Skip to the main content
PeabodyMake a free story

🧠 Learn how AI works · 💻 Coding · 🎓 Courses and learning platforms

fast.ai Practical Deep Learning

By fast.ai (Jeremy Howard). Visit fast.ai Practical Deep Learning

Desk research from the maker's own pages. Not yet tried by us. Checked 2026-10-24.

A free video course that teaches people who already code how to build and train deep learning models. Part 1 has nine lessons of about 90 minutes each, recorded at the University of Queensland; part 2 goes from foundations to Stable Diffusion. Learners write Python in free cloud notebooks on Kaggle or Paperspace, train image and text models, and publish small apps. A free book goes with it.

Our take: One of the best free routes into real AI for a strong teen coder; not for beginners.

Checked in a browser

A browser check on 2026-10-25: what loads, the ads and trackers we saw, and what a child meets first. Not a hands-on test.

Child safety: 4 out of 5No ads or chat; many links out to coding and AI sites (Kaggle, Hugging Face, OpenAI, Twitter, YouTube).
Data: 5 out of 5No ad, tracking or analytics networks; only GitHub and Google static file hosts, and no cookies.
Ease: 3 out of 5A long text page for adult coders; a 'Start watching lesson 1' link is clear, but it is not built for young children.
Cost honesty: 5 out of 5Free, with no upsell seen.

What we found

  • Contacted 3 third-party sites (githubusercontent.com, googleapis.com, gstatic.com); 0 cookies set.
  • No ads or trackers.
  • Nine lessons of about 90 minutes each; the course asks for some coding experience.
  • Links out to 22 sites, including Kaggle, Hugging Face, OpenAI, PyTorch and YouTube.

From the test: A free, ad-free and tracker-free deep learning course for teens and adults who already code.

The facts
Ages it suitsabout 16+ (needs a year of coding)
What its terms say about ageThe course site has no terms or age rule. The Kaggle notebooks it uses bar under-13s, and Kaggle says users 13 to 16 need verified parent consent before using it. (terms)
PriceFree. Free course and book; uses free tiers of Kaggle or Paperspace (Oct 2026).
Works onWeb
AccountNo account needed
AdsNone
In-app purchasesNone
The AI in itTeaches how to build, train and deploy AI models with PyTorch, fastai and Hugging Face.

What's good

  • Free, deep and hands-on
  • Only high school math needed
  • Real projects you can share

Watch for

  • Needs about a year of Python first
  • Part 1 was recorded in 2022, so some tools have moved on
  • Depends on third-party notebook sites with their own age rules

Data and privacy

The course site itself asks for no account. Kaggle (owned by Google) does not knowingly collect data from under-13s and requires parent consent for ages 13 to 16. Read the privacy policy.

Does it work?

No independent study found; the maker's own claims only.

In the Big Index

Alternatives to fast.ai Practical Deep Learning

  • Desk researchFreeNo ads

    AI Duet

    Yotam Mann with Google Magenta and Creative Lab · ages about 5 and up

    Worth a few minutes in a music or AI lesson; not a tool for learning piano.

  • Desk researchFreeNo ads

    AI Literacy Lessons for Grades K-12

    Common Sense Education · ages grades K-12 (AI-specific lessons start in grade 6)

    A good free starting point for teachers and parents who want to talk about AI with middle and high schoolers; anyone wanting hands-on AI building should pair it with a project-based course.

  • Desk researchFreeNo ads

    AI Unplugged

    Stefan Seegerer and Annabel Lindner (FAU Erlangen-Nürnberg) · ages about 10-18 (designed for lower and upper secondary; some activities adapt for primary)

    A strong free choice for teachers or homeschool parents who want kids to understand how AI learns without using AI tools; families wanting a ready-to-run app should look elsewhere.

  • Desk researchPaidNo ads

    Arduino UNO Q

    Arduino · ages about 13 and up with some coding experience

    A good next step for teens who already know Arduino and want to add AI vision; true beginners should start with a classic UNO or a starter kit.

  • Desk researchFreeNo ads

    AutoDraw

    Google Creative Lab · ages about 5 and up

    Handy for quick posters and class projects; skip it when the point is for kids to practice their own drawing.

  • Desk researchPaidNo ads

    BBC micro:bit

    Micro:bit Educational Foundation · ages about 8-14

    One of the best-value ways for 8- to 14-year-olds to learn coding and simple machine learning, with privacy-first tools; very young children will need help.

Every product in learn how ai works →

Where these facts come from

Prices, terms and features change. Nobody paid to be listed, nothing here is an affiliate link, and Peabody never rates itself. Where we and the maker disagree, the maker's own pages are the ones to trust.