AI Basics
A friendly guide to the big ideas: what they mean, how they relate, and where you see them in real life.
Quick Jump
How these ideas stack
Think of it like nested boxes: AI is the umbrella, ML sits inside AI, DL sits inside ML, and GenAI often uses DL.
Quick Jump
Artificial Intelligence (AI)
AI is the umbrella term for making computers act “smart” in useful ways—like understanding language, spotting patterns, planning steps, or making decisions with incomplete information. Some AI uses hand-written rules, but most modern AI learns from data.
Key ideas
- AI is about behavior: can the system do the job well?
- It can be narrow (great at one task) rather than generally intelligent.
- Good AI is usually designed with clear goals and careful testing.
Friendly examples
- A camera that recognizes faces or objects.
- A spam filter that learns what looks suspicious.
- A map app that finds the fastest route.
Machine Learning (ML)
Machine Learning is how many AI systems improve with experience. Instead of telling the computer every rule, you show it lots of examples, and it learns patterns that help it predict outcomes or choose actions.
Key ideas
- You train a model on data, then use it on new inputs.
- It’s great when rules are messy or too many to write by hand.
- Quality matters: the model learns whatever your data teaches it (including mistakes).
Friendly examples
- Predicting if an email is spam or not.
- Estimating delivery time based on past deliveries.
- Recommending videos based on what you watched.
Deep Learning (DL)
Deep Learning is machine learning powered by neural networks with many layers. The “deep” part means the model can learn complex features automatically—especially helpful for images, speech, and language.
Key ideas
- Often needs more data and compute than simpler ML methods.
- Very strong at perception tasks (vision, audio) and modern text systems.
- Bigger isn’t always better—cost and latency matter in real products.
Friendly examples
- Photo apps that detect people, pets, or text in images.
- Speech-to-text dictation.
- Modern translation and summarization.
Generative AI (GenAI)
Generative AI creates new content—like text, images, or code—instead of only labeling or scoring existing data. It’s great for drafting, brainstorming, and transforming content, but it still needs human judgment for accuracy and safety.
Key ideas
- You prompt it with instructions, examples, or constraints.
- It can “hallucinate” (sound confident but be wrong).
- Best practice: treat outputs like a helpful draft, not a final answer.
Friendly examples
- Drafting an email, then you edit the tone.
- Generating a few UI copy options for a button label.
- Creating code snippets, then you review and test them.
Large Language Models (LLMs)
LLMs are deep learning models trained on huge amounts of text to generate language. Many use transformers and work by predicting the next token (a small chunk of text). That simple goal can produce surprisingly useful skills like explaining, summarizing, and writing code.
Key ideas
- Tokens: text is split into small pieces for the model to process.
- Next-token prediction: it generates one token at a time, based on context.
- Great at patterns and language—but not a perfect fact database.
Friendly examples
- Summarizing a long document into key points.
- Turning a rough idea into a clearer outline.
- Explaining code or suggesting a refactor (with review).
Where to go next
Now that the core terms make sense, explore practical examples and learn how people use AI at work and in everyday life.