AI Periodic Table Explained: Mapping LLMs, RAG & AI Agent Frameworks

The video presents an innovative and highly effective pedagogical framework for understanding the AI technology stack. Martin Keen’s ‘AI Periodic Table’ is a brilliant conceptualization that organizes a chaotic field into a structured, memorable system. The use of chemistry analogies (groups, periods, reactivity) is not just a gimmick but a genuinely useful mental model that aids in understanding how different AI components relate and interact. The content is technically accurate, covering key concepts such as embeddings, vector databases, RAG, agents, and fine-tuning with clear definitions and examples. The progression from primitives to compositions to deployment to emerging technologies is logical and helps viewers grasp the evolutionary nature of AI systems. The demonstration of ‘reactions’ (RAG chatbot, agentic loop) effectively illustrates how the elements combine in real-world applications, making the framework actionable. The presenter’s expertise is evident, and the explanations are accessible without oversimplifying. However, the video is not without limitations. The periodic table is an original creation by the presenter, not an established standard, and it may omit or misclassify certain technologies. For instance, the placement of fine-tuning under ‘retrieval’ is debatable, as it is more about model adaptation than retrieval. Additionally, the ’emerging’ row is speculative and may quickly become outdated. The video also lacks citations to external sources, relying on the presenter’s authority and IBM’s credibility. Despite these minor issues, the video excels in its clarity, structure, and educational value. It successfully demystifies complex AI jargon and provides a valuable tool for both beginners and practitioners. The positive reception from viewers, as evidenced by comments, underscores its effectiveness. Overall, this is an excellent educational resource that offers a fresh perspective on AI architecture.

The title accurately reflects the content: the video explains an AI periodic table that maps LLMs, RAG, and AI agent frameworks.

8/10

The video presents a clear, structured framework for understanding AI components, grounded in established concepts (RAG, embeddings, agents). The presenter is an IBM Technology evangelist, and the content aligns with industry knowledge. However, the periodic table is an original conceptual model, not an official standard, and lacks peer-reviewed sources.

Markers derived by PSI from the transcript: the creator did not define chapters.

• 00:00Introduction: The chaos of AI terms and the need for structure.
• 00:45Introducing the AI Periodic Table concept and its layout.
• 02:00Explaining primitives: Prompts (Pr), Embeddings (Em), and LLMs (Lg).
• 04:00Row 2 compositions: Function calling (Fc), Vector databases (Vx), RAG (Rg), Guardrails (Gr), Multimodal models (Mm).
• 06:00Row 3 deployment: Agents (Ag), Fine-tuning (Ft), Frameworks (Fw), Red teaming (Rt), Small models (Sm).
• 08:00Row 4 emerging: Multi-agent systems (Ma), Synthetic data (Sy), Interpretability (In), Thinking models (Th).
• 10:00Demonstrating reactions: Building a RAG-based chatbot.
• 12:00Demonstrating the agentic loop with function calling.
• 14:00Challenge to viewers: Map any AI product to the table.
• 16:00Conclusion and call for comments.

• IBM watsonx Data Scientist - Associate certification — Mentioned in the video description as a promotion for IBM certification.
• AI Frameworks learning resources — Linked in the description as a resource to learn more about AI frameworks.
• IBM AI newsletter — Linked in the description for monthly AI updates from IBM.

• IBM watsonx.ai — IBM's AI platform, which aligns with the concepts discussed in the video.

• No direct discordant sources found — The video presents an original framework, so no direct contradictions with established sources were identified.