中文

Zero to AI: Beginner Path

From Python basics to understanding LLMs — build the foundation first, skip the hype

No experience needed ~3–4 months (1 hour/day) 7 steps

Path goal

Finish able to handle basic data processing in Python, understand core machine learning and deep learning concepts, follow mainstream LLM discussions, and complete a small practice project on your own.

Who it's for：Complete beginners with no programming or AI background who want a systematic start

  1. Step 1: Master Python basics

    Pick one tutorial and finish it. Focus on variables, control flow, functions and file I/O — type the code yourself instead of just reading.

    💡 Pick just one: Liao Xuefeng's tutorial is the most systematic, and Harvard CS50's Python suits video learners.

  2. Step 2: Build a big-picture view of AI

    Use two survey courses to see what AI can and cannot do and where it applies — skip the technical details for now.

    💡 Watch AI for Everyone with Chinese subtitles; Elements of AI has an official Chinese version.

  3. Step 3: Build math intuition (optional but recommended)

    No need for systematic calculus — use animations to build intuition for derivatives, gradients, matrices and probability; models will feel much easier later.

    💡 3Blue1Brown has official Chinese subtitles on Bilibili; StatQuest explains statistics in short videos.

  4. Step 4: Machine learning intro

    Understand supervised and unsupervised learning, the train/validation/test mindset, and what overfitting means.

    💡 Google MLCC has interactive exercises; Li Hongyi's Chinese course is the friendliest for newcomers.

  5. Step 5: Deep learning intro

    Learn neural networks, backpropagation and CNN basics; complete a first full training loop with Dive into Deep Learning or fast.ai.

    💡 The d2l Chinese edition pairs formulas with code best; Nielsen's book is the gentlest for beginners.

  6. Step 6: Build a small project

    Use free compute on AI Studio or Kaggle to finish an entry competition or course project, putting the first five steps to work.

    💡 Size doesn't matter — a complete data → train → evaluate loop is enough.

  7. Step 7: Understand LLMs and follow the frontier

    Learn LLM and prompt engineering basics, and build the habit of checking weekly frontier news to stay sensitive to the field.

    💡 You've crossed the beginner stage — dive deeper via the App Dev, Agent Dev or AI Engineer path.

Browse all resources →