This site’s other research tutorials cover hands-on capture techniques — photogrammetry, Gaussian splatting. There’s a different foundational skill for anyone exploring creative AI and machine-learning research: running the code. An enormous amount of research and creative-ML work is shared as notebooks with the instruction “just run this in Colab,” which is baffling if you’ve never used it. How to Use Google Colab: Complete Notebook Tutorial for Beginners demystifies it — Colab is a free, browser-based environment with access to cloud GPUs, where you can run real machine-learning models without installing anything or owning a powerful computer.
What a notebook is, and why it matters
A notebook is a document that mixes runnable code, its output, and explanatory text in one place, broken into cells you execute one at a time. That format is why it dominates research and creative ML: you can run a step, see the result (an image, a number, a chart), adjust, and re-run, without executing an entire program at once. Colab is Google’s hosted version — the notebook runs on Google’s computers, not yours, so nothing installs locally. For an artist or researcher, that means opening a shared notebook and running a cutting-edge model is often as simple as pressing play on each cell in order.
The free cloud GPU is the real gift
The reason Colab matters for creative AI specifically is hardware. Running machine-learning models — image generators, style transfer, audio models — demands a powerful GPU that many people don’t own. Colab provides access to cloud GPUs for free (with usage limits, and paid tiers for more), so you can run models that your own laptop couldn’t handle. The tutorial covers the essentials: opening and running notebooks, the crucial step of enabling a GPU runtime, uploading your own files and data, and installing extra libraries a notebook needs. Those few skills unlock the vast library of shared creative-ML notebooks floating around the community.
Where to go next
Once you can open a notebook, switch on a GPU, and run cells, an enormous amount opens up: community notebooks for image generation and manipulation, research code released alongside papers, and your own experiments in Python. It’s the practical entry point to the whole machine-learning side of this site’s coverage — the models behind the AI tools, the techniques behind the research. Be mindful of session limits and save your work (Colab clears the machine when a session ends), and start by running someone else’s simple notebook end to end; understanding one is the key to all the rest.