Skip to content

Two Awesome Tools to Demystify AI: Teachable Machine and Quick Draw

Tool Tuesday

If you haven’t dived deep into Artificial Intelligence, it may feel a little bit like magic.  This feeling is also what drives some of the fear when dealing with artificial intelligence. As exciting as AI is and as much as it feels like magic, it is not.  This week I will look at two tools that will help us understand (at a very basic level) how AI works and demystify it.  Both tools are great for also teaching students about AI. 

Teachable Machine

First up is Teachable Machine a tool developed by Google and follows Google’s privacy policy.  There is no requirement to sign in to Teachable Machine unless you choose to upload data from your Google Drive.  Certain features do require the use of your computer’s camera and microphone.  

Now that we have the nitty gritty out of the way, What exactly is Teachable Machine? Teachable Machine is a place where anyone can train an AI model.  Imagine teaching your computer to tell the difference between two sets of data, say cats and dogs. This tool not only simplifies machine learning but offers the opportunity to see how artificial intelligence is influence by the data given. Check out the video below to see more of a glimpse into this tool.

Demystify with Teachable Machine

Exploring a simple example will help demystify AI and illustrate how teachable machines work.
MIT provides lessons and data around Teachable Machine. The set of data (below) is a set of cats and dogs that you input into the machine. 

Take a moment to observe this small set of data and then decide what you think the machine will identify the animals below as.

Keep in mind that when you go to the Teachable machine you will plug in this data and then click “train the machine.”  Once you do that you’ll input new piece of data (like the cat and dog above) and ask the machine to identify it.  There are a variety of reasons why the machine may or may not be able to identify the new data.  In our case of the cat and dog, it is most likely going to label our cat as a dog, and our dog as a cat.  When looking back at our original data you will see that most of our cats are fluffy, so when we introduce a cat that is not the machine struggles to identify the new piece of data.  Same with our dog, it is fluffy and most of our original dogs are not. 

How do I use this with students?

Using the examples above and teaching students what AI is and what AI is not is very important. This allows students to make good choices on how to use AI ethically. Teachable machine is a great example of how AI and machine learning work. Students will see that what data is put into the machine matter for the outcomes.  If our data is incorrect, there is not enough data or it is biased then what is generated is going to have issues.

Challenge students to enter the dog and cat data and have conversations about what is happening.  MIT’s lessons will help you to walk you through these conversations.  

Here are some other questions you can use with your students around biases: 

  • How could I get the machine to recognize the hairless cat or the fluffy dog?
  • Bias can happen unintentionally based on who is training the machine. (The person is putting information in from their perspective, they may miss something) Ask students what’s missing or what’s not here in this data set.
  • Could we train the machine incorrectly, what would be the implications? **This is important to note that students may try to “break” the machine, and instead of getting angry have a constructive conversation around what if?
  • When these biases happen how might it make someone feel to be left out?
  • Show students that sometimes the machine gets it right (with our previous data) – If the tool is right 50% of the time is it useful too? What if this were recognizing faces and only got half of them right, is it still a useful tool?

Another Challenge:

Once students have put the set of data of cats and dogs in the machine have them train the machine on their own set of data.  This can be anything in the classroom; pencils and pens, thumbs up and thumbs down.  Want to up the engagement?  Have students bring in items from home.  The coolest example of this is the teacher who had their students bring in Squishmallows and other plushies to have the machine identify them.  Then have conversations around their findings, ask “what’s missing.”

I hope this will be an effective way for you to teach students how AI works as well as have some important ethical conversations about the data that is used with artificial intelligence. Another tool that does this well is Quick Draw, which has an incredible amount of data.

Quick Draw

First things first, a WARNING.  This game will suck you in.  Every time I click on Quick Draw, even if it’s just to check something out to share, I end up playing.  You and your students are going to have fun with it!  There is no sign-in so that is a bonus for using it.  This is also powered by Google, so it follows their privacy policies.  

Quick draw is just that. The AI gives you a word and you draw it, and then the AI guesses it. What is neat about Quick Draw is you can investigate all of the data that is inputted into the machine. Again, this is an easy entry place to have a conversation about biases.  Ask your students if anything is missing from the data set.  This is also a part of MIT lessons.  This blog entry from Google helps to visualize the data in a cool way by showing 1000 drawings of an object from all around the world.  Yet another entry point to the conversation, this time about culture, that AI is helping us have.

These two tools are simple yet they provide ethically important conversations that our students need to have.  These conversations foster a safe place for our students that makes them feel included and represented with an easy entry point through drawing and data.  I encourage you to try them out for yourself but also explore them with your students.  Come share about how you’re teaching students about AI and the data behind it on our Facebook page.

Remember all of our Tool Tuesday’s Tools can be found here.

-Stay Curious, Stay Innovative!

apple with Nicole

Leave a Reply

Your email address will not be published. Required fields are marked *