The MAD Podcast with Matt Turck
Top AI Researcher on GPT 4.5, DeepSeek and Agentic RAG | Douwe Kiela, CEO, Contextual AI
Mar 6, 2025
Join Douwe Kiela, CEO of Contextual AI and co-author of the groundbreaking RAG paper, as he delves into the latest advancements in AI, including GPT-4.5 and the revolutionary DeepSeek. He shares insights into the evolution and architecture of Retrieval-Augmented Generation (RAG), explaining its transition to Agentic RAG and the significance of self-learning systems. The conversation also tackles the challenges enterprises face in deploying AI, balancing accuracy with inherent inaccuracies, and navigating the future landscape of contextual AI.
50:44
Episode guests
AI Summary
AI Chapters
Episode notes
Podcast summary created with Snipd AI
Quick takeaways
- The evolution of Retrieval-Augmented Generation (RAG) has established it as a crucial framework for enhancing AI accuracy and knowledge updates.
- Contextual AI's focus on agentic RAG emphasizes creating integrated models using synthetic data to improve decision-making in complex environments.
Deep dives
The Importance of Retrieval Augmented Generation (RAG)
Retrieval Augmented Generation (RAG) is a significant development in AI, allowing language models to work with data they were not originally trained on. This approach answers fundamental questions such as how to keep AI systems updated with current information without constant retraining. The evolution of RAG has made it the dominant paradigm in the AI space, as it enables models to generate accurate responses grounded in verified data sources, like Wikipedia. By integrating retrieval processes into generative models, RAG enhances accuracy, reduces the risk of hallucinations, and allows for seamless updates in knowledge.
