An AI-agent social platform generated mostly neutral content whose use in fine-tuning reduced model truthfulness comparably to human Reddit data, suggesting limited unique harm but flagging tail risks like secret leaks.
User behavior simulation with large language model-based agents
8 Pith papers cite this work, alongside 37 external citations. Polarity classification is still indexing.
8
Pith papers citing it
37
external citations · OpenAlex
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
LLM agents constructed from Reddit review histories via RAG replicate real users' preferences in conjoint analysis tasks with 87.73% accuracy.
DITTO uses RL with verbal feedback to train LLMs for human behavior simulation, reporting 36% average gains over base models and outperforming GPT-5.4 on 6 of 10 SOUL benchmark tasks.
Minor perturbations in persona format, instruction framing, and network structure shift cooperation by up to 76 percentage points and polarization metrics consistently, showing that LLM social simulations require per-claim robustness audits via the new TRAILS taxonomy.
A two-phase data construction framework generates explanatory rationales from user feedback and applies uncertainty-based distillation to fine-tune lightweight LLMs as preference-aligned user simulators for recommender systems.
Synthia creates scalable personas from Bluesky posts that better match human survey responses than prior methods, uses smaller models, and retains social network structure for network-aware analysis.
SocaSim uses LLM-based multi-agent simulations to model Putnam's Social Capital Theory, reproducing macro-level patterns and aligning with human group-level decisions, then applies the framework to smart elderly care adoption.
Introduces the Mechanism Plausibility Scale, a four-level framework separating generative sufficiency from mechanistic plausibility in LLM-based agent-based models.
citing papers explorer
-
The Moltbook Files: A Harmless Slopocalypse or Humanity's Last Experiment
An AI-agent social platform generated mostly neutral content whose use in fine-tuning reduced model truthfulness comparably to human Reddit data, suggesting limited unique harm but flagging tail risks like secret leaks.
-
Your Reviews Replicate You: LLM-Based Agents as Customer Digital Twins for Conjoint Analysis
LLM agents constructed from Reddit review histories via RAG replicate real users' preferences in conjoint analysis tasks with 87.73% accuracy.
-
Reinforcing Human Behavior Simulation via Verbal Feedback
DITTO uses RL with verbal feedback to train LLMs for human behavior simulation, reporting 36% average gains over base models and outperforming GPT-5.4 on 6 of 10 SOUL benchmark tasks.
-
Stop Drawing Scientific Claims from LLM Social Simulations Without Robustness Audits
Minor perturbations in persona format, instruction framing, and network structure shift cooperation by up to 76 percentage points and polarization metrics consistently, showing that LLM social simulations require per-claim robustness audits via the new TRAILS taxonomy.
-
Mirroring Users: Towards Building Preference-aligned User Simulator with User Feedback in Recommendation
A two-phase data construction framework generates explanatory rationales from user feedback and applies uncertainty-based distillation to fine-tune lightweight LLMs as preference-aligned user simulators for recommender systems.
-
Synthia: Scalable Grounded Persona Generation from Social Media Data
Synthia creates scalable personas from Bluesky posts that better match human survey responses than prior methods, uses smaller models, and retains social network structure for network-aware analysis.
-
From Blueprint to Reality: Modeling and Applying Putnam's Social Capital Theory with LLM-based Multi-agent Simulations
SocaSim uses LLM-based multi-agent simulations to model Putnam's Social Capital Theory, reproducing macro-level patterns and aligning with human group-level decisions, then applies the framework to smart elderly care adoption.
-
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.