Generative agents with memory streams, reflection, and planning using LLMs exhibit believable individual and emergent social behaviors in a simulated town.
hub
https://arxiv.org/abs/2301.07543
16 Pith papers cite this work. Polarity classification is still indexing.
16
Pith papers citing it
hub tools
citation-role summary
background 1
other 1
citation-polarity summary
representative citing papers
Expanding AI technologies in game-theoretic markets creates a 'Poisoned Apple' effect where agents release unused technologies to manipulate regulators into choosing market designs that benefit them at the expense of opponents and fairness.
LLMs display prompt-sensitive risk behavior and a linearly decodable realization-status signal in Gemma's residual stream, yet activation steering along this direction fails to shift downstream risk choices.
LLM agents make collective belief dynamics programmable, with simulations showing coordinated agents induce stable belief shifts, and four structural properties that complicate detection and defense.
A new battery of 30 cognitive tasks demonstrates that process-level behavioral features distinguish humans from frontier AI agents better than performance metrics (mean AUC 0.88), with process-specific fine-tuning improving mimicry but limited cross-task transfer.
DoubleAgents shows that a distributed-cognition design with coordination agent, dashboard, and policy module increases user comfort and reliance on AI agents for coordination tasks over time.
LLM embeddings enable strong retrodiction of masked GSS opinions via cross-validation and external validation but only modest performance on entirely unasked opinions.
A PMT-constrained LLM framework with A-TLM configuration outperforms classical imputation methods on RMSE and bias for block-wise missing disaster survey data.
Introduces the Mechanism Plausibility Scale, a four-level framework separating generative sufficiency from mechanistic plausibility in LLM-based agent-based models.
Causal localization via attribution and patching identifies a temporal preference subgraph in mid-to-upper layers of Qwen3-4B-Instruct-2507, with time-horizon geometry in the residual stream and initial evidence for steering-vector control.
Demographic-only LLM agents for retirement survey prediction exhibit central tendency bias, fail to reproduce incorrect or 'don't know' answers, and miss factor interactions in regressions, unlike survey-anchored agents.
Three independent LLMs exhibit correlated forecasting errors on 568 binary questions but human predictions show no activation of this shared bias.
AgentDynEx introduces nudging and a Configuration Matrix to help set up and maintain balanced mechanics and dynamics in multi-agent LLM simulations.
Proposes extending preregistration practices to AI agent experiments and supplies a tailored template to limit researcher degrees of freedom.
A three-dimensional decision matrix concludes that AI-simulated focus groups cannot replace human ones for observing emergent political meanings and identities but may be conditionally suitable for testing campaign messages depending on risk and grounding.
citing papers explorer
-
Generative Agents: Interactive Simulacra of Human Behavior
Generative agents with memory streams, reflection, and planning using LLMs exhibit believable individual and emergent social behaviors in a simulated town.
-
The Poisoned Apple Effect: Strategic Manipulation of Mediated Markets via Technology Expansion of AI Agents
Expanding AI technologies in game-theoretic markets creates a 'Poisoned Apple' effect where agents release unused technologies to manipulate regulators into choosing market designs that benefit them at the expense of opponents and fairness.
-
Representation Without Control: Testing the Realization Effect in Language Models
LLMs display prompt-sensitive risk behavior and a linearly decodable realization-status signal in Gemma's residual stream, yet activation steering along this direction fails to shift downstream risk choices.
-
LLM Agents Make Collective Belief Dynamics Programmable: Challenges and Research Directions
LLM agents make collective belief dynamics programmable, with simulations showing coordinated agents induce stable belief shifts, and four structural properties that complicate detection and defense.
-
Process Matters more than Output for Distinguishing Humans from Machines
A new battery of 30 cognitive tasks demonstrates that process-level behavioral features distinguish humans from frontier AI agents better than performance metrics (mean AUC 0.88), with process-specific fine-tuning improving mimicry but limited cross-task transfer.
-
DoubleAgents: Human-Agent Alignment in a Socially Embedded Workflow
DoubleAgents shows that a distributed-cognition design with coordination agent, dashboard, and policy module increases user comfort and reliance on AI agents for coordination tasks over time.
-
AI-Augmented Surveys: Leveraging Large Language Models and Surveys for Opinion Prediction
LLM embeddings enable strong retrodiction of masked GSS opinions via cross-validation and external validation but only modest performance on entirely unasked opinions.
-
Can Large Language Models Revolutionize Survey Research? Experiments with Disaster Preparedness Responses
A PMT-constrained LLM framework with A-TLM configuration outperforms classical imputation methods on RMSE and bias for block-wise missing disaster survey data.
-
Mechanism Plausibility in Generative Agent-Based Modeling
Introduces the Mechanism Plausibility Scale, a four-level framework separating generative sufficiency from mechanistic plausibility in LLM-based agent-based models.
-
Temporal Preference Concepts and their Functions in a Large Language Model
Causal localization via attribution and patching identifies a temporal preference subgraph in mid-to-upper layers of Qwen3-4B-Instruct-2507, with time-horizon geometry in the residual stream and initial evidence for steering-vector control.
-
From Demographics to Survey Anchors: Evaluating LLM Agents for Modeling Retirement Attitudes
Demographic-only LLM agents for retirement survey prediction exhibit central tendency bias, fail to reproduce incorrect or 'don't know' answers, and miss factor interactions in regressions, unlike survey-anchored agents.
-
The Oracle's Fingerprint: Correlated AI Forecasting Errors and the Limits of Bias Transmission
Three independent LLMs exhibit correlated forecasting errors on 568 binary questions but human predictions show no activation of this shared bias.
-
AgentDynEx: Nudging the Mechanics and Dynamics of Multi-Agent Simulations
AgentDynEx introduces nudging and a Configuration Matrix to help set up and maintain balanced mechanics and dynamics in multi-agent LLM simulations.
-
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.
-
Fake Plastic Voters: When Political Parties Can Use AI-Simulated Focus Groups
A three-dimensional decision matrix concludes that AI-simulated focus groups cannot replace human ones for observing emergent political meanings and identities but may be conditionally suitable for testing campaign messages depending on risk and grounding.
- When Agent Markets Arrive