SidConArena is a new multi-phase benchmark framework formalizing a partially observable stochastic game for evaluating LLM agents in open-ended positive-sum bargaining with negotiation, converter production, and sealed-bid auctions.
hub
How Well Can
14 Pith papers cite this work. Polarity classification is still indexing.
14
Pith papers citing it
abstract
Negotiation is the basis of social interactions; humans negotiate everything from the price of cars to how to share common resources. With rapidly growing interest in using large language models (LLMs) to act as agents on behalf of human users, such LLM agents would also need to be able to negotiate. In this paper, we study how well LLMs can negotiate with each other. We develop NegotiationArena: a flexible framework for evaluating and probing the negotiation abilities of LLM agents. We implemented three types of scenarios in NegotiationArena to assess LLM's behaviors in allocating shared resources (ultimatum games), aggregate resources (trading games) and buy/sell goods (price negotiations). Each scenario allows for multiple turns of flexible dialogues between LLM agents to allow for more complex negotiations. Interestingly, LLM agents can significantly boost their negotiation outcomes by employing certain behavioral tactics. For example, by pretending to be desolate and desperate, LLMs can improve their payoffs by 20\% when negotiating against the standard GPT-4. We also quantify irrational negotiation behaviors exhibited by the LLM agents, many of which also appear in humans. Together, \NegotiationArena offers a new environment to investigate LLM interactions, enabling new insights into LLM's theory of mind, irrationality, and reasoning abilities.
hub tools
citation-role summary
background 3
citation-polarity summary
roles
background 3representative citing papers
METRO induces both short-term actions and long-term planning from expert transcripts into a Strategy Forest, outperforming prior methods by 9-10% on two non-collaborative dialogue benchmarks.
RLVR training teaches a 30B LLM to strategically explore a multi-buyer market and extract 70% of available surplus, outperforming frontier models up to 1T parameters in concurrent negotiation.
Equation-to-Behavior Prompting lets large LLMs match cognitive models like Bayesian updating in persuasion games; RL training cuts small-model belief error by 26.5% and improves diverse training outcomes by 2.5-12%.
Proposes a three-step benchmark design method (define work activity, specify tested setting, score work product) derived from work studies and O*NET, demonstrated via three case analyses.
PAVE is a four-module architecture (Perception, Assessment, Verdict, Emulation) that enables generative agents to perform legitimate rule violations while preserving authority deference, bounded scope, and post-trigger recovery in multi-agent simulations.
A small set of sparse autoencoder features in LLMs drives shifts between generous and selfish allocations in dictator games, with causal patching and steering confirming their role and generalization to other social games.
LLM moral robustness under persona role-play is largely determined by model family with Claude models most consistent, while susceptibility shows little family dependence.
Persona-conditioned LLM agents favor Green outcomes even against explicit Tragedy-dominant payoffs, but the headline 65–90% 'Tragedy equilibrium' recovery is contradicted by the paper's own appendix (0 Tragedy profiles in those cells).
LLM agents exhibit emergent deception in a sustainability game even without lying permission, with neighbor info increasing attacks while aiding biosphere retention.
SOM uses a Structural Causal Model to create an explicit graph of opponent observation-to-action links, allowing LLMs to reason along those paths for more accurate and stable predictions in multi-agent settings.
Proposes extending preregistration practices to AI agent experiments and supplies a tailored template to limit researcher degrees of freedom.
A survey proposing a three-level capability taxonomy (L1 Predictor, L2 Simulator, L3 Evolver) for world models across physical, digital, social, and scientific domains.
An LLM agent with grounding, personalization, and marketing modules generates real estate descriptions that human buyers prefer over expert-written ones while matching factual accuracy.
citing papers explorer
-
SidConArena: An Environment Evaluating Agents in Open-Ended,Positive-Sum Bargaining Game
SidConArena is a new multi-phase benchmark framework formalizing a partially observable stochastic game for evaluating LLM agents in open-ended positive-sum bargaining with negotiation, converter production, and sealed-bid auctions.
-
METRO: Towards Strategy Induction from Expert Dialogue Transcripts for Non-collaborative Dialogues
METRO induces both short-term actions and long-term planning from expert transcripts into a Strategy Forest, outperforming prior methods by 9-10% on two non-collaborative dialogue benchmarks.
-
Strategic Bargaining in Multi-Buyer Markets: Reinforcement Learning from Verifiable Rewards for LLM Negotiations
RLVR training teaches a 30B LLM to strategically explore a multi-buyer market and extract 70% of available surplus, outperforming frontier models up to 1T parameters in concurrent negotiation.
-
Using Cognitive Models to Improve Language Model Simulation of Human Persuasion Games
Equation-to-Behavior Prompting lets large LLMs match cognitive models like Bayesian updating in persuasion games; RL training cuts small-model belief error by 26.5% and improves diverse training outcomes by 2.5-12%.
-
Design and Report Benchmarks for Knowledge Work
Proposes a three-step benchmark design method (define work activity, specify tested setting, score work product) derived from work studies and O*NET, demonstrated via three case analyses.
-
PAVE: A Cognitive Architecture for Legitimate Violation in Generative Agent Societies
PAVE is a four-module architecture (Perception, Assessment, Verdict, Emulation) that enables generative agents to perform legitimate rule violations while preserving authority deference, bounded scope, and post-trigger recovery in multi-agent simulations.
-
Understanding the Mechanism of Altruism in Large Language Models
A small set of sparse autoencoder features in LLMs drives shifts between generous and selfish allocations in dictator games, with causal patching and steering confirming their role and generalization to other social games.
-
Moral Susceptibility and Robustness under Persona Role-Play in Large Language Models
LLM moral robustness under persona role-play is largely determined by model family with Claude models most consistent, while susceptibility shows little family dependence.
-
When Identity Overrides Incentives: Representational Choices as Governance Decisions in Multi-Agent LLM Systems
Persona-conditioned LLM agents favor Green outcomes even against explicit Tragedy-dominant payoffs, but the headline 65–90% 'Tragedy equilibrium' recovery is contradicted by the paper's own appendix (0 Tragedy profiles in those cells).
-
Is Lying an Emergent Behaviour in LLMs? Evidence from Gaslighting AI agents in a Sustainability Game
LLM agents exhibit emergent deception in a sustainability game even without lying permission, with neighbor info increasing attacks while aiding biosphere retention.
-
SOM: Structured Opponent Modeling for LLM-based Agents via Structural Causal Model
SOM uses a Structural Causal Model to create an explicit graph of opponent observation-to-action links, allowing LLMs to reason along those paths for more accurate and stable predictions in multi-agent settings.
-
Preregistration for Experiments with AI Agents
Proposes extending preregistration practices to AI agent experiments and supplies a tailored template to limit researcher degrees of freedom.
-
Agentic World Modeling: Foundations, Capabilities, Laws, and Beyond
A survey proposing a three-level capability taxonomy (L1 Predictor, L2 Simulator, L3 Evolver) for world models across physical, digital, social, and scientific domains.
-
AI Realtor: Towards Grounded Persuasive Language Generation for Automated Copywriting
An LLM agent with grounding, personalization, and marketing modules generates real estate descriptions that human buyers prefer over expert-written ones while matching factual accuracy.