LLM Agent
LLM agents are autonomous systems that combine large language models with the ability to interact with their environment, achieving complex tasks through planning, tool use, and iterative refinement. Current research focuses on improving their reliability and safety, including aligning their behavior with human values, enhancing their decision-making processes (e.g., through Q-value models and tree search algorithms), and mitigating vulnerabilities to adversarial attacks. This field is significant because it bridges the gap between theoretical AI and practical applications, impacting diverse areas such as game development, software testing, healthcare, and financial markets by automating tasks and improving decision-making.
Papers
DAWN: Designing Distributed Agents in a Worldwide Network
Zahra Aminiranjbar, Jianan Tang, Qiudan Wang, Shubha Pant, Mahesh Viswanathan
AgentHarm: A Benchmark for Measuring Harmfulness of LLM Agents
Maksym Andriushchenko, Alexandra Souly, Mateusz Dziemian, Derek Duenas, Maxwell Lin, Justin Wang, Dan Hendrycks, Andy Zou, Zico Kolter, Matt Fredrikson, Eric Winsor, Jerome Wynne, Yarin Gal, Xander Davies
OpenCity: A Scalable Platform to Simulate Urban Activities with Massive LLM Agents
Yuwei Yan, Qingbin Zeng, Zhiheng Zheng, Jingzhe Yuan, Jie Feng, Jun Zhang, Fengli Xu, Yong Li