Multi-agent AI Systems: From Simple Rules to Complex Problem Solving

Watch a 40-minute conference talk from Data Science Conference Europe 2022 exploring how multi-agent systems offer a compelling alternative to large language models in AI development. Discover why training increasingly larger models may have limited practical value and learn how breaking down problems into smaller, interconnected agents can lead to more effective solutions. Explore the advantages of multi-agent systems including improved prototyping capabilities, enhanced flexibility, better robustness, greater explainability, increased control, and easier integration. Understand how these systems can evolve from simple rules to complex models while remaining practical for real-world applications, making AI more accessible beyond just large tech companies with extensive resources. Presented by Livio Bencik in Belgrade, this talk challenges conventional thinking about AI development and presents a more distributed, manageable approach to solving complex problems.