Exécution de notre outil de synthèse sur un nouvel article

Résumé : Large Language Models (LLMs) have achieved remarkable success in diverse natural language processing tasks, yet a critical challenge remains: the tendency to hallucinate or generate plausible but factually incorrect information. Retrieval-Augmented Generation (RAG) offers a promising approach by combining LLMs with external retrieval systems to ground generation in factual evidence. This paper reviews key developments in RAG architectures, analyzes empirical results demonstrating improved factual accuracy, and discusses technical challenges related to retrieval quality, latency, and adversarial robustness. We further identify open research areas such as scalable retrieval, multi-hop reasoning, and trustworthy human-AI collaboration. The survey synthesizes foundational research and practical insights, providing a comprehensive understanding of RAG’s role in advancing reliable AI-generated knowledge.