Revising Context, Shifting Simulated Stance: Auditing LLM-Based Stance Simulation in Online Discussions

SUMMARY

Research shows LLM stance simulations are highly sensitive to irrelevant context changes. Counterfactual framework reveals models shift predicted user positions based on semantically independent discussion elements. #llm #nlp #ai

Get research like this, matched to your field

Distill AI tracks arXiv, Nature, NeurIPS, CVPR, GitHub, HuggingFace and more — then surfaces the papers that matter to you, every morning. Track any custom topic, get 2-sentence summaries, and chat with any paper.