What is AI? — Beginner-Friendly Notes
1. Simple Definition (remember this)
Artificial Intelligence (AI) = making computers do things that normally need human intelligence — like seeing, understanding language, learning from experience, and making decisions.
If a task needs you to think (not just calculate), and a machine can do it — that is AI.
Beginner analogy: A calculator is not AI (it just follows fixed arithmetic rules). Google Maps suggesting a faster route by learning traffic patterns is AI (it learns and predicts).
2. What AI Is NOT
- Not magic, not a brain in a box — it is programs + data + mathematics.
- Not just robots — most AI has no body (search, recommendations, chat).
- Not always perfect — AI learns from data, so biased or small data → biased/wrong AI.
3. The 4 Ways People Define AI (Russell & Norvig)
Textbooks frame AI in a 2×2 grid. You only need to recognise it:
| Human-like | Rational (ideal) | |
|---|---|---|
| Think | Systems that think like humans (e.g., cognitive models) | Systems that think rationally (logic, reasoning) |
| Act | Systems that act like humans (Turing Test) | Systems that act rationally (agents achieving goals — modern view) |
Exam tip: The course takes the acting rationally / agent view as the main definition.
4. AI vs Related Terms
- AI (broadest): any intelligent behaviour.
- Machine Learning (ML): a subset of AI where the system learns from data instead of being hand-coded for every rule.
- Deep Learning: a subset of ML using neural networks with many layers.
AI ⊃ Machine Learning ⊃ Deep Learning
5. Core Capabilities You Will See All Semester
- Perception — sensors → understanding (vision, speech)
- Reasoning — logic, search, knowledge
- Learning — improving from data
- Action — agents acting in an environment
- Language & Interaction — NLP, dialogue, robotics
6. Quick Examples by Category
- Perception: face unlock, medical image diagnosis
- Reasoning: chess engine, route planning, puzzle solver
- Learning: spam filter, Netflix recommendations, price prediction
- Action: vacuum robot, self-driving car, game agent
- Language: translation, ChatGPT, voice assistants
7. How to Spot an AI Problem (3 checks)
- Environment is uncertain / large (too many possibilities to hard-code)
- Needs learning or search (pattern from data, or exploring options)
- Needs human-like judgement (classify, predict, decide, generate)
If all three are low, plain programming is enough — you don't need AI.
8. Common Mistakes Beginners Make
- Thinking AI = human consciousness — no, it is narrow task intelligence today.
- Thinking more data always helps — only clean, relevant data helps.
- Thinking AI writes its own goals — goals are still given by humans (PEAS in next topics).
9. One-Line Summary to Write in Exam
AI is the study and design of intelligent agents — systems that perceive their environment and act to maximise their chance of achieving goals, exhibiting capabilities such as learning, reasoning, perception and language understanding.
10. Quick Self-Check
- Is a rule
if temperature > 30 → turn on ACAI? (No — fixed rule, no learning/search) - Is email spam detection AI? (Yes — learns patterns from examples)
- Name one AI you used today and which capability it used.