation within social networks. This methodology aims to explain the internal information states of agents and employs probabilistic model checking to analyze these models. Initially, I provide a comprehensive overview of the current research on information diffusion in social networks. Following this, I delve into the fundamentals of model checking and illustrate how this technique can be used to assess model accuracy, particularly in managing large networks with high precision. To demonstrate the effectiveness of this approach, I conduct a series of experiments. The results show that this paper provides a robust and effective method for analyzing complex information diffusion and network behavior.

With the explosive growth of the Coronavirus Pandemic (COVID-19), misinformation on social media has developed into a global phenomenon with widespread and detrimental societal effects. Despite recent progress and efforts in detecting COVID-19 misinformation on social media networks, this task remains challenging due to the complexity, diversity, multi-modality, and high costs of fact-checking or annotation. In this research, we introduce a systematic and multidisciplinary agent-based modeling approach to limit the spread of COVID-19 misinformation and interpret the dynamic actions of users and communities in evolutionary online (or offline) social media networks. Our model was applied to a Twitter network associated with an armed protest demonstration against the COVID-19 lockdown in Michigan state in May, 2020. We implemented a one-median problem to categorize the Twitter network into six key communities (nodes) and identified information exchange (links) within the network. We measured the response time to COVID-19 misinformation spread in the network and employed a cybernetic organizational method to monitor the Twitter network. The overall misinformation mitigation strategy was evaluated, and agents were allocated to interact with the network based on the measured response time and feedback. The proposed model prioritized the communities based on the agents response times at the operational level. It then optimized agent allocation to limit the spread of COVID19 related misinformation from different communities, improved the information diffusion delay threshold to up to 3 minutes, and ultimately enhanced the mitigation process to reduce misinformation spread across the entire network.

No abstract available

The successful adoptionof innovationsdependson theprovisionof adequate information to farmers. In rural areas of developing countries, farmers usually rely on their social networks as an information source. Hence, policy-makers and program-implementers can benefit from social di usion processes to e ectively disseminate information. This study aims to identify the set of farmers who initially obtain information (‘seeds’) that optimises di usion through the network. It systematically evaluates di erent criteria for seed selection, number of seeds, and their interaction e ects. An empirical Agent-Based Model adjusted to a case study in rural Zambia was applied to predict di usion outcomes for varying seed sets ex ante. Simulations revealed that informing farmers with the most connections leads to highest di usion speed and reach. Also targeting village heads and farmers with high betweenness centrality, who function as bridges connecting di erent parts of the network, enhances di usion. An increased number of seeds improves reach, but the marginal e ects of additional seeds decline. Interdependencies between seed set size and selection criteria highlight the importance of considering both seed selection criteria and seed set size for optimising seeding strategies to enhance information di usion.

This paper investigates the interplay between information diffusion in social networks and its impact on financial markets with an Agent-Based Model (ABM). Agents receive and exchange information about an observable stochastic component of the dividend process of a risky asset \`a la Grossman and Stiglitz. A small proportion of the network has access to a private signal about the component, which can be clean (information) or distorted (misinformation). Other agents are uninformed and can receive information only from their peers. All agents are Bayesian, adjusting their beliefs according to the confidence they have in the source of information. We examine, by means of simulations, how information diffuses in the network and provide a framework to account for delayed absorption of shocks, that are not immediately priced as predicted by classical financial models. We investigate the effect of the network topology on the resulting asset price and evaluate under which condition misinformation diffusion can make the market more inefficient.

In recent years, the problem of maximizing influence has been an important topic in social network research, but existing methods usually ignore the subjective willingness of selected seed nodes and the uncertainty of the probability of propagation, which leads to deviations of marketing results from the plan. This paper proposes a dynamic social network information diffusion influence maximization algorithm NUIM. First, the historical data is used to estimate the probability that the selected seed node is willing to become a seed, and a new propagation probability function Φ is further designed to make the information propagation probability follow the distribution related to the network topology instead of a fixed value. Secondly, to enable the agent to learn the optimal seed node selection strategy, the cyclic learning rate is used to replace the artificially set initial value of the learning rate and the learning rate decay strategy in reinforcement learning. The experimental results show that on the power-law network data set, the influence rate obtained by NUIM can reach 16.43%, which is better than other algorithms with better performance.

We study a campaigner who wants to learn the structure of a social network by observing the underlying diffusion process and intervening on it. Using synchronous majoritarian updates on binary opinions as the underlying dynamics, we offer upper bounds on the campaigner's budget for learning any network with certainty, considering both observation and intervention resources, and further improving them for the case of clique networks. Additionally, we investigate the learning progress of the campaigner when her budget falls below these upper bounds. For such cases, we design a greedy campaigning strategy aimed at optimising the campaigner's information gain at each opinion diffusion step.

No abstract available

ABSTRACT We present a group dynamics model that shows knowledge integration as a process occurring over time. As each individual in the group contact others, his own knowledge changes, and over time the collective knowledge is obtained. This allows modeling knowledge diffusion in a social network and while the models presented in this paper are not competitive in that area, they approach the problem from previously unconsidered direction. We test the behavior of the model in a multi-agent simulation and we test a simple advertisement campaign in a social network. We provide discussion of elements needed for making model more competitive.

No abstract available

A typical epidemic often involves the transmission of a disease, the flow of information regarding the disease, and the spread of human preventive behaviors against the disease. These three processes diffuse simultaneously through human social networks, and interact with one another, forming negative and positive feedback loops in the complex human-disease systems. Few studies, however, have been devoted to coupling all the three diffusions together and representing their interactions. To fill the knowledge gap, this article proposes a spatially explicit agent-based model to simulate a triple-diffusion process in a metropolitan area of 1 million people. The individual-based approach, network model, behavioral theories, and stochastic processes are used to formulate the three diffusions and integrate them together. Compared to the observed facts, the model results reasonably replicate the trends of influenza spread and information propagation. The model thus could be a valid and effective tool to evaluate information/behavior-based intervention strategies. Besides its implications to the public health, the research findings also contribute to network modeling, systems science, and medical geography.

No abstract available

A large scale agent-based model of common Facebook users was designed to develop an understanding of the underlying mechanism of information diffusion within online social networks at a micro-level analysis. The agent-based model network structure is based on a sample from Facebook. Using an erased configuration model and the idea of common neighbours, a new correction procedure was investigated to overcome the problem of missing graph edges to construct a representative sample of the Facebook network graph. The model parameters are based on assumptions and general activity patterns (such as posting rate, time spent on Facebook etc.) taken from general data on Facebook. Using the agent-based model, the impact of post length, post score and publisher's friend count on the spread of wall posts in several scenarios was analyzed. Findings indicated that post content has the highest impact on the success of post propagation. However, amusing and absorbing but lengthy posts (e.g. a funny video) do not spread as well as short but unremarkable ones (e.g. an interesting photo). In contrast to product adoption and disease spread propagation models, the absence of a similar "epidemic" threshold in Facebook post diffusion is observed.

Currently, China is in the period of social transformation. Such transformation continuously results in high group polarization behaviors, which attracts many attentions. In order to explore the evolutionary mechanism and formation process of group polarization behavior, this paper proposes a group polarization model which is integrated into the Susceptible-Infected-Recovered-Susceptible (SIRS) epidemic model. In this paper, firstly, the SIRS epidemic model and the factors of relationship strength are introduced based on the J-A model (proposed by Jager and Amblard) to enhance the information transmission and interaction among individuals. In addition, the BA network (proposed by Barabasi and Albert) model is used as the agent adjacency model due to its closeness to the real social network structure. After that, the Monte Carlo method is applied to conduct experimental simulation. Subsequently, this paper analyzes the simulation results in threefold: (1) comparison of polarization processes with and without integration of the SIRS epidemic model; (2) adjusting the immune recovery parameter γ and the relationship strength z to explore the role of these two parameters in the polarization process; and (3) comparing the polarization effects of different network structures. Through the experiments, we find that BA network is more polarized than small-world network in the same scale. Finally, corresponding measures are proposed to prevent and mitigate the occurrence of group polarization.

For maximizing influence spread in a social network, given a certain budget on the number of seed nodes, we investigate the effects of selecting and activating the seed nodes in multiple phases. In particular, we formulate an appropriate objective function for two-phase influence maximization under the independent cascade model, investigate its properties, and propose algorithms for determining the seed nodes in the two phases. We also study the problem of determining an optimal budget-split and delay between the two phases.

We introduce a computational agent‐based model of innovation diffusion that allows us to analyse the influence of information and communication technology (ICT) development on decision‐making. Model dynamics are based on local emulation between pairs of individuals that generate an evolving social network on which an innovation is virally spread (by word of mouth). Results suggest that ICT development affects the data usefulness for decision‐making by changing the topology of the social network (the means whereby the innovation is propagated). Paradoxically, a higher level of ICT development (providing a larger volume of data) narrows the differences between better and worse launch strategies, thus reducing data‐driven decision‐making usefulness, which then shows diminishing returns on the ICT level.

Influence maximization is a well-investigated problem which asks for key individuals who have significant influence in a given social network. This paper addresses this problem when the social network structure is hidden. We adopt the framework of influence learning from samples and build a neural network model to represent the information diffusion process. Based on the model, we propose two new algorithms NeuGreedy and NeuMax. NeuGreedy simulates the traditional greedy algorithm whilst NeuMax utilizes the weights of connections between neurons. We test the algorithms on both synthetic and real-world datasets. The results verify the effectiveness of the proposed methods as compared to existing algorithms with or without the network structure.

Renewable energy resources and energy-efficient technologies, as well as building retrofitting, are only some of the possible strategies that can achieve more sustainable cities and reduce greenhouse gas emissions. Subsidies and incentives are often provided by governments to increase the number of people adopting these sustainable energy efficiency actions. However, actual sales of green products are currently not as high as would be desired. The present paper applies a hybrid agent-based model (ABM) integrated with a Geographic Information System (GIS) to simulate a complex socio-economic-architectural adaptive system to study the temporal diffusion and the willingness of inhabitants to adopt photovoltaic (PV) systems. The San Salvario neighborhood in Turin (Italy) is used as an exemplary case study for testing consumer behavior associated with this technology, integrating social network theories, opinion formation dynamics and an adaptation of the theory of planned behavior (TPB). Data/characteristics for both buildings and people are explicitly spatialized with the level of detail at the block scale. Particular attention is given to the comparison of the policy mix for supporting decision-makers and policymakers in the definition of the most efficient strategies for achieving a long-term vision of sustainable development. Both variables and outcomes accuracy of the model are validated with historical real-world data.

: Meeting climate goals requires radical changes in the consumption behaviour of individuals. This necessitates an understanding of how the diffusion of low-carbon behaviour will occur. The speed and inter-dependency of these changes in behavioural choices may be modulated by individuals’ culture. We develop an agent-based model to study how behavioural decarbonisation interacts with longer-term cultural change, composed of individuals with multiple behaviours that evolve due to imperfect social learning in a social network. Using the definition of culture as socially transmitted information, we represent individuals’ environmental identity as an aggregation of attitudes towards multiple relevant behaviours. The strength of interaction be-tween individuals is determined by the similarity in their environmental identity, leading to inter-behavioural dependencyand spilloversin greenattitudes. Ourresultsshow thatthe initialdistributionof agentattitudes towards behaviours and asymmetries in social learning, such as confirmation bias, are the main drivers of model dynamics, helping to generate awareness of what roadblocks may appear to deep decarbonisation. To assess the impact of culture beyond a purely diffusive regime, we introduce green influencers as a minority of individuals who broadcast a green attitude. The greatest emissions reduction is achieved with the inclusion of culture, relative to a behavioural independence case, and with low confirmation bias. However, green influencers fail to achieve deep behavioural decarbonisation through solely voluntary action. We identify areas for further research regarding how culture, through inter-behavioural dependence, may be leveraged for climate policy.

No abstract available

ABSTRACT This article reports the findings from simulating the spatial diffusion processes of memes over social media networks by using the approach of agent-based modeling. Different from other studies, this article examines how space and distance affect the diffusion of memes. Simulations were carried out to emulate and to allow assessment of the different levels of efficiency that memes spread spatially and temporally. Analyzed network structures include random networks and preferential attachment networks. Simulated spatial processes for meme diffusion include independent cascade models and linear threshold models. Both simulated and real-world social networks were used in the analysis. Findings indicate that the numbers of information sources and opinion leaders affect the processes of meme diffusion. In addition, geography is still important in the processes of spatial diffusion of memes over social media networks.

Influence Maximization aims to select a subset of elements in a social network to maximize information spread under a diffusion model. While existing work primarily focuses on selecting influential nodes, these approaches assume unrestricted message propagation-an assumption that fails in closed social networks, where content visibility is constrained and node-level activations may be infeasible. Motivated by the growing adoption of privacy-focused platforms such as Signal, Discord, Instagram, and Slack, our work addresses the following fundamental question: How can we learn effective edge activation strategies for influence maximization in closed networks? To answer this question we introduce Reinforcement Learning for Link Activation (RELINK), the first DRL framework for edge-level influence maximization in privacy-constrained networks. It models edge selection as a Markov Decision Process, where the agent learns to activate edges under budget constraints. Unlike prior node-based DRL methods, RELINK uses an edge-centric Q-learning approach that accounts for structural constraints and constrained information propagation. Our framework combines a rich node embedding pipeline with an edge-aware aggregation module. The agent is trained using an n-step Double DQN objective, guided by dense reward signals that capture marginal gains in influence spread. Extensive experiments on real-world networks show that RELINK consistently outperforms existing edge-based methods, achieving up to 15% higher influence spread and improved scalability across diverse settings.

No abstract available

In the last years, the study of rumor spreading on social networks produced a lot of interest among the scientific community, expecially due to the role of social networks in the last political events. The goal of this work is to reproduce real-like diffusions of information and misinformation in a scale-free network using a multi-agent-based model. The data concerning the virtual spreading are easily obtainable, in particular the diffusion of information during the announcement for the discovery of the Higgs Boson on Twitter was recorded and investigated in detail. We made some assumptions on the micro behavior of our agents and registered the effects in a statistical analysis replying the real data diffusion. Then, we studied an hypotetical response to a misinformation diffusion adding debunking agents and trying to model a critic response from the agents using real data from a hoax regarding the Occupy Wall Street movement. After tuning our model to reproduce these results, we measured some network properties and proved the emergence of substantially separated structures like echochambers, independently from the network size scale, i.e. with one hundred, one thousand and ten thousand agents.

Among popular social media platforms, Reddit stands out for its decentralized approach to moderation and community management. Due to this and its community-based network structure, Reddit provides a unique environment for studying the diffusion of knowledge and beliefs over social media. While assortativity, polarization, and user behavior have been examined within empirical contexts, having the ability to model the impacts of different moderation policies and rules across communities could provide useful insights for limiting the spread of misinformation. In this work, we introduce an agent-based model of Reddit interactions and moderating actions. By simulating interactions at the user level and specifying user-specific attributes, our model allows practitioners to conduct experiments with various types of actors and moderators and study their potential impact on Reddit-facilitated discussions and information diffusion. Additionally, subreddit-specific attributes enable communities to have different standards and thresholds for user conduct. To validate this model, we rely on an empirical dataset of over 100K posts and 800K comments across three U.S. political events in addition to user surveys and studies.

No abstract available

No abstract available

No abstract available

Social media platforms, taken in conjunction, can be seen as complex networks; in this context, understanding how agents react to sentiments expressed by their connections is of great interest. Here, the authors show how Network Knowledge Bases help represent the integration of multiple social networks, and explore how information flow can be handled via belief revision operators for local (agent-specific) knowledge bases. They report on preliminary experiments on Twitter data showing that different agent types react differently to the same information — this is a first step toward developing tools to predict how agents behave as information flows in their social environment.

No abstract available

No abstract available

The way of information diffusion among individuals can be quite complicated, and it is not only limited to one type of communication, but also impacted by multiple channels. Meanwhile, it is easier for an agent to accept an idea once the proportion of their friends who take it goes beyond a specific threshold. Furthermore, in social networks, some higher-order structures, such as simplicial complexes and hypergraph, can describe more abundant and realistic phenomena. Therefore, based on the classical multiplex network model coupling the infectious disease with its relevant information, we propose a novel epidemic model, in which the lower layer represents the physical contact network depicting the epidemic dissemination, while the upper layer stands for the online social network picturing the diffusion of information. In particular, the upper layer is generated by random simplicial complexes, among which the herd-like threshold model is adopted to characterize the information diffusion, and the unaware-aware-unaware model is also considered simultaneously. Using the microscopic Markov chain approach, we analyze the epidemic threshold of the proposed epidemic model and further check the results with numerous Monte Carlo simulations. It is discovered that the threshold model based on the random simplicial complexes network may still cause abrupt transitions on the epidemic threshold. It is also found that simplicial complexes may greatly influence the epidemic size at a steady state.

During the special period of the COVID-19 outbreak, this project investigated the driving factors in different information diffusion modes (i.e. broadcasting mode, contagion mode) based on the nomination relations in a charitable social relay campaign on Sina Weibo. Specifically, we mapped a nomination social network and tracked the core communicators in both modes. Besides, we also observed the network from perspectives such as relationships between core communicators and modularity of the whole network. We extracted homophily factors and tested them on representative communities within the largest component of the network. We found that some core communicators distributed in a co-dependent way. At last, we presented several explanations to the phenomenon which can be explored in further research.

The role of misinformation diffusion during a pandemic is crucial. An aspect that requires particular attention in the analysis of misinfodemics is the rationale of the source of false information, in particular how the behavior of agents spreading misinformation through traditional communication outlets and social networks can influence the diffusion of the disease. We studied the process of false information transmission by malicious agents, in the context of a disease pandemic based on data for the COVID-19 emergency in Italy. We model communication of misinformation based on a negative trust relation, supported by findings in the literature that relate the endorsement of conspiracy theories with low trust level towards institutions. We provide an agent-based simulation and consider the effects of a misinfodemic on policies related to lockdown strategies, isolation, protection and distancing measures, and overall negative impact on society during a pandemic. Our analysis shows that there is a clear impact by misinfodemics in aggravating the results of a current pandemic.

No abstract available

No abstract available

No abstract available

No abstract available

There has been a growing interest in enhancing rule-based agent-based models (ABMs) for social media platforms (i.e., X, Reddit) with more realistic large language model (LLM) agents, thereby allowing for a more nuanced study of complex systems. As a result, several LLM-based ABMs have been proposed in the past year. While they hold promise, each simulator is specifically designed to study a particular scenario, making it time-consuming and resource-intensive to explore other phenomena using the same ABM. Additionally, these models simulate only a limited number of agents, whereas real-world social media platforms involve millions of users. To this end, we propose OASIS, a generalizable and scalable social media simulator. OASIS is designed based on real-world social media platforms, incorporating dynamically updated environments (i.e., dynamic social networks and post information), diverse action spaces (i.e., following, commenting), and recommendation systems (i.e., interest-based and hot-score-based). Additionally, OASIS supports large-scale user simulations, capable of modeling up to one million users. With these features, OASIS can be easily extended to different social media platforms to study large-scale group phenomena and behaviors. We replicate various social phenomena, including information spreading, group polarization, and herd effects across X and Reddit platforms. Moreover, we provide observations of social phenomena at different agent group scales. We observe that the larger agent group scale leads to more enhanced group dynamics and more diverse and helpful agents' opinions. These findings demonstrate OASIS's potential as a powerful tool for studying complex systems in digital environments.

We release a large-scale dataset that captures interactions between human users and CommentRobert, an LLM-based social media agent on Weibo. The dataset contains Weibo posts in which users actively mention the LLM agent account @CommentRobert, indicating that the users are interested in interacting with the platform-empowered LLM agent. The dataset contains 557,645 interactions from 304,400 unique users over 17 months. We detail our data collection methodology, user attributes, and content characteristics, underscoring the dataset's value in examining real-world human-LLM agent interactions. Our analysis offers insights into the demographic and behavioral traits of users interested in the selected LLM agent, interaction dynamics between humans and the agent, and linguistic patterns in comments. These interactions provide a unique lens through which to explore how humans perceive, trust, and communicate with LLMs. This dataset enables further research into modeling human intent understanding, improving LLM agent design, and studying the evolution of human-LLM agent relationships. Potential applications also include long-term user engagement prediction and AI-generated comment detection on social platforms. This constructed dataset is available at https://zenodo.org/records/16921462.

No abstract available

No abstract available

No abstract available

While social media usage has taken a prominent role in large social movements, societal constraints are catching up with privacy and transparency in preparation and coordination. To cope with this problem, we propose a method combining agent-based simulation and big data analysis to gain insight into upcoming protests while ensuring a well-weighted and clean analysis process. This simulation method aims to strengthen the information position of governmental officials in advance of a large-scale protest to allocate resources better and take suitable measures to ensure public safety while reducing the preliminary privacy invasive interventions. The proposed method is tested on a real-world case study where posts from Black Lives Matter protests are used to simulate social interaction in advance. Results show that behavioural constructs such as the spillover effect can be predicted based on previous data, contributing to gaining information from a wider perspective.

Cities are complex, dynamic, evolving adaptive systems comprised of people as well as interconnected physical infrastructure. Simulation modeling can help us understand and shape the evolution of our cities. In this paper, we describe an agent-based simulation modeling framework applied to Chicago, called chiSIM (for the Chicago Social Interaction Model). Each person residing in Chicago is represented as an agent in chiSIM; all places where people can be located in Chicago also are represented. The model simulates the movements of people between locations on an hourly basis during the course of a typical day. Co-located agents engage in various kinds of social interactions, such as exchanging information, engaging in business transactions, or simply sharing physical proximity. We discuss technical approaches to large-scale urban modeling including development of synthetic populations, efficiency gains through distributed processing, logging and analysis of simulation results, and visualization.

Housing markets are inherently spatial, yet many existing models fail to capture this spatial dimension. Here, we introduce a new graph-based approach for incorporating a spatial component in a large-scale urban housing agent-based model (ABM). The model explicitly captures several social and economic factors that influence the agents’ decision-making behaviour (such as fear of missing out, their trend-following aptitude, and the strength of their submarket outreach), and interprets these factors in spatial terms. The proposed model is calibrated and validated with the housing market data for the Greater Sydney region. The ABM simulation results not only include predictions for the overall market, but also produce area-specific forecasting at the level of local government areas within Sydney as arising from individual buy and sell decisions. In addition, the simulation results elucidate agent preferences in submarkets, highlighting differences in agent behaviour, for example, between first-time home buyers and investors, and between both local and overseas investors.

In this work, we study the risks of collective financial fraud in large-scale multi-agent systems powered by large language model (LLM) agents. We investigate whether agents can collaborate in fraudulent behaviors, how such collaboration amplifies risks, and what factors influence fraud success. To support this research, we present MultiAgentFraudBench, a large-scale benchmark for simulating financial fraud scenarios based on realistic online interactions. The benchmark covers 28 typical online fraud scenarios, spanning the full fraud lifecycle across both public and private domains. We further analyze key factors affecting fraud success, including interaction depth, activity level, and fine-grained collaboration failure modes. Finally, we propose a series of mitigation strategies, including adding content-level warnings to fraudulent posts and dialogues, using LLMs as monitors to block potentially malicious agents, and fostering group resilience through information sharing at the societal level. Notably, we observe that malicious agents can adapt to environmental interventions. Our findings highlight the real-world risks of multi-agent financial fraud and suggest practical measures for mitigating them. Code is available at https://github.com/zheng977/MutiAgent4Fraud.

No abstract available

No abstract available

No abstract available

No abstract available

Many agent-based models (ABMs) try to explain large-scale phenomena by reducing them to behaviors at lower scales. At these scales in social systems are functional groups such as households, religious congregations, coops and local governments. The intra-group dynamics of functional groups often generate inefficient or unexpected behavior that cannot be predicted by modeling groups as basic units. We introduce a framework for modeling intra-group decision-making and its interaction with social norms, using the household as our focus. We select phenomena related to women’s empowerment in agriculture as examples influenced by both intra-household dynamics and gender norms. Our framework proves more capable of replicating these phenomena than two common types of ABMs. We conclude that it is not enough to build multi-scale models; explaining social behaviors entails modeling intra-scale dynamics.

No abstract available

No abstract available

Abstract Designing cooperation-enhancing protocols for large-scale multiagent networked systems has been a grand transdisciplinary challenge. In recent years, tag-based interactions and conformity bias have been studied extensively but separately as two viable mechanisms for cooperation enhancement in such systems. Inspired by recent studies on interaction effects in social dilemmas, we herein develop a hybrid, multiagent-based, co-evolutionary model of tag-mediated cooperation and conformity with conditional and unconditional strategies. Through a series of extensive Monte Carlo simulation experiments, we study four variants of this computational model, finding that under the majority rule, the nonconforming unconditional cooperators and conformity biased transmission of other strategies can lead to global altruistic dominance. Employing a random pinning control mechanism, we further observe that only a small fraction of nonconforming altruists is actually required to drive the system towards a robust persistence of pure altruism. Our analytic results in combination with further computational experiments reveal that spatial structure and nonconformity of cooperators are the two indispensable ingredients for the stable dominance of altruistic behavior in tag-based multiagent systems. Our findings can be beneficial for developing novel cooperation-controlling techniques in distributed self-organizing systems such as peer-to-peer networks or in various social networking and viral marketing technologies.

Significance Digital media and online social networks have upended how narratives are constructed and shared, shaping cognition and culture in unexpected ways. Individuals within these networks have increased narrative agency, which enables them to directly contribute to and share evolving stories. Understanding the reflexive processes between individual and networked group narrative dynamics requires new forms of behavioral experimentation and modeling. We conducted a large-scale online social network experiment on narrative interaction, analyzed language dynamics using agent-based modeling, and developed quantitative measures of narrative alignment. Results reveal how network structure interacts with individual decision-making to influence the dynamics and semantic content of shared beliefs, with implications for understanding how narrative information flows through online networks with different neighborhood connections.

As embodied intelligence emerges as a core frontier in artificial intelligence research, simulation platforms must evolve beyond low-level physical interactions to capture complex, human-centered social behaviors. We introduce FreeAskWorld, an interactive simulation framework that integrates large language models (LLMs) for high-level behavior planning and semantically grounded interaction, informed by theories of intention and social cognition. Our framework supports scalable, realistic human-agent simulations and includes a modular data generation pipeline tailored for diverse embodied tasks. To validate the framework, we extend the classic Vision-and-Language Navigation (VLN) task into a interaction enriched Direction Inquiry setting, wherein agents can actively seek and interpret navigational guidance. We present and publicly release FreeAskWorld, a large-scale benchmark dataset comprising reconstructed environments, six diverse task types, 16 core object categories, 63,429 annotated sample frames, and more than 17 hours of interaction data to support training and evaluation of embodied AI systems. We benchmark VLN models, and human participants under both open-loop and closed-loop settings. Experimental results demonstrate that models fine-tuned on FreeAskWorld outperform their original counterparts, achieving enhanced semantic understanding and interaction competency. These findings underscore the efficacy of socially grounded simulation frameworks in advancing embodied AI systems toward sophisticated high-level planning and more naturalistic human-agent interaction. Importantly, our work underscores that interaction itself serves as an additional information modality.

Social interaction occurs across many time scales and varying numbers of agents; from one-on-one to large-scale coordination in organizations, crowds, cities, and colonies. These contexts, are characterized by emergent self-organization that implies higher order coordinated patterns occurring over time that are not due to the actions of any particular agents, but rather due to the collective ordering that occurs from the interactions of the agents. Extant research to understand these social coordination dynamics (SCD) has primarily examined dyadic contexts performing rhythmic tasks. To advance this area of study, we elaborate on attractor dynamics, our ability to depict them visually, and quantitatively model them. Primarily, we combine difference/differential equation modeling with mixture modeling as a way to infer the underlying topological features of the data, which can be described in terms of attractor dynamic patterns. The advantage of this approach is that we are able to quantify the self-organized dynamics that agents exhibit, link these dynamics back to activity from individual agents, and relate it to other variables central to understanding the coordinative functionality of a system's behavior. We present four examples that differ in the number of variables used to depict the attractor dynamics (1, 2, and 6) and range from simulated to non-simulated data sources. We demonstrate that this is a flexible method that advances scientific study of SCD in a variety of multi-agent systems.

No abstract available

In this study, we develop a model of hepatitis C virus (HCV) transmission dynamics capable of analyzing the health, economic and epidemiological impact of treatment and large-scale screening in the Indian context. The model simulates the interaction of infected and uninfected agents in environments wherein key risk factors of HCV transmission operate. The natural history of disease is simulated using a previously published and validated Markov model. The agent interaction/transmission environments simulated by the model include a home environment for transmission via unprotected sex, a medical environment for transmission via unsafe medical practices, educational and social interaction environments for conversion of non-injecting drug user (IDU) agents to IDUs and transmission via sharing of injecting equipment among IDUs. The model is calibrated to current HCV and IDU prevalence targets. We present model calibration results and preliminary results for the impact of treatment uptake rates on HCV and IDU prevalence.

No abstract available

Criminal behaviour exists in many variations, each with its own cause. A large group of offenders only shows criminal behaviour during adolescence. This kind of behaviour is largely influenced by the interaction with others, through social learning. This paper contributes a dynamical agent-based approach to simulate social learning of adolescence-limited criminal behaviour, illustrated for a small school class. The model is designed in such a way that it can be compared with data resulting from a large scale empirical study.

No abstract available

It is common to define the structure of interactions among a population of agents by a network. Most of agent-based models were shown highly sensitive to that network, so the relevance of simulation results directely depends on the descriptive power of that network. When studying social dynamics in large populations, that network cannot be collected, and is rather generated by algorithms which aim to fit general properties of social networks. However, more precise data is available at a country scale in the form of socio-demographic studies, census or sociological studies. These "scattered statistics" provide rich information, especially on agents' attributes, similar properties of tied agents and affiliations. In this paper, we propose a generic methodology to bring up together these scattered statistics with bayesian networks. We explain how to generate a population of heterogeneous agents, and how to create links by using both scattered statistics and knowledge on social selection processes. The methodology is illustrated by generating an interaction network for rural Kenya which includes familial structure, colleagues and friendship constrained given field studies and statistics.