12 weekend AI projects for families

Take photos of the dog, the cat, and the empty couch, and train a model that knows who is on the sofa. Kids learn why more and more varied photos give a better model.

1. Take 30+ photos of each pet in different spots and light.
2. Make one class per pet plus a nobody class.
3. Train and test with new photos.
4. Try a stuffed animal and see what it guesses.

Use Scratch's ask-and-answer blocks to build a character that replies to what you type with if-then rules. Kids discover how early chatbots worked and why today's ones needed huge amounts of text instead.

1. Pick a character sprite.
2. Use ask and wait, then if-then blocks for a few keywords.
3. Add a fallback reply for anything else.
4. Have a family member try to stump it.

Watch for: Kids under 13 need a parent's email to create a Scratch account; projects can be shared publicly, so check settings.

Train a model on words or pictures, then use it inside Scratch, for example a character that reacts to compliments and insults. Dozens of guided projects with worksheets make it easy to follow at home.

Watch for: Sign-up options vary; a parent or teacher sets up the account. Read the site's getting-started notes first.

Photograph your child's drawing of a person or creature and the tool makes it dance, jump or wave. Kids stay the artist; the AI only adds motion. Great for a family movie night.

1. Child draws a figure with arms and legs apart.
2. Parent uploads a photo of it.
3. Adjust the joints the tool finds.
4. Pick a dance and save the video.

Watch for: You upload a photo to Meta. Read the consent choice about letting them use drawings for research.

Train an image model in the browser, then drop it into an MIT App Inventor app that runs on your phone. It is a real, shareable app, and teens see the whole path from data to product.

1. Train a classifier on a few objects.
2. Export the model.
3. Build a simple App Inventor screen with the camera.
4. Test on the phone and fix weak classes.

Sit outside for 20 minutes while Merlin's sound ID suggests which birds are singing. Make a tally chart. Kids practice checking an AI's guess with their own eyes and a field guide.

Watch for: Uses mic and location while running.

Zooniverse projects ask volunteers to label photos of animals, galaxies or old documents. Many of those labels train research AI. Kids see that behind every model are lots of careful human labels.

Watch for: Pick a project together; some images (like camera traps) can show predators and prey.

Code a dance video with blocks, then use the AI feature to describe a mood and let it pick moves and backgrounds. Then change what it chose. A playful way to show AI as a helper you edit.

Step-by-step projects from the Raspberry Pi Foundation walk kids through building small machine learning projects. Clear instructions mean a parent with no coding background can follow along.

Shoot a short film with toys or clay, frame by frame. Afterward, talk about how AI video tools create frames too, and why a handmade film shows the maker's choices in every frame.

Watch for: In-app purchases for extra features.

Three short videos with family guides, conversation cards and hands-on activities. Do all three over a weekend: what AI is, privacy and fairness, and AI for schoolwork. Available in several languages.

Browse thousands of real doodles people drew for one word, like 'cat', then each draw your own and see how yours compares. Kids see the training data an AI learned from, including its odd and surprising examples.