📜 [Paper] LLM User Simulator Audit Accepted at the NeurIPS 2026 UserSim Workshop (Non-archival)
🎉 Our paper, “From Next-Action Fidelity to Rollout Validity: A Human-Grounded Audit of LLM User Simulators in Repeated Social Interaction,” has been accepted at the NeurIPS 2026 Workshop on Grounded User Simulation for Model Evaluation and Training (UserSim). All three reviewers recommended acceptance.
The paper was written by Hyeongseo Jeon (first author, M.S. student) and Prof. Donghyun Lee (corresponding). It checks whether LLMs that stand in for people still behave like people over many rounds of social interaction. Congratulations, Hyeongseo! 🎊
Non-archival: UserSim is a non-archival workshop. The paper is not part of the NeurIPS proceedings, and the authors keep the right to publish a full version elsewhere. The workshop is held in Paris on December 12, 2026.
Citation: Jeon, H., & Lee, D. (2026). From next-action fidelity to rollout validity: A human-grounded audit of LLM user simulators in repeated social interaction. NeurIPS 2026 Workshop on Grounded User Simulation for Model Evaluation and Training (UserSim). Non-archival.