Agentic AI Ecosystems: Navigating Cultural-Awareness, Biases and Misinformation in Multi-agent and Human-agent Interactions

  • Angana Borah, University of Michigan

In an era where artificial intelligence (AI) increasingly mediates human communication, understanding the dynamics of human-AI interaction is critical. This talk explores the potential of multi-agent AI systems in fostering inclusive, and culturally aware human-AI interactions. Drawing from three key research directions, we first delve into the power of multi-agent large multimodal models (LMMs) for cultural image captioning, showcasing how collaborative LMM systems can generate richer and nuanced interpretations of cultural artifacts. Next, we address the challenges of implicit biases in multi-agent large language model (LLM) interactions, presenting methods for detecting and mitigating biases to ensure fair and equitable outcomes. Finally, we examine the role of persuasion in the context of misinformation in demographic-aware human-LLM interactions, focusing on strategies to counteract harmful dynamics while preserving the benefits of AI-driven communication. Together, these ideas help create AI systems that boost human creativity and support fairness, inclusion, and trust.

MSRAfrica.AgenticAIEcosystems.2025Download

다음 볼만한 동영상