新闻
SocietyBench: Forecasting Counterfactual Social-World Evolution
arXiv cs.AI · 发布于 · 阅读约3分钟
30秒读懂
- 发生了什么
- SocietyBench is a benchmark that measures how well large language models forecast social events by predicting outcomes from anonymized, counterfactual timelines built from real news and social media.
- 为何重要
- Matters for engineers building LLM agents that need to understand and predict real-world social dynamics beyond task completion like bug fixing or browser control.
- 注意
- Top models score only 75 out of 100 against a 50-point baseline. Performance splits into two independent axes: probability calibration and temporal accuracy, meaning strength in one does not guarantee the other.
收听本摘要
- agent
- llm
- language model
- distill
- benchmark
The Agent Architect
每周一个模式、一个权衡、一个生产事故案例。为构建智能体系统的人准备的每周简报。
每周一封邮件,一键退订。您的地址仅用于发送简报。