Learning AI

How to learn AI for free: a beginner's roadmap

A realistic, free roadmap to learn AI from scratch: understand the ideas first, learn to use AI tools well, then choose whether to build. With the best free courses for each step.

5 min read

Short answer

Start by learning what AI is and how it learns, which needs no maths or code. Next, practise using AI tools well. Then choose a direction: using AI in your work, or building AI, which means learning Python and basics. Ten to twenty minutes a day is enough to start, and everything you need is free, from Elements of AI and Google's Machine Learning Crash Course to fast.ai and Hugging Face.

Step 1: understand the big ideas (one to four weeks)

Before any code, get a feel for what AI is, how machines learn from data, what do and how like ChatGPT work. This is the part most people skip, and it's why so many give up on their first technical course.

Step 2: learn to use AI tools well

Most people will get more from AI by using it well than by building it. Learn to write clear , give context, check answers and keep private data safe.

Practise on your real tasks: summarising a report, planning a trip, drafting an email. You'll learn faster than with made-up exercises.

Step 3: decide where you're heading

Here the road splits. Both are worthwhile.

  • AI user or professional. You want to use AI in marketing, teaching, law, healthcare or management. Go deeper on prompting, AI tools in your field, risks and ethics. You can stop at no-code.
  • AI builder. You want to make models or AI products: machine learning engineer, data scientist, AI developer. Carry on to step 4.

Step 4 (builders): Python, a little maths, then machine learning

  1. Python. It's the language of AI. Start with Kaggle's free Python course or the official Python tutorial. You need variables, loops, functions and lists, then the NumPy and pandas libraries.
  2. Just enough maths. Vectors and matrices, what a derivative is, and basic probability. Learn it alongside the code rather than first. Our guide do you need maths to learn AI? explains what matters and why.
  3. Machine learning. Google's Machine Learning Crash Course and Kaggle Learn are free and hands-on. Our Intermediate level covers the same ideas visually.
  4. Deep learning. Practical Deep Learning for Coders from fast.ai starts with working code, and Andrej Karpathy's Neural Networks: Zero to Hero builds a small GPT from scratch.
  5. Language models and agents. The free Hugging Face LLM course and agents course, with our Advanced level for the ideas.

Step 5: build small things

Courses give you knowledge; projects make it stick and show others what you can do. Good first projects are small and personal:

  • A classifier that sorts your own photos or emails.
  • A chatbot that answers questions about a document you care about, using RAG.
  • A Kaggle competition entry, even if it ranks low.

Share your work on GitHub with a short write-up of what you tried and what went wrong.

How long does it take to learn AI?

It depends on the goal. As a rough guide for someone studying a little each day: the core ideas take a few weeks, comfortable everyday use of AI tools takes a month or two of practice, and building machine learning models with Python takes several months. Job-ready skills usually take a year or more, and you keep learning after that, because the field moves fast.

Mistakes to avoid

  • Starting with heavy maths. Many people quit before seeing anything work. Learn the ideas, then the maths you need.
  • Collecting courses. Finishing one course beats starting five. Pick one per step.
  • Only watching. Pause and try things. Run the code, change it, break it.
  • Chasing every new model. The fundamentals change slowly. Learn them well and new releases make sense quickly.

Want a certificate along the way? See our guide to free AI courses with certificates, or the full list of the best free AI courses.

Key takeaways

  • Learn the ideas before the code: it makes every later course easier.
  • Using AI well is a real skill, and the quickest win for most people.
  • Builders need Python, some maths and a project, in that order of urgency.
  • Short, regular sessions beat weekend binges.

Learn it properly, free

Terms in this post

Questions

Can I learn AI without coding?

Yes. You can understand how AI works and use it well with no coding at all. You only need code, usually Python, if you want to build or train models yourself.

Can I learn AI on my own for free?

Yes. Some of the best AI courses are free, including Elements of AI, Google's Machine Learning Crash Course, fast.ai, Kaggle Learn, Hugging Face courses and AI in Bed. Many universities also put their lectures online for free.

How long does it take to learn AI?

A few weeks for the core ideas, a month or two to use AI tools confidently, several months to build machine learning models with Python, and usually a year or more to be job-ready as a builder. It varies a lot with time and background.

What is the best free AI course for beginners?

For no maths or code, Elements of AI from the University of Helsinki and the AI in Bed Beginner level are good places to start. If you can code a little, Google's Machine Learning Crash Course and Kaggle Learn are excellent free next steps.

Sources and further reading

Read next

Spotted a mistake? Suggest an edit