Google AI game for parents and children to learn together - Quick, Draw! -AI4kids

The AI ​​craze is on the rise. How can we explain the basic concepts of AI to children? You can start with the AI ​​experimental game developed by Google - Quick, Draw, and use the game to help your children understand how AI artificial intelligence learns and the principles behind it.

Click on the Quick, Draw game page ( https://quickdraw.withgoogle.com/# ). Although it is an English interface, you can easily start the game. It is just like using hands and feet, but you use drawing! There are six questions in total, and each question has 20 seconds. You have to draw the specified items, which are usually objects and animals you see in life. The computer will guess what you are drawing while you draw. You can compete to see who can make the computer understand you better!

While playing the game, you can observe how the computer understands you and your child's doodles. Although the computer seems to be making random guesses, the principle behind it is actually guessing through a machine learning algorithm. Each game is allowing the game to learn to guess, making the guesses more accurate. Just like people, the more you play gestures, the more you will understand each other's way of conveying messages, the better your tacit understanding will be, and the faster you will guess.

You can watch the video below to explain how the Quick, Draw game works:

https://www.youtube.com/watch?v=X8v1GWzZYJ4

After playing the Quick, Draw game, you can discuss with your children and think about how this game applies new AI technology? The technology it relies on is largely similar to that of Google Translate. Google Translate and the keyboard provide handwriting input functions. How to make the machine recognize your handwriting is not only about comparing the font size, but also the order of the strokes will affect the judgment result. Learning to recognize handwriting and images requires a large database for computer learning and training. The more diverse the data, the more accurate the judgment criteria.

To put it simply, when the game starts, suppose the topic you get is "belt", you can see that many topics in the picture are unfinished or drawn very sloppily. But how does AI technology tell whether a drawing is of a belt or a snake? It will compare the image of the painting with all past databases, and analyze according to everyone's information, how to start drawing the brush usually means drawing a belt and how to start drawing the brush means drawing a snake.

After the game is over, you can click on the picture to see what shape the belt you drew is similar to. It may also be one of the options, or you can click on the Quick, Draw game database to see how others draw. When the graphic features or strokes you draw are similar to these shapes, the computer can guess the shape you drew.

Of course, the algorithm behind the Quick, Draw game is not completely accurate and sometimes there are mistakes. However, AI's learning and judgment is a matter of probability. The more people play this game, the larger the database will be. The more comparison processes and details there are, the higher the probability of guessing. This is also the process of AI machine learning, learning through "data", just like Google Translate. As the database increases, the translation will become more accurate and closer to the original meaning of the language.

From the Quick, Draw game, you can learn the basics of how AI artificial intelligence learns, and explain to children how machines learn and what the current limitations are. If you are also a technology geek or a friend who likes group games, try the Quick, Draw game and learn about AI while playing!

Want to learn more? Click the button to get more information about the program!