The AI Periodic Table: Your Mental Model
Just as chemistry's periodic table organizes elements into families with predictable properties, the AI Periodic Table organizes AI concepts into families and complexity levels. Once you understand this structure, you can decode any AI architecture, any product demo, any technical pitch.
The AI Periodic Table concept was presented by Martin Keen from IBM Technology. We've adapted and extended their framework to create Learn AI at Jahnel Group.
The Table
Rows = Complexity Levels
From atomic primitives to cutting-edge emerging technologies:
| Row | Name | Description |
|---|---|---|
| Row 1 | Primitives | The atomic building blocks that can't be broken down further |
| Row 2 | Compositions | Combinations of primitives that form practical tools |
| Row 3 | Deployment | Production-ready systems and practices |
| Row 4 | Emerging | Cutting-edge developments still rapidly evolving |
Groups (Columns) = Functional Families
Concepts that serve similar purposes at different complexity levels:
| Group | Family | Theme |
|---|---|---|
| G1 | Reactive | Action—from instructions to autonomous operation |
| G2 | Retrieval | Memory—storing, finding, and adapting information |
| G3 | Orchestration | Coordination—combining components into systems |
| G4 | Validation | Quality—ensuring systems work correctly and safely |
| G5 | Models | Intelligence—the raw capabilities that power everything |
How to Use This Table
-
Learn the families first. Understanding the five groups helps you categorize any new AI concept you encounter.
-
Follow the rows for complexity. Row 1 concepts are prerequisites for Row 2. Row 2 enables Row 3. And so on.
-
Notice the connections. RAG (Orchestration) depends on Embeddings and Vector DBs (Retrieval). Agents (Reactive) often use Function Calling (Reactive) and Frameworks (Orchestration).
-
Map real systems. When you see an AI product or architecture, identify which elements are in play. It demystifies the complexity.