AI vs ML vs Deep Learning

Introduction

These three terms are often used interchangeably, but they're actually nested: Deep Learning ⊂ Machine Learning ⊂ Artificial Intelligence. Knowing where each sits stops you from over-engineering simple problems.

Beginner analogy: AI is the field of medicine. ML is the specialty of cardiology. Deep Learning is the technique of open-heart surgery. Each is more specialised than the last.

Understanding the topic

Core concepts to understand:

• AI — any technique that mimics intelligence (rules, search, ML).
• Machine Learning — algorithms that learn from data (regression, decision trees, SVMs).
• Deep Learning — ML using deep neural networks (CNNs, RNNs, transformers).
• Generative AI is a kind of deep learning — typically transformer-based.
• Pick the simplest tool: regex > rules > classical ML > deep learning > LLM.

Syntax reference

Visual workflow / architecture:

┌────────────── AI ──────────────────┐
│  rule-based · search · expert sys  │
│ ┌─────────── ML ────────────────┐  │
│ │ regression · trees · k-means  │  │
│ │ ┌───── Deep Learning ──────┐  │  │
│ │ │ CNN · RNN · Transformer  │  │  │
│ │ │ ┌──── Generative AI ───┐ │  │  │
│ │ │ │ GPT · Claude · SD    │ │  │  │
│ │ │ └──────────────────────┘ │  │  │
│ │ └──────────────────────────┘  │  │
│ └───────────────────────────────┘  │
└────────────────────────────────────┘

Real-world use

Spam filter? Classical ML (logistic regression). Image search in Google Photos? Deep Learning (CNN). ChatGPT? Generative deep learning (transformer LLM). Choose the right tool for the right job.

Best practices

• Don't reach for an LLM when regex or a small classifier solves it cheaper.
• Use deep learning when you have lots of data and unstructured inputs (images, text, audio).
• Benchmark classical ML baselines before assuming deep learning wins.

Hands-on exercise

Interview preparation — practice these questions:

• Q1. How are AI, ML and deep learning related?
• Q2. Give one production example of each.
• Q3. Why would you NOT use deep learning for a problem?
• Q4. Where does Generative AI fit in the hierarchy?