Fully autonomous teams of LLM-powered AI agents are emerging that collaborate to perform complex tasks for users. What challenges do developers face when trying to build and debug these AI agent teams? In formative interviews with five AI agent developers, we identify core challenges: difficulty reviewing long agent conversations to localize errors, lack of support in current tools for interactive debugging, and the need for tool support to iterate on agent configuration. Based on these needs, we developed an interactive multi-agent debugging tool, AGDebugger, with a UI for browsing and sending messages, the ability to edit and reset prior agent messages, and an overview visualization for navigating complex message histories. In a two-part user study with 14 participants, we identify common user strategies for steering agents and highlight the importance of interactive message resets for debugging. Our studies deepen understanding of interfaces for debugging increasingly important agentic workflows.

The very few studies that have attempted to formulate multi-agent reinforcement learning (RL) algorithms for adaptive traffic signal control have mainly used value-based RL methods although recent literature has shown that policy-based methods may perform better in partially observable environments. Additionally, because of the simplifying assumptions on signal timing made almost universally across previous studies, RL methods remain largely untested for real-world signal timing plans. This study formulates a multi-agent proximal policy optimization (MA-PPO) algorithm to implement adaptive and coordinated traffic control along an arterial corridor. The formulated MA-PPO has centralized critic architecture under the centralized training and decentralized execution framework. All agents are formulated to allow selection and implementation of up to eight signal phases as commonly implemented in the field controllers. The formulated algorithm is tested on a simulated real-world corridor with seven intersections, actual/complete traffic movements and signal phases, traffic volumes, and network geometry including intersection spacings. The performance of the formulated MA-PPO adaptive control algorithm is compared with the field implemented coordinated and actuated signal control (ASC) plans modeled using Vissim-MaxTime software in the loop simulation (SILs). The speed of convergence for each agent largely depended on the size of the action space which in turn depended on the number and sequence of signal phases. Compared with the currently implemented ASC signal timings, MA-PPO showed a travel time reduction of about 14% and 29%, respectively for the two through movements across the entire test corridor. Through volume sensitivity experiments, the formulated MA-PPO showed good stability, robustness and adaptability to changes in traffic demand.

We formulate the problem of fake news detection using distributed fact-checkers (agents) with unknown reliability. The stream of news/statements is modeled as an independent and identically distributed binary source (to represent true and false statements). Upon observing a news, agent $i$ labels the news as true or false which reflects the true validity of the statement with some probability $1-\pi_i$. In other words, agent $i$ misclassified each statement with error probability $\pi_i\in (0,1)$, where the parameter $\pi_i$ models the (un)trustworthiness of agent $i$. We present an algorithm to learn the unreliability parameters, resulting in a distributed fact-checking algorithm. Furthermore, we extensively analyze the discrete-time limit of our algorithm.

We study fair division of indivisible goods under the maximin share (MMS) fairness criterion in settings where agents are grouped into a small number of types, with agents within each type having identical valuations. For the special case of a single type, an exact MMS allocation is always guaranteed to exist. However, for two or more distinct agent types, exact MMS allocations do not always exist, shifting the focus to establishing the existence of approximate-MMS allocations. A series of works over the last decade has resulted in the best-known approximation guarantee of $\frac{3}{4} + \frac{3}{3836}$. In this paper, we improve the approximation guarantees for settings where agents are grouped into two or three types, a scenario that arises in many practical settings. Specifically, we present novel algorithms that guarantee a $\frac{4}{5}$-MMS allocation for two agent types and a $\frac{16}{21}$-MMS allocation for three agent types. Our approach leverages the MMS partition of the majority type and adapts it to provide improved fairness guarantees for all types.

Personality Computing is a field at the intersection of Personality Psychology and Computer Science. Started in 2005, research in the field utilizes computational methods to understand and predict human personality traits. The expansion of the field has been very rapid and, by analyzing digital footprints (text, images, social media, etc.), it helped to develop systems that recognize and even replicate human personality. While offering promising applications in talent recruiting, marketing and healthcare, the ethical implications of Personality Computing are significant. Concerns include data privacy, algorithmic bias, and the potential for manipulation by personality-aware Artificial Intelligence. This paper provides an overview of the field, explores key methodologies, discusses the challenges and threats, and outlines potential future directions for responsible development and deployment of Personality Computing technologies.

Social biases and belief-driven behaviors can significantly impact Large Language Models (LLMs) decisions on several tasks. As LLMs are increasingly used in multi-agent systems for societal simulations, their ability to model fundamental group psychological characteristics remains critical yet under-explored. In this study, we present a multi-agent framework that simulates belief congruence, a classical group psychology theory that plays a crucial role in shaping societal interactions and preferences. Our findings reveal that LLMs exhibit amplified belief congruence compared to humans, across diverse contexts. We further investigate the implications of this behavior on two downstream tasks: (1) misinformation dissemination and (2) LLM learning, finding that belief congruence in LLMs increases misinformation dissemination and impedes learning. To mitigate these negative impacts, we propose strategies inspired by: (1) contact hypothesis, (2) accuracy nudges, and (3) global citizenship framework. Our results show that the best strategies reduce misinformation dissemination by up to 37% and enhance learning by 11%. Bridging social psychology and AI, our work provides insights to navigate real-world interactions using LLMs while addressing belief-driven biases.

Consider the decision-making setting where agents elect a panel by expressing both positive and negative preferences. Prominently, in constitutional AI, citizens democratically select a slate of ethical preferences on which a foundation model is to be trained. There, in practice, agents may both approve and disapprove of different ethical principles. Proportionality has been well-studied in computational social choice for approval ballots, but its meaning remains unclear when negative sentiments are also considered. In this work, we propose two conceptually distinct approaches to interpret proportionality in the presence of up and down votes. The first approach treats the satisfaction from electing candidates and the impact of vetoing them as comparable, leading to combined proportionality guarantees. The second approach considers veto power separately, introducing guarantees distinct from traditional proportionality. We formalize axioms for each perspective and examine their satisfiability by suitable adaptations of Phragm\'en's rule, Proportional Approval Voting rule and the Method of Equal Shares.

Large Language Models (LLMs) have shown remarkable capabilities as autonomous agents, yet existing benchmarks either focus on single-agent tasks or are confined to narrow domains, failing to capture the dynamics of multi-agent coordination and competition. In this paper, we introduce MultiAgentBench, a comprehensive benchmark designed to evaluate LLM-based multi-agent systems across diverse, interactive scenarios. Our framework measures not only task completion but also the quality of collaboration and competition using novel, milestone-based key performance indicators. Moreover, we evaluate various coordination protocols (including star, chain, tree, and graph topologies) and innovative strategies such as group discussion and cognitive planning. Notably, gpt-4o-mini reaches the average highest task score, graph structure performs the best among coordination protocols in the research scenario, and cognitive planning improves milestone achievement rates by 3%. Code and datasets are public available at https://github.com/MultiagentBench/MARBLE.

In this paper, we adapt a two species agent-based cancer model that describes the interaction between cancer cells and healthy cells on a uniform grid to include the interaction with a third species -- namely immune cells. We run six different scenarios to explore the competition between cancer and immune cells and the initial concentration of the immune cells on cancer dynamics. We then use coupled equation learning to construct a population-based reaction model for each scenario. We show how they can be unified into a single surrogate population-based reaction model, whose underlying three coupled ordinary differential equations are much easier to analyse than the original agent-based model. As an example, by finding the single steady state of the cancer concentration, we are able to find a linear relationship between this concentration and the initial concentration of the immune cells. This then enables us to estimate suitable values for the competition and initial concentration to reduce the cancer substantially without performing additional complex and expensive simulations from an agent-based stochastic model. The work shows the importance of performing equation learning from agent-based stochastic data for gaining key insights about the behaviour of complex cellular dynamics.

Most microeconomic models of interest involve optimizing a piecewise linear function. These include contract design in hidden-action principal-agent problems, selling an item in posted-price auctions, and bidding in first-price auctions. When the relevant model parameters are unknown and determined by some (unknown) probability distributions, the problem becomes learning how to optimize an unknown and stochastic piecewise linear reward function. Such a problem is usually framed within an online learning framework, where the decision-maker (learner) seeks to minimize the regret of not knowing an optimal decision in hindsight. This paper introduces a general online learning framework that offers a unified approach to tackle regret minimization for piecewise linear rewards, under a suitable monotonicity assumption commonly satisfied by microeconomic models. We design a learning algorithm that attains a regret of $\widetilde{O}(\sqrt{nT})$, where $n$ is the number of ``pieces'' of the reward function and $T$ is the number of rounds. This result is tight when $n$ is \emph{small} relative to $T$, specifically when $n \leq T^{1/3}$. Our algorithm solves two open problems in the literature on learning in microeconomic settings. First, it shows that the $\widetilde{O}(T^{2/3})$ regret bound obtained by Zhu et al. [Zhu+23] for learning optimal linear contracts in hidden-action principal-agent problems is not tight when the number of agent's actions is small relative to $T$. Second, our algorithm demonstrates that, in the problem of learning to set prices in posted-price auctions, it is possible to attain suitable (and desirable) instance-independent regret bounds, addressing an open problem posed by Cesa-Bianchi et al. [CBCP19].