Module 1: Foundation
Introduction to Agentic AI Course
🔗 View the Excel Workbook Online
Program
In this lesson, I define what I mean by a computer program and why it’s important to distinguish it from AI. A program follows explicit instructions written by humans—if I tell the machine to “add 2” to every number, it will do exactly that, over and over, without deviation. Programming is fundamentally about encoding step-by-step instructions and having the machine execute them precisely as written.
AI
In this lesson, I explain how AI differs from traditional programs by focusing on pattern learning rather than instruction following. Instead of telling the system what rule to apply, I show how AI learns patterns from data and generalizes those patterns to new situations it has never seen before. This ability to generalize—something humans do naturally—is the key capability that turns computation into intelligence.
Discriminative AI
In this lesson, I revisit discriminative AI, the type of model I spent much of my early research career working on. I show how discriminative models take many input features and reduce them to a simple decision or label, such as classifying whether an image represents a rectangle. The core idea is learning a mapping from complex data to a clear outcome, with generalization allowing the model to handle slightly new or noisy inputs.
Generative AI
In this lesson, I contrast discriminative models with generative AI, which performs the reverse operation. Instead of collapsing many features into a single decision, generative models start from a small input—such as a seed or prompt—and expand it into rich, high-dimensional outputs like images or text. I show how generative AI can also generalize, producing plausible new outputs even for cases it has never explicitly seen during training.