Google gives away more AI training than most companies manage to sell. The catch is finding it. The company’s free courses sit scattered across Google Cloud Skills Boost, Kaggle, a developer site most people never open, and a Coursera catalogue where “free” carries a small asterisk.
Below is the map, plus an honest read on which of these free Google AI courses respect your time, which ones quietly want a credit card, and how to get through one before your enthusiasm runs out.
Where Google actually hides its free AI courses
Four places cover almost everything, and none of them link to each other in any obvious way. Bookmark these:
- Google Cloud Skills Boost — the home of the Generative AI learning path, plus dozens of short courses and skill badges on Vertex AI, Gemini, and BigQuery ML.
- Kaggle Learn — Google owns Kaggle, and its micro-courses run in the browser with zero setup. Python, pandas, machine learning, deep learning.
- Machine Learning Crash Course — roughly 15 hours of TensorFlow-based fundamentals sitting on developers.google.com, free since 2018.
- Google AI Studio docs and codelabs — less a course than a guided sandbox for Gemini, but it’s the fastest way to make an API call and see something happen.
Most people bounce between these tabs for twenty minutes, feel overwhelmed, and close all four. If that sounds familiar, the fix is picking one path and treating the rest as later reading. There’s a useful breakdown of where the good free AI classes are and how to actually finish one if the starting problem is really the finishing problem.
Start with the Generative AI learning path
This is the closest thing Google has to an official beginner track, and it’s genuinely free to work through. Ten courses, most of them 30 to 45 minutes long, ending in a short quiz and a skill badge you can stick on LinkedIn.
You’ll cover how large language models work, what responsible AI means in practice, how attention mechanisms function, and how to write prompts that don’t produce mush inside Vertex AI. None of it requires code. An evening gets you through the first three courses.
The honest limitations
It’s a tour, not a training programme. You’ll finish knowing the vocabulary and almost none of the mechanics. The hands-on labs that accompany some modules run on credits, which brings us to the money question a few sections down. If you want a second opinion on which generative AI courses are worth the evening, there’s a comparison of free generative AI courses that separates the useful from the email traps.
Kaggle Learn is the underrated one
Almost nobody mentions Kaggle when they talk about free Google AI courses, which is strange given the quality. Each micro-course takes three to five hours, runs entirely in a browser notebook, and issues a certificate on completion. No installs, no GPU setup, no credit card.
Intro to Machine Learning is seven lessons and builds up to submitting a real entry to the Titanic competition. Intermediate Machine Learning covers missing values, categorical variables, and pipelines. Intro to Deep Learning gets you training a Keras model by lesson three.
The pace is what makes it work. Each lesson is a short concept, then a code cell you edit and run. You learn by breaking things in a sandbox small enough to understand. If you want a wider shortlist of free machine learning courses that are worth your weekends, Kaggle usually sits near the top of that pile.
Machine Learning Crash Course: free, but not gentle
Google’s ML Crash Course is about 15 hours across more than 30 modules, with interactive Colab exercises throughout. It’s free, it’s well made, and it will quietly punish anyone who skipped the prerequisites.
You need comfort with Python and a tolerance for basic algebra. The course opens with linear regression and loss functions, not with “what is AI”. That’s a feature if you want depth and a wall if you don’t. Try the first two modules. If the notation makes you wince, spend a fortnight on Kaggle’s Python and pandas courses first, then come back. The course will still be there.
What “free” really means in Google’s catalogue
Here’s where the marketing and the checkout page drift apart slightly.
- Skill badges are not accredited certificates. They’re verifiable completion records. Useful on a CV, but they aren’t a qualification, and recruiters know the difference.
- Hands-on labs eat credits. Many Skills Boost labs draw on a monthly allowance. Some accounts get enough for a handful of labs; run dry and you pay per lab. The video courses and quizzes stay free.
- Google AI Essentials is not free. It’s $49 a month on Coursera unless you apply for financial aid, which is granted reasonably often. It’s a decent ten-hour primer, just budget for it.
- Kaggle certificates cost nothing and there’s no cap on how many you collect. That’s the most generous deal in the whole catalogue.
If certification is the actual goal, this guide to free artificial intelligence courses with certificates that are actually free will save you from signing up for something that asks for card details at step four.
A four-week schedule that survives contact with real life
Three evenings a week, 45 minutes each. That’s roughly nine hours a week, which is enough to finish without wrecking your sleep.
Week one: the entire Generative AI learning path. Front-load the vocabulary so the later material has something to attach to.
Week two: Kaggle’s Python course, then Intro to Machine Learning. Submit the Titanic entry even if your score is mediocre. Submitting is the point.
Week three: the first half of Machine Learning Crash Course. Stop at the point where you can explain overfitting to a friend without notes.
Week four: build one small thing. A Gemini call in AI Studio that summarises your own notes. A Colab notebook that charts a dataset from your job. Anything with your fingerprints on it.
The badge means nothing without the project
A skill badge sitting on a profile with no work behind it is a decoration. The people who get value out of free Google AI courses are the ones who finish a course and immediately break something with it, then fix it.
That’s where the last week matters more than the first three. Pick a task you already do badly or slowly, and automate a slice of it. Drafting replies, cleaning a spreadsheet, labelling a folder of images. Google’s free stack covers all three without a subscription, and the API is generous enough at low volume.
Once you’ve built one thing, the next step is usually code. The best free AI coding courses are the ones that end with something running, not a quiz score, and Google’s own codelabs sit comfortably in that category. Start there, ship one rough tool, and let the certificate be the side effect rather than the goal.