Persona Following Is Not Selective Control: The Neutrality Gap in LLM User Simulation
cs.AI
Submitted: 2026-09-28
Updated: 2026-09-28
Sources
- Seed2.0 Model Card: Towards Intelligence Frontier for Real-World Complexity
- Persona Vectors: Monitoring and Controlling Character Traits in Language Models
- DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence
- Gemma 2: Improving Open Language Models at a Practical Size
- Gemma 3 Technical Report
- Gemma 4 Technical Report
- GLM-5: from Vibe Coding to Agentic Engineering
- The Llama 3 Herd of Models
- The Challenge of Using LLMs to Simulate Human Behavior: A Causal Inference Perspective
- Verbalizable Representations Form a Global Workspace in Language Models
- This human study did not involve human subjects: Validating LLM simulations as behavioral evidence
- Kimi K2.5: Visual Agentic Intelligence
- Statistical realism is not evidence that LLMs can estimate treatment effects in social science experiments
- The Illusion of Intervention: Your LLM-Simulated Experiment is an Observational Study
- Empirical Bernstein Bounds and Sample Variance Penalization
- Nemotron 3 Super: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning
- Forecasting Side Effects of Activation Steering
- gpt-oss-120b & gpt-oss-20b Model Card
- Qwen2.5 Technical Report
- Steering Language Models With Activation Engineering
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Agents' Last Exam
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems