At the recent TCEA Convention and Exposition, I learned about an experiment from Google called Quick, Draw!. Quick, Draw!, which launched last November, is a game built with machine learning. Players are prompted to sketch an object on a 20-second clock using the mouse or touchscreen. While the player doodles, the neural network throws out its best guesses of the subject, stopping mid-sketch if it guesses correctly.
How Quick, Draw Works
I gave it a try on my laptop and my iPhone. However, I have to admit, I’m not very good at it. My first time playing on the laptop, I only got two out of six. My iPhone try went a bit better, but that’s probably because the objects were fairly simple.
An interesting feature of Quick, Draw! is the insight players can gain about how the network works. After you complete a game, just select one of the drawings and you’ll see two sets of results. Let’s use my pathetic attempt at a bee as an example (try not to laugh). First, you see what kind of object the computer thought you were going to draw.
Scroll down and you’ll see the images that the neural network thinks a bee looks like.
Give it a try and see how many of your doodles can be recognized.
Classroom Uses of Quick, Draw!
So, I asked myself, how can this be used in the classroom? I was quickly reminded of the Quick Draw strategy, where students are asked to respond to a piece of text (literary or expository) through illustrations. Students are encouraged to quickly draw as many simple sketches as possible. The notion is for students to capture as many details as they can during the time period.
I can easily see Quick, Draw! being used with English Language Learners, especially those who are at the low-beginning to mid-beginning levels in writing English as their second language.
Some ideas for classroom application include:
- Have students summarize one of their illustrations, describing what they just drew through a voluntary share aloud or a Timed RoundRobin.
- Ask students to screen capture one of their images and write a story about it.
- Have students listen to the words the computer is guessing to see how many they recognize.
- For images that weren’t recognized, have students identify and define the images the machine thought they were drawing.
- During whole-class sharing, allow students to speak about one of their illustrations that couldn’t be guessed and have the entire class share details of how they would have drawn the image.
- Allow the students to explain some of their illustrations that might have special connections in their lives.
- Let students make predictions or inferences about their illustrations.
Quick, Draw! by Google is a fun and quick way for students to learn. If you are looking for more ways to incorporate drawing into your classroom, check out Drawing to Learn on the iPad for more lesson ideas.