A Deep Dive into the Three Building Blocks of GenAI

On this page

Generative AI and Icebergs: What do they have in common?

The observable effects of Artificial Intelligence are just the tip of the iceberg. To fully grasp Generative AI, it’s essential to understand its three fundamental building blocks. Our recent webinar, “Beyond Limits: Transforming Member Experience with AI,” explored these generative AI building blocks using the iceberg analogy.

Diving deeper, the visible AI application layer sits atop the iceberg, serving as the interface for user interactions. Below this layer, the LLM (Large Language Model) layer and the Foundation Model layer form the core building blocks of gen AI. These components of generative AI are critical, functioning collaboratively to enable seamless understanding and generation of content across various formats, illustrating the comprehensive structure and functionality of Generative AI.

Foundation Model Layer: The Unseen Powerhouse

At the base of the Generative AI iceberg lies the Foundation Model layer. This layer is the result of extensive training on vast amounts of unstructured data in an unsupervised manner. Think of it as the bedrock upon which the entire generative process stands. These models serve as the robust foundation, enabling AI to generate, comprehend, and interact seamlessly across diverse formats and contexts. The significance of the Foundation Model cannot be overstated, as it forms the backbone of the AI’s capabilities.

LLM (Large Language Model) Layer: The Linguistic Intelligence

Moving up the layers, we encounter the Large Language Model (LLM) layer, situated in the middle of the generative iceberg. This layer represents the linguistic intelligence of Generative AI. Specifically designed for applications involving human interactions, content creation, and language understanding, LLMs play a pivotal role in sectors such as customer service, education, and content generation. The LLM layer is the bridge between the foundational model and the visible application, adding a critical linguistic dimension to the generative capabilities.

Application Layer: Unveiling the Visible Facade

Finally, emerging above the waterline is the Application layer. Here, Generative AI takes on tangible forms like ChatGPT, designed to crawl through the vast expanse of the World Wide Web. The Application layer utilizes the knowledge acquired from the Foundation and LLM layers to answer questions and perform tasks. ChatGPT, for instance, exemplifies the power of Generative AI in action, showcasing its ability to leverage vast knowledge to provide insightful responses and execute tasks.

Generative AI is a multi-layered process, with each layer playing a crucial role in shaping its capabilities. The Foundation Model layer provides the necessary strength, the LLM layer adds linguistic intelligence, and the Application layer brings AI into our daily lives. Understanding these three building blocks is paramount in appreciating the depth and sophistication behind the AI experiences we encounter. As we continue to explore the possibilities of Generative AI, it’s clear that its true potential lies beneath the surface, much like the submerged mass of an iceberg.

Scroll to Top

Let's Talk