$τ^2$-Bench: Evaluating Conversational Agents in a Dual-Control Environment

• URL: http://arxiv.org/abs/2506.07982v1
• Date: Mon, 09 Jun 2025 17:52:18 GMT
• Title: $τ^2$-Bench: Evaluating Conversational Agents in a Dual-Control Environment
• Authors: Victor Barres, Honghua Dong, Soham Ray, Xujie Si, Karthik Narasimhan,
• Abstract summary: Existing benchmarks for AI agents simulate single-control environments.<n>We introduce $tau2$-bench, where both agent and user make use of tools to act in a shared, dynamic environment.<n>In particular, our experiments show significant performance drops when agents shift from no-user to dual-control.
• Score: 32.345011712015435
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Existing benchmarks for conversational AI agents simulate single-control environments, where only the AI agent can use tools to interact with the world, while the user remains a passive information provider. This differs from real-world scenarios like technical support, where users need to actively participate in modifying the state of the (shared) world. In order to address this gap, we introduce $\tau^2$-bench, with four key contributions: 1) A novel Telecom dual-control domain modeled as a Dec-POMDP, where both agent and user make use of tools to act in a shared, dynamic environment that tests both agent coordination and communication, 2) A compositional task generator that programmatically creates diverse, verifiable tasks from atomic components, ensuring domain coverage and controlled complexity, 3) A reliable user simulator tightly coupled with the environment, whose behavior is constrained by tools and observable states, improving simulation fidelity, 4) Fine-grained analysis of agent performance through multiple ablations including separating errors arising from reasoning vs communication/coordination. In particular, our experiments show significant performance drops when agents shift from no-user to dual-control, highlighting the challenges of guiding users. Overall, $\tau^2$-bench provides a controlled testbed for agents that must both reason effectively and guide user actions.

Related papers

• UserBench: An Interactive Gym Environment for User-Centric Agents [110.77212949007958]
  Large Language Models (LLMs)-based agents have made impressive progress in reasoning and tool use, but their ability to proactively collaborate with users remains underexplored.<n>We introduce UserBench, a user-centric benchmark designed to evaluate agents in multi-turn, preference-driven interactions.
  arXiv  Detail & Related papers  (2025-07-29T17:34:12Z)
• Breaking Single-Tester Limits: Multi-Agent LLMs for Multi-User Feature Testing [22.10696272175415]
  We propose MAdroid, a novel multi-agent approach powered by the Large Language Models (LLMs) to automate the multi-user interactive task for app feature testing.<n>Specifically, MAdroid employs two functional types of multi-agents: user agents (Operator) and supervisor agents (Coordinator and Observer)<n>Our evaluation, which included 41 multi-user interactive tasks, demonstrates the effectiveness of our approach, achieving 82.9% of the tasks with 96.8% action similarity.
  arXiv  Detail & Related papers  (2025-06-21T01:38:53Z)
• ThinkGeo: Evaluating Tool-Augmented Agents for Remote Sensing Tasks [54.52092001110694]
  ThinkGeo is a benchmark designed to evaluate tool-augmented agents on remote sensing tasks via structured tool use and multi-step planning.<n>Inspired by tool-interaction paradigms, ThinkGeo includes human-curated queries spanning a wide range of real-world applications.<n>Our analysis reveals notable disparities in tool accuracy and planning consistency across models.
  arXiv  Detail & Related papers  (2025-05-29T17:59:38Z)
• Get Experience from Practice: LLM Agents with Record & Replay [16.179801770737892]
  This paper proposes a new paradigm called AgentRR (Agent Record & Replay), which introduces the classical record-and-replay mechanism into AI agent frameworks.<n>We detail a multi-level experience abstraction method and a check function mechanism in AgentRR.<n>In addition, we explore multiple application modes of AgentRR, including user-recorded task demonstration, large-small model collaboration and privacy-aware agent execution.
  arXiv  Detail & Related papers  (2025-05-23T10:33:14Z)
• Mobile-Bench-v2: A More Realistic and Comprehensive Benchmark for VLM-based Mobile Agents [33.899782380901314]
  VLM-based mobile agents are increasingly popular due to their capabilities to interact with smartphone GUIs and XML-structured texts.<n>Existing online benchmarks struggle with obtaining stable reward signals due to dynamic environmental changes.<n>Mobile-Bench-v2 includes a common task split, with offline multi-path evaluation to assess the agent's ability to obtain step rewards.
  arXiv  Detail & Related papers  (2025-05-17T07:58:34Z)
• Collaborative Instance Object Navigation: Leveraging Uncertainty-Awareness to Minimize Human-Agent Dialogues [54.81155589931697]
  Collaborative Instance object Navigation (CoIN) is a new task setting where the agent actively resolve uncertainties about the target instance.<n>We propose a novel training-free method, Agent-user Interaction with UncerTainty Awareness (AIUTA)<n>First, upon object detection, a Self-Questioner model initiates a self-dialogue within the agent to obtain a complete and accurate observation description.<n>An Interaction Trigger module determines whether to ask a question to the human, continue or halt navigation.
  arXiv  Detail & Related papers  (2024-12-02T08:16:38Z)
• SPA-Bench: A Comprehensive Benchmark for SmartPhone Agent Evaluation [89.24729958546168]
  Smartphone agents are increasingly important for helping users control devices efficiently.<n>We present SPA-Bench, a comprehensive SmartPhone Agent Benchmark designed to evaluate (M)LLM-based agents.
  arXiv  Detail & Related papers  (2024-10-19T17:28:48Z)
• $τ$-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains [43.43344028212623]
  $tau$-bench is a benchmark emulating dynamic conversations between a user and a language agent. We employ an efficient and faithful evaluation process that compares the database state at the end of a conversation with the annotated goal state.
  arXiv  Detail & Related papers  (2024-06-17T19:33:08Z)
• Learning to Use Tools via Cooperative and Interactive Agents [58.77710337157665]
  Tool learning empowers large language models (LLMs) as agents to use external tools and extend their utility. We propose ConAgents, a Cooperative and interactive Agents framework, which coordinates three specialized agents for tool selection, tool execution, and action calibration separately. Our experiments on three datasets show that the LLMs, when equipped with ConAgents, outperform baselines with substantial improvement.
  arXiv  Detail & Related papers  (2024-03-05T15:08:16Z)
• AgentCF: Collaborative Learning with Autonomous Language Agents for Recommender Systems [112.76941157194544]
  We propose AgentCF for simulating user-item interactions in recommender systems through agent-based collaborative filtering. We creatively consider not only users but also items as agents, and develop a collaborative learning approach that optimize both kinds of agents together. Overall, the optimized agents exhibit diverse interaction behaviors within our framework, including user-item, user-user, item-item, and collective interactions.
  arXiv  Detail & Related papers  (2023-10-13T16:37:14Z)
• MUG: Interactive Multimodal Grounding on User Interfaces [12.035123646959669]
  We present MUG, a novel interactive task for multimodal grounding where a user and an agent work collaboratively on an interface screen. Prior works modeled multimodal UI grounding in one round: the user gives a command and the agent responds to the command. MUG allows multiple rounds of interactions such that upon seeing the agent responses, the user can give further commands for the agent to refine or even correct its actions.
  arXiv  Detail & Related papers  (2022-09-29T21:08:18Z)
• Realistic simulation of users for IT systems in cyber ranges [63.20765930558542]
  We instrument each machine by means of an external agent to generate user activity. This agent combines both deterministic and deep learning based methods to adapt to different environment. We also propose conditional text generation models to facilitate the creation of conversations and documents.
  arXiv  Detail & Related papers  (2021-11-23T10:53:29Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.