"AI isn't magic and it isn't a robot uprising — it's software that does tasks we used to think needed human intelligence."
Level: Beginner · Time: ~12 min · Prerequisites: none — this is your starting line
Learning Objectives
By the end of this module, you will be able to:
- Define artificial intelligence in plain language
- Tell AI, machine learning, and deep learning apart
- Recognize the main "approaches" to building AI
- Spot AI in the tools you use every day
1. What AI Really Means
Artificial Intelligence (AI) is the field of building computer systems that can do tasks normally associated with human intelligence — recognizing a face, understanding a sentence, making a recommendation, or winning a game.
Notice the word tasks. AI today is narrow: each system is brilliant at one thing (spotting spam, translating text) and clueless outside it. The all-knowing, human-like AI from movies — "general" AI, sometimes called AGI — doesn't exist yet, and no one knows for sure when (or whether) it will. A chess engine that crushes grandmasters cannot tell you why the sky is blue; a translation model that handles fifty languages has no idea what a chessboard is.
Explain like I'm new: AI is a very well-trained apprentice for one job. A spam filter is an apprentice who only sorts mail; it can't drive a car or write a poem. Each AI you meet is a specialist, not a genius-at-everything.
Concept: People often confuse automation with AI. A thermostat that switches on at a fixed temperature is automation — it follows a rule someone set. It becomes AI-flavored only when it learns your habits and predicts when you'll want heat. The dividing line is whether the system adapts from data or just follows fixed instructions.
2. AI vs. Machine Learning vs. Deep Learning
These three nest inside each other like measuring cups:
| Term | What it is |
|---|---|
| Artificial Intelligence | The whole goal: machines doing "smart" tasks |
| Machine Learning (ML) | A way to do AI: learn patterns from data instead of hand-written rules |
| Deep Learning | A kind of ML using brain-inspired neural networks |
So deep learning is a type of machine learning, and machine learning is one path to AI. You'll go deep on each in the coming modules. A quick way to keep them straight: if someone says "AI," they mean the goal; "ML," the method of learning from data; "deep learning," the specific technique using layered neural networks. All three words might describe the very same product — a voice assistant is an AI, built with ML, powered by deep learning.
Key idea: Modern AI mostly means machine learning — systems that learn from examples rather than following rules a programmer typed by hand. This shift, from "write the rules" to "show examples," is the single biggest idea in the field.
3. The Big Approaches to AI
Over the decades, people have tried several routes to make machines smart:
- Symbolic AI (rules & logic): encode human knowledge as explicit rules and facts. Great for clear domains; brittle when the world is messy. (Module 2)
- Machine learning (learn from data): let the computer find patterns in examples. Dominant today. (Modules 3–9)
- Search & optimization: explore many possibilities to find a good answer (how a chess engine "thinks ahead").
- Evolutionary & agent-based: mimic evolution or independent agents reacting to an environment. (Module 9)
Most real systems today blend these — mostly machine learning, with a dash of the others. Think of them as tools in a workshop rather than rival teams: a self-driving car uses learned models to see the road, search to plan a route, and hand-written rules to never run a red light. A good AI engineer picks the right tool for each part of the problem instead of forcing everything through one approach.
Real-world use case: Your phone's photo app uses deep learning to group faces, search to find photos by keyword, and simple rules to organize albums by date. One app, several AI approaches working together.
4. AI Is Already Everywhere
You almost certainly used AI today:
- Unlocking your phone with your face
- Autocomplete finishing your sentences
- Recommendations on Netflix, Spotify, and shops
- Maps predicting traffic and arrival times
- Spam filters quietly protecting your inbox
- Voice assistants turning speech into action
The goal of this track is to take you from using these systems to understanding and building them.
Hands-On: Try This
Try this: For the next hour, jot down every time you suspect AI is involved in an app or device. For each, guess which approach it uses — learned from data, or hand-written rules? This habit of "spotting the AI" will make every later module click faster.
Common Mistakes
Common mistake: Believing today's AI "understands" like a person. It finds statistical patterns extremely well, but it has no common sense or awareness. That's why it can be confidently wrong — a theme we'll return to often.
✅ Checkpoint
- Is deep learning a type of machine learning, or the other way around?
- What's the key difference between symbolic AI and machine learning?
- Name three places you meet AI daily.
Answers: 1) Deep learning is a type of ML. 2) Symbolic AI follows hand-written rules; ML learns patterns from data. 3) e.g., face unlock, recommendations, spam filters.
Key Takeaway: AI is software that performs tasks we associate with human intelligence — and today it's narrow (one task each), not general. Machine learning (learning from data) is the dominant approach, with deep learning a powerful subset. Other approaches — symbolic rules, search, evolution — still play supporting roles. You already rely on AI daily; now you'll learn to build it.
Further Learning
Adapted from Microsoft's AI for Beginners (MIT License). Sketchnote by Tomomi Imura.