What do you need…

1. Computer, laptop, tablet and/or smartphone.

2. Internet Connection

3. 10-20 minutes

Pro-tip: This activity works best with a tablet or touch screen because it involves sketching. Most youngsters will struggle to draw using a mouse. I recommend a tablet with stylist.

Today is all about participating a real machine learning experiment! The game itself is quite simple and you can probably just click the link and get going without the provided instructions. What we are really after here is a discussion (even a brief one) about what is actually going on with this game. Don’t forget to explore the data set after playing.

Goal: Children gain an increased understanding of Machine Learning practices and are excited about having participated in a real AI experiment.

Step 1. Play

Go to Quick, Draw and play a round or two. If you need help, follow the guide below to get started.

Tip: Have a little fun here, but not too much. We want to get to the rest of the activity while we still have our kiddos attention. We can always come back to this step for more fun later!

Step 2. Learn

Watch the video created by the team behind Quick, Draw to see how the AI researchers are using the game to teach the AI.

Step 3. Explore

Now that we really see what’s going on here. Take some time to explore the Data Set.

Kids will have a great time giggling over the thousands of silly drawings. Maybe you’ll even find some outliers to flag. They’ll love that they are helping the scientists train the AI!

Step 4. Discuss

Take some time to reflect on the activity. Use these questions to spark discussion:

1. "How do you think the computer learns to recognize what you're drawing?"

   • This question invites you to explore how the computer was trained. The answer lies in supervised learning: Quick, Draw! uses millions of drawings from people worldwide, each labeled with what it represents (like "cat" or "tree"). The computer learns patterns from these examples to recognize new sketches. Discussing this can reveal how AI mimics human learning through practice and examples!

2. "Why do you think the computer sometimes guesses wrong?"

   • This gets at the limitations of machine learning. The computer might guess incorrectly if your drawing style differs from the examples it was trained on or if it hasn’t seen enough similar sketches. It’s a great way to discuss how the AI needs diverse, representative data to improve—just like how we need varied practice to master a skill.

3. "If you draw something unusual, like a made-up animal, what do you think the computer might guess?"

   • This question explores how the AI generalizes from its training. Since Quick, Draw! matches your sketch to patterns it knows, a made-up animal might be guessed as something familiar (like a "dog" if it has four legs). It’s a fun way to discuss how the computer relies on what it’s already learned to make predictions about new things.

4. "How do you think the computer gets better at guessing over time?"

   • This opens a conversation about improvement in machine learning. While Quick, Draw!’s core model is pre-trained, the idea of adding new drawings to its dataset can make it smarter. It’s like how you get better at something with more practice—discussing this can highlight the role of data and feedback in AI.

5. "What do you think would happen if everyone started drawing things in a completely different way?"

   • This question digs into the importance of training data matching real-world use. If everyone drew differently from the examples Quick, Draw! was trained on, it might struggle to guess correctly. It’s a playful way to discuss how AI depends on seeing a wide variety of examples to work well for everyone.

Key Vocabulary Words and Definitions

1. Machine Learning

   • Definition: Machine learning is when computers learn to do tasks by looking at lots of examples, instead of being told exactly what to do.

   • Why It’s Important: In Quick, Draw!, the computer figures out how to recognize drawings—like a cat or a tree—by studying tons of examples from other players.

2. Neural Network

   • Definition: A neural network is a special kind of computer program that works a bit like a brain. It has many small parts connected together that help it learn and make decisions.

   • Why It’s Important: The neural network is the part of Quick, Draw! that looks at your drawing and guesses what it is. It’s trained to spot patterns, kind of like how your brain knows what a dog looks like.

3. Prediction

   • Definition: A prediction is a guess that the computer makes about something new, based on what it has learned from examples.

   • Why It’s Important: When you draw in Quick, Draw!, the computer predicts what you’re making—like a car or a house—by comparing it to drawings it’s seen before. It’s its best guess!