What is Artificial Intelligence?

Why This Matters

AI already decides what you watch, filters your email, unlocks your phone, and routes your commute. It is also rapidly changing how every profession works. Whether you write code, manage teams, teach students, or run a business, understanding what AI actually is (and is not) is the difference between using it well and being misled by hype or fear. Every other lesson in this curriculum builds on the mental model you form here.

Everyday Analogy

Think about how a child learns to recognize animals. You do not hand them a rulebook: "a cat has whiskers between 4 and 8 cm, ears shaped like triangles…" You simply point at many cats and say "cat." Big ones, small ones, fluffy ones, cats in shadow. Over time the child builds an internal sense of cat-ness and recognizes a cat they have never seen before.

AI systems learn the same way: not from rules, but from examples. Show a system a million labelled photos and it discovers the patterns itself. That single shift, from writing rules to learning from data, is the heart of modern AI.

What Is AI, Exactly?

Artificial Intelligence is the field of computer science focused on building systems that perform tasks which normally require human intelligence: recognizing images, understanding speech, translating languages, making decisions from incomplete information, and generating text or pictures.

The critical contrast is with traditional programming:

A spam filter built from rules breaks the moment spammers invent a new trick. A learned spam filter adapts, because it learned the statistical shape of spam, not a fixed checklist.

The Russian Dolls: AI ⊃ Machine Learning ⊃ Deep Learning

These three terms are used interchangeably in the media, but they nest inside each other:

• AI is the whole field: any technique that makes machines behave intelligently, including old-school rule systems.
• Machine Learning (ML) is the subset where systems learn from data instead of being explicitly programmed. Nearly all modern AI is ML (lesson AI-002).
• Deep Learning is the subset of ML that uses many-layered neural networks, the technology behind image recognition, speech, and chatbots like ChatGPT and Claude (lessons AI-003 and AI-004).

When someone says "the AI decided…", they almost always mean a machine-learned model made a statistical prediction.

Worked Example: AI You Used Today

Trace one ordinary morning:

1. Face unlock: a deep learning model compares your face to stored patterns (vision).
2. Email inbox: a learned classifier already removed ~100 spam messages you never saw.
3. Maps route: models predict traffic 20 minutes ahead from millions of phones.
4. "You might like…": a recommendation model matched your history against millions of users.
5. Autocomplete: a language model predicted your next word, the same core idea behind ChatGPT.

None of these use hand-written rules. All of them learned from data. You interacted with more AI systems before breakfast than existed in the world in 1990.

Real-World Showcase

• Netflix estimates its recommendation system saves over $1B per year by keeping subscribers engaged: pattern learning as core business.
• Medical imaging models, trained on hundreds of thousands of labelled images, now detect certain cancers in scans at accuracy matching specialist radiologists.
• DeepMind's AlphaFold predicted the 3D structure of nearly every known protein, a decades-old scientific grand challenge that data alone solved.

Common Misconceptions

• "AI thinks like a human." No: it finds statistical patterns. There is no understanding, intention, or consciousness inside (see AI-068 for the AGI debate).
• "AI is one technology." It is an umbrella covering many techniques, from decades-old decision trees to frontier language models.
• "AI is always right." Models are confident even when wrong, because they learned patterns, and patterns have exceptions (see AI-060 on hallucinations).
• "AI follows rules someone wrote." The defining feature of modern AI is precisely that nobody wrote the rules.

Try It Yourself

1. Count your AI touchpoints for one hour: unlocking your phone, every recommendation, every autocomplete, every photo enhancement. Most people find 10+ in an hour.
2. Explain AI to someone using the child-and-cats analogy, then check their understanding with one question: "So does the programmer write the rules?" If they answer "no, it learns them from examples," you both understand the core idea.
3. Open ChatGPT, Claude, or Gemini and ask: "Explain how you were trained, in three sentences." Compare the answer to this lesson.

Key Takeaways

• AI = systems performing tasks that normally require human intelligence.
• The defining shift: learning patterns from data instead of following hand-written rules.
• AI ⊃ Machine Learning ⊃ Deep Learning: nested subsets, not synonyms.
• You already use dozens of AI systems daily; it is infrastructure, not science fiction.
• AI predicts from patterns; it does not understand, and it is not always right.