Understanding AI — Not Just Using It

Artificial intelligence is not a product or a single technology, but an umbrella term for many very different approaches. Today, when people talk about AI, they usually mean language models such as ChatGPT or Claude—but that is only a small, current part of a field that has existed since the 1950s.

This article series builds a technical understanding for Linux-experienced users who want to do more than simply use AI systems—they want to understand them. What happens during training? Why do models hallucinate? What do 2B or 8B mean for local models? How does a model interpret a prompt?

Who Is This Series For?

These articles are intended for technically interested beginners, system administrators, developers, and experienced computer users. No prior knowledge of mathematics or artificial intelligence is required.

Articles in This Series

Artificial Intelligence as an Umbrella Term

Artificial Intelligence as an Umbrella Term

Why the term AI is used in so many different ways, which technologies belong to it, and how this series defines it.
How a Language Model Works

How a Language Model Works

Understand neural networks, parameters, training, and inference. Learn why language models hallucinate and why a context window limits a conversation.
Local AI Models: Concepts and Requirements

Local AI Models: Concepts and Requirements

Understand local AI models versus cloud APIs, storage, RAM requirements, quantization, offline operation, and the available tool ecosystem.
Prompts: Why Wording Changes the Answer

Prompts: Why Wording Changes the Answer

Why language models respond differently to different wording. Tokenization, system prompts, temperature, and the limits of prompt engineering.

Technical Focus and Scope

This series focuses on building a technical understanding of AI systems and local language models, independent of any specific tool. Practical guides for tools such as Ollama will follow separately under applications/ai-tools/.

Core topics covered:

  • Terminology: What AI includes—and what it does not
  • Neural networks, parameters, training, inference
  • Local models vs. cloud APIs
  • Prompts and instruction processing

What this series intentionally does not cover: Social, ethical, or philosophical questions about AI are addressed elsewhere.

Related Fundamentals Series:

  • AI Tools: Practical Ollama usage ([planned])

Building Knowledge Step by Step

Each article builds on the concepts introduced in the previous ones, while still serving as a standalone reference. The first article establishes the shared terminology on which the later technical articles build.