Webinar 2026 From Hallucination to Grounded Answers: An Introduction to Retrieval-Augmented Generation

Large language models can generate remarkably fluent answers, but they may also produce convincing misinformation or fabricate facts, especially when asked about documents they have never seen. For researchers, students, and professionals, an unsupported answer is often worse than no answer at all. This seminar introduces Retrieval-Augmented Generation (RAG), a practical approach that combines large language models with information retrieval to produce answers grounded in trusted sources rather than the model's memory. Through live demonstrations, we will show how RAG answers questions over local documents and research papers — and how the same pipeline extends to web pages and other sources. We will introduce the core ideas behind RAG (document chunking, embeddings, vector databases, and semantic retrieval), and discuss when RAG is the right tool compared with alternatives. The goal: turn a confident guesser into a research tool you can trust — and check.