What Is AI? A Beginner's Map of the Landscape

AI, ML, and generative AI — what's the difference?

Every few months brings a new buzzword, so let's pin down the core ideas first. Three terms get used interchangeably in job listings and conference talks, but they describe genuinely different things, and knowing which is which makes the rest of the field much easier to navigate.

Artificial intelligence is the broad field of building systems that perform tasks that normally require human intelligence — understanding language, recognizing images, making decisions, navigating a room. As a field it goes back to the 1950s, and it includes plenty of approaches that are not machine learning at all. A program that follows a fixed decision tree written by a human is AI. So is a chess engine. Neither learns anything from data.

Machine learning is the subfield where systems learn from data instead of following hand-written rules. Instead of a human encoding "if the email contains these words, mark it spam", you show the system thousands of labelled emails and it tunes itself until it can classify new ones. The rules are never written — they emerge from the data.

Generative AI is the slice of ML that creates rather than classifies. It produces text, images, code, or audio. Large Language Models, the ones behind chatbots like ChatGPT or Claude, are generative models trained on enormous amounts of text. Generative AI is a recent arrival: the field existed for decades but only became broadly useful once models, datasets, and compute grew large enough together.

The relationship, in one picture

Think of it as nested circles. AI contains ML. ML contains generative AI. Generative AI is not a replacement for the others — a spam filter is machine learning but not generative AI, and it is still a complete and legitimate product.

A useful second framing is discriminative versus generative. Discriminative models answer "which class is this?" — spam or not spam, cat or dog, approve or decline. Generative models answer "what would this look like?" — they model the distribution of the data and can produce new samples from it. Most classical ML is discriminative. All of the AI that writes code or answers questions is generative.

Where AI sits in an engineering stack

When you see an AI feature in a product, it is almost never the whole system. It is one component with ordinary software on either side.

• Data layer — where the information lives, and how it gets cleaned and structured.
• Model layer — the ML or LLM component doing the reasoning or generation.
• Application layer — the API, interface, and business logic around it.
• Evaluation and monitoring — how anyone knows whether it still works.

This matters because the hardest problems in production AI are rarely the model itself. They are the data, the retrieval, the latency, the cost, and the fact that the system must be trusted to make decisions. Engineering students who can work confidently across all four layers are the ones who get hired.

Why this matters for students

• AI is a tool, not a magic wand — it fails without clear problems and clean data.
• Every engineering field now touches AI, so basics are a career edge rather than a niche.
• Job titles change fast; durable skills (math, data, debugging) age well.
• The gap between "I have used ChatGPT" and "I can build with AI" is the entire job.

Start with the map above, then learn by building small projects. One working chatbot or classifier teaches you more than ten theory lectures, because the constraints you hit while building are the constraints you will discuss in interviews.

A concrete first project

Build a text classifier before you build a chatbot. Take a labelled dataset, train a simple model, and then deliberately feed it bad input to see how it fails. Understanding failure modes is what separates someone who has used an API from someone who can engineer a system.

Go deeper

• See how large language models really work for the mechanics behind chatbots.
• Explore our AI and engineering training tracks to turn basics into a career path.
• Get a week-by-week plan with the Zero to AI Engineer roadmap.