Skip to content

AI for Mere Mortals

Plain-language guides to understanding AI, written for people who want to use it thoughtfully and understand what's happening — including when running it on their own computers at home.

No math degree required. No hype. Just what things are, why they matter, and how to get useful help while checking the results.

Start with everyday AI

Getting Useful Help from AI covers clear requests, useful context, and improving an answer in small rounds. When Should You Trust an AI Answer? explains how to judge claims, check sources, and match verification to the task. Both apply to hosted chatbots and local models.

Start here on Windows

Your first local model on Windows walks through installation, a first answer, checks that it worked, and common fixes.

Windows validation results record tested versions, measurements, and remaining checks, with a repeatable script.

The guides

  1. How LLMs Work — what a language model actually is, in one page. Start here.
  2. Tokens — the unit everything is measured in: speed, memory, and cost.
  3. Reading a Hugging Face Model Page — what the numbers and labels mean: 7B, Q4_K_M, GGUF, context length, instruct.
  4. Choosing a Model — matching a model to your hardware and your task.
  5. Running Models at Home — RAM math, CPU vs GPU, and the settings that matter (Ollama-focused).
  6. Ollama, In Depth — every command, every useful setting, front ends and utilities, and common use cases.
  7. Open WebUI, In Depth — the private ChatGPT-style web page: install, documents and RAG, functions, users, and common setups.
  8. RAG and Embeddings — supplying relevant passages from your documents.
  9. Getting Useful Help from AI — goals, context, examples, and useful follow-ups.
  10. When Should You Trust an AI Answer? — evidence, citations, uncertainty, and checking what matters.
  11. Glossary — quick definitions, A to Z.

A note on language

AI has a habit of sounding complicated on purpose. These guides try to say the simple thing plainly. If a term appears before it's explained, check the glossary — and if something's still confusing, that's the guide's fault, not yours.

Every guide follows the same writing criteria: plain words, no hype, honest limits, and checkable steps.