Artificial Intelligence for Beginners: How LLM and AI Works

Four terms everyone uses, and how they fit inside each other like nested dolls.

The six steps that turned decades-old ideas into the assistants you use today.

A model is a file full of numbers. What those numbers are, and why bigger models need more memory.

Why making a model costs millions but using it costs almost nothing, and whether it learns from your chats.

The one small job a large language model repeats: guessing the next token, shown live.

What a token is, how text is split into tokens, and why prices and limits are counted in them.

Why an assistant seems to forget the start of a long chat, and how to work with that limit.

Why the same question gets different answers, and when to turn that variety up or down.

Open ChatGPT, send a first message, and see what the free plan gives you and what the paid plans add.

Open Claude, build a small tool right inside the chat, and compare the free and paid plans.

Open Gemini, try a first chat, and see what the free plan gives you and what the paid plans add.

Open Copilot, send a first message, and see how it helps inside Word, Excel and PowerPoint.

Meet two assistants that live inside social apps, and see what each one gives you for free and what costs extra.

Search the web with an assistant that writes one short answer and shows clickable sources you can check.

Try three open-model assistants, see what each does best, and learn where your data goes when you use them.

Meet the assistant from Europe for everyday work and coding, and learn what it costs and what to watch for.

Put every assistant plan side by side, from free to top tier, and decide whether you really need to pay.

Pick the right assistant by task: writing, research, office work, coding, images and privacy.

Take a chatbot apart into five parts, so you can sit down at any assistant and feel at home.

Talk to your assistant out loud, and point your phone camera at the world to ask what you see.

Describe a picture in words, get an image back, and learn how to edit it with follow-up requests.

Make a short video clip from one sentence, and see what the free daily credits really buy.

Hand a big question to an assistant, get back a report with sources, and learn when to trust it.

Chat with your own PDFs and files, and learn what the word RAG means in plain words.

See what an assistant remembers about you, and how to view, edit, pause or delete it.

Let an assistant reach your calendar, files and email through connectors, while you stay in charge.

See how an agent plans, acts, checks and repeats until the job is done, and where it still needs you.

Try AI that clicks and types in your browser, and learn five simple rules to use it safely.

Build a notebook from your own material, get answers with citations from those sources only, and turn them into audio, video and slides.

Run your first prompt in Google's test bench, shape the answers with settings, and create a free API key.

Turn one sentence into a short video with the free daily credits, step by step.

Try Microsoft Copilot and Perplexity on the same question, and see which one fits which job.

Turn a paragraph of text into a clear diagram in seconds, with no drawing skills.

Describe a presentation in one sentence, edit the first draft in Canva, and meet Gamma for slides.

Make a song with Suno and a spoken sentence with ElevenLabs, and learn what the free plans allow.

Make your first AI images with a fresh free daily allowance and no credit card.

Install a free design app on your computer and make your first design with your own coding agent.

One page that matches every free tool in this section to the job it does best.

See how a model learns language by guessing the next token across a huge amount of text.

Learn how a pretrained model becomes a helpful assistant, and the LoRA trick that makes tuning cheaper.

Meet RLHF and RLVR, the two ways feedback teaches a model what a good answer looks like.

See what a model does when it thinks before answering, what the effort setting means, and why thinking costs more.

Learn how a huge model can wake only a few experts per token, and how one model reads text, photos and audio.

See how a small model learns from a big one, and why that lets AI fit on a phone or a laptop.

Learn the difference between a closed model and an open-weight one you can download and run yourself.

Meet the nine model families you will hear about most, and learn to read the size in a model name.

Open Hugging Face, find a model, and learn to read its page, step by step.

Learn why one model comes in different file sizes, and what letters like Q4 in a name mean.

Read the most common model licenses in plain words, and learn what you may do with each one.

A quick tour of the open-source AI projects everyone mentions, and what each one is for.

Set up an assistant that lives in your chat apps and uses tools on your computer, then learn the safety rules.

Build a small n8n workflow with an AI step that turns rough notes into a polite email.

See the real reasons to run AI on your own computer, and the one price you pay for it.

Download Ollama, install it, and have your first chat with a model running on your own computer.

Install LM Studio, find a model in its built-in catalog, and chat, all in one app.

Get a full chat window with history and document upload on top of your own models, using Open WebUI.

Get answers that are working calculators, charts and 3D models instead of plain text.

Set up AnythingLLM and ask questions about your own document, privately on your computer.

Learn to measure your AI's speed, what each speed feels like, and what really decides it.

See why a long chat uses more memory, what the KV cache is, and a trick that shrinks it.

Solve a speed puzzle and learn how draft tokens let a model write several tokens in one step.

See what happens when a model is too big for your memory: no error, just a big slowdown, and how to avoid it.

Learn the one rule that decides whether a model runs on your computer: it has to fit in memory.

See what an 8 GB card or laptop can really run, and which model sizes fit.

See why 12 and 16 GB are the sweet spot, and which model sizes they run comfortably.

See what 24 GB can run, and the used-card route to get there.

See what the extra memory at 32 GB buys you: room for very long chats and a higher-quality version of a model.

Learn that AMD and Intel cards run AI too, what they offer for the money, and what to watch for.

Learn to read a gaming laptop by its memory, not by its name, before you spend real money.

Learn how unified memory lets a MacBook run AI without a graphics card, and what to expect from the Air and the Pro.

Turn the whole section into a simple buying guide, starting with what you already own.

Learn how unified memory lets a small box load models that a graphics card cannot even start.

See what a 70B model needs, and the three routes to 48 to 64 GB of memory.

Meet NVIDIA's DGX Spark and its partner versions, and see what the box is good at.

See how ordinary Windows machines with AMD Ryzen AI Max chips reach big memory.

Compare the Mac mini and the Mac Studio, and see which big models each one can hold.

See how an ordinary gaming PC runs a 180 billion parameter model that asks for datacenter hardware.

Turn the whole section into one decision: pick the big-memory machine that fits your model size and your patience.