The site is supposed to improve awareness of machine learning, explaining what it is and isn’t. Called Teachable Machine, the online experience lets you “teach” an AI how to perform tasks using just your web browser and a camera.
Teachable Machine is meant to be accessible to everyone. There’s no code to learn or complex technical details to understand. All that’s asked is you follow the onscreen instructions and have a camera to hand. You’ll need to take at least 30 photos each time you run the machine. These will be used to “train” the onscreen AI to replicate your actions.
Some of Google’s examples include making the AI’s hand say “moo” and forcing it to mimic an air guitar by wiggling your fingers. Google warned that the machine can’t comprehend high-level concepts, such as faces and objects, so it learns by following the gestures you provide. You can reset the learning process at any time if you want to try something else.
Google said Teachable Machine is supposed to improve peoples’ understanding of how machine learning actually works. The concept is becoming increasingly widespread as it finds its way into more services, devices and apps. Although many web users have now experienced the tech in some form, most are still unfamiliar with what “machine learning” really means.
“From helping you find your favorite dog photos, to helping farmers in Japan sort cucumbers, machine learning is changing the way people use code to solve problems. But how does machine learning actually work? We wanted to make it easier for people who are curious about this technology to learn more about it,” said Google. “So we created Teachable Machine, a simple experiment that lets you teach a machine using your camera—live in the browser, no coding required.”
On the technical side, Teachable Machine was created using a browser-based deep learning framework. It allows web developers to run neural networks within the context of a web browser, facilitating on-device AI.
All the images you create using Teachable Machine are stored on your device and never sent to Google’s servers, keeping the site safe to use. Developers interested in the project can find the source code on GitHub.
