Agent Smith
Research on "Agent Smith" (a placeholder name, as the provided papers don't refer to a specific entity named Agent Smith) focuses on developing autonomous AI agents capable of complex reasoning and interaction within various environments, leveraging large language models (LLMs) as their core decision-making component. Current research emphasizes improving agent capabilities through techniques like knowledge graph integration, multi-agent collaboration, and the incorporation of error-correction mechanisms, often within specialized frameworks designed for specific tasks (e.g., medical question answering, social simulation, or software engineering). This work is significant for advancing AI capabilities in complex domains and improving the reliability and safety of autonomous systems, with potential applications ranging from scientific research to healthcare and industrial automation.
Papers
ClinicalLab: Aligning Agents for Multi-Departmental Clinical Diagnostics in the Real World
Weixiang Yan, Haitian Liu, Tengxiao Wu, Qian Chen, Wen Wang, Haoyuan Chai, Jiayi Wang, Weishan Zhao, Yixin Zhang, Renjun Zhang, Li Zhu, Xuandong Zhao
From Single Agent to Multi-Agent: Improving Traffic Signal Control
Maksim Tislenko, Dmitrii Kisilev
Quantifying Misalignment Between Agents: Towards a Sociotechnical Understanding of Alignment
Aidan Kierans, Avijit Ghosh, Hananel Hazan, Shiri Dori-Hacohen
Aligning Agents like Large Language Models
Adam Jelley, Yuhan Cao, Dave Bignell, Sam Devlin, Tabish Rashid
Contrastive Sparse Autoencoders for Interpreting Planning of Chess-Playing Agents
Yoann Poupart