KarAIoke: When AI Art Meets Music

I’ve spent the greater part of the past year, talking with teachers and leaders around the country around AI. They’ve share a major concern that AI will cause students to lose creativity and critical thinking. I’ve been thinking about this concern more and more over the past few months as schools dive deeper down the AI rabbit hole. Then something interested happened…

Last month I was speaking at the FETC conference about AI and specifically, AI tools that expand creativity. The night before my presentation I had an idea hit me. What if you could take lyrics from a song, enter them into an AI image generator, and then have people guess the song based on those image outputs? Kind of like “name that tune” but only where the audience has images instead of the actual song.

I called this new creation “KarAIoke” and set out to test it with the audience at FETC, TCEA and schools where I have been doing a lot of AI training (including my daughter’s high school!). After a few tweaks, the end results have been amazing. It immediately hooks in the audience and in some cases causes spontaneous sing-alongs, a great way to start a session or have a brain break in between.

Before I get into how I set this up, I just want to say that this creative activity would NOT have been possible without AI. I have some mild artistic ability, but drawing out the images of lyrics in multiple creative ways would take me months. I’ve done a lighter version of this using emojis in the past, but the image generators out there today really make this activity sing! (pun intended)

The Set-Up

Using an AI image generator of your choosing (I use Copilot, Canva, or Adobe Firefly, depending on my mood), find some lyrics of a popular song and copy/paste them into your image generator. Depending the on the lyrics (the more visual the better), you may have to re-engineer your prompts to get the best outputs. Here’s a one of my early examples (don’t scroll too far if you want to guess):

You can almost hear that legendary Alicia Keys riff as she sings, “This girl is on fyyyyrrrrrrrrr” just by looking at the image. I create 4-6 similar type examples and have them queued up on my slide along with a brief section of the song ready to play should they get it right.

Playing KarAIoke:

Now that I have my images and songs ready to go, I set up this “AI-inspired” icebreaker with the audience. I generally start with a couple of easy examples first to get their confidence going. Rather than have the audience shout out the answer right away, I ask them to have a quick 10-15 second discussion with the people next to them as they try to come up with the answer. Then I count them down and ask them to all shout what the name of the song is.

The best reactions are to songs where the audience can sing along so I have to throw in a couple of classics like this next example, where I pulled a set of lyrics from a classic sing-along to jog their thinking:

Indeed, good times never seemed so good (So Good! So Good! So Gooood!). Again, it takes some tweaking of prompts to get the right output for your song. Just putting in a “weeping guitar” might not generate enough visual clues for the audience. I also like to give them multiple images of the same song to dissect.

Follow-up activity:

Depending on the size of the audience (I’ve done this with up to 500 people), you could do a follow-up activity with the crowd. For my workshop at TCEA, I had 50 people in the room which made it the perfect size to have them put on their creative thinking hats and add suggestions for KarAIoke classics to my Padlet wall. Now that Padlet has the “I Can’t Draw” feature built into their platform, you don’t even need a log-in or other tool. You can just have them create right within the platform and post! As you can see from some of the results, they did not disappoint:

Summary:

Whether you are doing the KarAIoke warm-up (5 minutes) or the full activity (15 minutes), what I like about this is that it gets the audience immediately thinking creatively. And most importantly, it shows them that AI could actually help with creativity rather than hinder it. Before I wrap up, I have to add one more that is my favorite one to end on:

It’s not a midnight train to Georgia or a Runaway Train. However, if you look in the bottom right corner you might see that she’s just a small-town girl…living in a lonely world….

Creating with AI has been quite a journey for sure. Want to come up with your own? Be sure to use the #KarAIoke hashtag on socials so we can all join in the fun!

Carl Hooker is an international speaker and trainer. He works with schools and events across the country to thoughtfully integrate tools like AI into learning. His latest book Learning Evolution shares several examples, strategies and ideas like this one. If you are interested in booking Carl for your next event or professional development day, fill out this speaking form to get more information.