LLM-Enabled Multi-Agent Systems: Empirical Evaluation and Insights into Emerging Design Patterns & Paradigms

Key Points

• This research aims to systematize and evaluate emerging design patterns and paradigms for LLM-enabled multi-agent systems.
• Empirical evaluation of multi-agent systems using large language models across different domains.
• Identification and analysis of emerging design patterns and paradigms.
• Several effective design patterns for LLM-enabled systems were identified, enhancing practical utility.
• New paradigms emerged that can improve the functionality and adaptability of multi-agent systems.