Building a Multimodal RAG That Responds with Text, Images, and Tables from Sources | Towards Data Science

Why do few chatbots return figures from source documents in their responses?