Fadé Adeniyi
Friday 30 October 2026
Four free courses taught by the people who shaped modern AI, which one to start with for your level and three prompts that turn watching into understanding.
Andrew Ng, Andrej Karpathy, Grant Sanderson of 3Blue1Brown and Jeremy Howard of fast.ai all teach AI for free online. If you don't code, start with Andrew Ng's AI For Everyone. If you can code a little, start with Karpathy. Use 3Blue1Brown to understand the maths visually.
At Techlerator we build Xeruit, an AI talent platform, so I spend a lot of time around people deciding what AI skills matter. The ones who learn fastest go straight to the source rather than to whoever repackaged it last week.
Not sure where you fit? Ask Claude to choose for you. Tell it the truth about your coding level.
I want to learn AI from one of these four free courses: 1. Andrej Karpathy's Neural Networks: Zero to Hero 2. Andrew Ng's AI For Everyone 3. 3Blue1Brown's neural networks videos 4. fast.ai's Practical Deep Learning About me: I work as [YOUR JOB]. My coding level is [NONE / A LITTLE / CONFIDENT]. I want to learn AI so I can [YOUR GOAL]. Tell me which one to start with and why, in three sentences. Then tell me which one to do second.
AI For Everyone, by Andrew Ng. The best first course for understanding what AI can do for a business, with no code at all. It's free to audit on Coursera. The certificate costs extra. More than 2.6 million people have taken it. Open the course.
Neural networks, by 3Blue1Brown. Grant Sanderson's animated series is the clearest explanation of how a neural network learns that I've seen. Start with chapter 1. 3blue1brown.com.
Neural Networks: Zero to Hero, by Andrej Karpathy. An OpenAI co-founder builds a GPT from scratch in plain Python and explains every line. It's free on YouTube. karpathy.ai/zero-to-hero.
Practical Deep Learning, by fast.ai. Jeremy Howard's course is for people with some coding experience who want to build working models quickly. Free and self-paced. course.fast.ai.
Every one of these courses has a moment where it loses you. Don't skip it and don't start again from the beginning. Ask Claude to explain that one part in your language.
I'm watching [COURSE AND LESSON NAME]. I didn't understand this part: [PASTE THE BIT OR DESCRIBE IT] Explain it again using an example from my world. I work in [YOUR INDUSTRY]. Keep it under 150 words, then ask me one question to check I've got it.
Use your own industry. An example from your world sticks far better than the course's example.
Here's the transcript of a lesson I've just watched: [PASTE THE TRANSCRIPT] Turn it into one page of notes: the five main ideas in plain English, any words I need to know with a one-line meaning each and one thing I could try this week to use it.
I don't sit through these start to finish. I pick one lesson, put it on while I'm doing something light and then ask Claude to explain the part I didn't get using an example from my own business. Andrew Ng's course is the one I send founders to first, because it's about what AI can do for a company, not how to code it.
Starting all four at once. You'll finish none of them. Pick one, finish it and only then start the next.
You've got four of the best free AI teachers, a way to pick your starting point and three prompts that make the lessons stick.
What this won't do: watching won't make you an AI engineer on its own. Coursera's free audit doesn't include the certificate. Course pages can change too, so check each one before you start.
Learn from the source. Then build something with it.
Related guides: Stop paying for AI certificates. Get these 4 free., Stop watching AI tutorials. Start building. and The free AI certifications worth it.
Sources, opened 30 October 2026: karpathy.ai; Coursera, AI For Everyone; 3Blue1Brown; fast.ai. Screenshots are the course pages.
The Audacity to Build is my community for people doing exactly that. It is opening soon. Join the waitlist and you are first in when it does.
If this was useful, follow me. I share what I am actually building, the tools that survive my own testing, and the ones I quietly cancel. New plays every week.
Pop in your name and email and the guide is yours, along with the occasional new guide when I publish one. No spam, ever.