Starting and learning
Machine learning for kids
Machine learning lets a computer learn from examples instead of fixed instructions, like recognising a drawing of a cat, telling penguin species apart or sorting pictures into groups. Children can try it with free, visual tools, guided by a parent or teacher.
How kids can see machine learning at work
Guessing what you draw
You draw a picture and the computer guesses what it is, after learning from lots of drawings other people made. It is a fun way to see that the computer learned from examples.
Real exampleQuick, Draw!Google's Quick, Draw! asks a neural network to recognise your doodle and adds drawings to a large doodle dataset shared publicly for machine learning research.
Teaching a computer with your own examples
Children show the webcam a few examples of each thing, such as a thumbs up and a wave, and the computer learns to tell them apart. Then they test it and see where it makes mistakes.
Real exampleTeachable MachineGoogle's Teachable Machine lets anyone train a computer to recognise their own images, sounds or poses, with no expertise or coding required.
Building games that learn
Children collect examples, train a model and use it in a game or chatbot they build themselves. This shows that the model only knows what it was taught.
Real exampleMachine Learning for KidsAn educational tool that lets children train a computer to recognise text, pictures, numbers or sounds, then use it in tools such as Scratch.
Turning scribbles into pictures
The computer looks at a rough sketch, guesses what you are trying to draw and offers neater versions to choose from.
Real exampleAutoDrawGoogle's AutoDraw pairs machine learning with drawings from artists, suggesting finished drawings that match what you are sketching.
Can kids learn machine learning?
Yes. Children can try machine learning with visual tools, collecting examples and training models that recognise pictures or sounds, with an adult nearby. Older children who enjoy coding can move on to Scratch projects and then to Python.
- Play and noticeTry a game such as Quick, Draw! and talk about how the computer might be guessing.
- Train a model with examplesTeach a computer two or three categories in a browser tool, then test it with new examples it has not seen.
- Build something in ScratchUse the trained model in a Scratch game, which links machine learning to coding children may already know.
- Move to PythonTeenagers ready for typed code can learn Python and repeat the same ideas with real datasets.Learn it in Python
Helping kids go further with machine learning
If you already work in the field
Parents and teachers do not need to be experts. Work through a tool together, ask what examples the model learned from and why it sometimes gets things wrong, and let the child lead the project.
If you come from data science or ML
People who work in machine learning and want to teach children should keep sessions short and hands-on, start from examples rather than code, and treat mistakes and fairness as part of the activity. School code clubs and STEM outreach programmes are good places to start.
Roles that do this workComputing teacherSTEM outreach coordinatorMachine learning engineer
Tools you will use
| Tool | What you use it for |
|---|---|
| Machine Learning for Kids | Trains simple text, image and number models and uses them in Scratch projects, with teacher accounts available. |
| Teachable Machine | Teaches a computer to recognise poses, sounds or objects from webcam examples in the browser. |
| Scratch | Turns a trained model into a game or story using drag-and-drop code blocks. |
| Orange Data Mining | Shows a decision tree visually so children can see how the computer splits examples into groups. |
Public data to practise on
- Palmer Penguins: Measurements of three penguin species, a friendly table for making and testing sorting rules.
- Quick, Draw! dataset: Doodles drawn by players of Google's Quick, Draw! game, browsable online by category.
Your first project
With an adult, open Palmer Penguins, invent a rule such as a flipper-length cut-off to guess the species, then train a decision tree in Orange and compare how many held-out penguins each gets right.
Common questions
At what age can kids start learning machine learning?
Children can explore the idea with visual tools once they can follow simple instructions, usually with an adult. Typed coding in Python suits teenagers, or younger children who already enjoy coding.
Do kids need to know coding to learn machine learning?
No. Tools such as Teachable Machine need no code. Scratch, and later Python, let children use their models in projects of their own.
Is machine learning safe for kids to try?
Choose tools from known organisations, read each tool's privacy notes, especially for webcam or microphone input, and supervise younger children.
What do kids learn from machine learning activities?
That computers learn from examples, that more varied examples make better models and that models make mistakes. These ideas help them use and question AI tools sensibly.
Courses for these skills
- Deep Learning with PyTorchUnderstand nonlinear hidden layers and network training.Syllabus available · lessons open as produced
- Python for AI and Machine LearningBuild the basic programming skills used to handle image records.33 published lessons · all planned lessons open
- Practical Machine Learning in PythonStudy distances, clustering and evaluating unsupervised results.51 published lessons · more lessons in production
- Data Tools: NumPy, pandas, Matplotlib and scikit-learnInspect feature scales and representative rows.23 published lessons · all planned lessons open
- Maths for Machine LearningConnect vectors and linear boundaries to separability.26 published lessons · all planned lessons open
New to programming or maths? Start with Python, or follow the machine learning path in order. Published lessons are free to read; compare Free and Pro.