AI Learning Roadmap for Beginners
A clear roadmap for beginners learning modern AI, from foundations and LLMs to prompting, RAG, agents, AI coding, and building with AI.
What this guide gives you
A beginner AI roadmap should explain what to learn first, what to postpone, and how each topic connects to real use.
- The roadmap should move from concepts to use, then from use to systems.
- Prompting comes before RAG and agents because context is the basic control surface.
- AI coding is easier to learn after you understand agents, tool use, and verification.
A practical AI roadmap
A useful beginner roadmap is not a directory. It is an order of dependency. Learn the concepts that make later ideas less mysterious.
For LearnAI0, the order is: AI foundations, large language models, prompting and context, RAG and knowledge, AI agents, AI coding, and building with AI.
Stage 1: Foundations
Start by learning what AI is doing at a high level. Understand that a language model is not searching the web by default, does not truly know whether a statement is correct, and can produce fluent wrong answers.
This stage protects you from two common mistakes: trusting AI too much and dismissing it too quickly.
Stage 2: LLMs and prompting
Next, learn the basic behavior of large language models: tokens, training, inference, context windows, and hallucinations. Then learn prompting as context design.
Good prompting is not magic wording. It is the work of defining a role, goal, source material, constraints, examples, and checks.
Stage 3: RAG and agents
RAG matters when the model needs outside knowledge. It combines retrieval with generation so the AI can work with reference material instead of relying only on its internal parameters.
Agents matter when the AI needs to act across steps: choose a tool, observe a result, update the plan, and continue. This is powerful, but it introduces new failure modes.
Stage 4: AI coding and building
AI coding is not just asking for code. It is a workflow: define rules, keep context clean, ask for plans, run tests, inspect diffs, and verify behavior.
Once you understand the earlier stages, AI coding becomes less like guessing and more like supervising a capable but imperfect collaborator.
Related LearnAI0 lessons
Understand the language-model mental model.
What exactly is AI? Why can ChatGPT, DeepSeek, and Doubao chat with you? A plain-language explanation for total beginners: AI isn't really thinking — it's a 'super sentence-finishing machine' that has read an enormous amount of text. A 3-minute interactive demo to build your first correct intuition about AI.Base Model: The Token-by-Token Prediction MachineLearn what a base model is.
What is the core of large language models? The Base model learns statistical patterns of Token sequences from massive corpora. Each generation does one thing: predict the most probable next Token. Understanding this mechanism is where everything begins.RAG: Let the Model "Take an Open-Book Exam"See why retrieval helps with hallucination.
RAG (Retrieval-Augmented Generation) is the most widely used engineering solution to reduce LLM hallucinations. A deep dive into how RAG works: vector retrieval, context injection, and grounded generation.Agent: AI That Gets Things DoneStart the agent section.
What is an AI Agent? Four core capabilities — Plan / Tool / Memory / Act — rooted in Lilian Weng's profoundly influential architecture, each with step-by-step demonstrations in real-world scenarios.Why You Need to Set Rules for AIBegin the AI coding methodology section.
Vibe Coding Methodology Lesson 1: AI writes code fast, but without rules it easily reworks, misses edits, and breaks things. Rules are the most reliable way to inject context.