Google wants you to help train their AI by tagging images in Google Photos

Google has updated its Google Photos app on Android with a new option that allows users to notify the search giant of the content of their images. By tagging these images, Google can improve its object recognition algorithms, making the photos more useful. It’s a virtuous cycle of AI development that is best deployed by tech giants like Google, which have lots of data and lots of users.

This is not at all an unusual practice. Machine learning systems don’t just learn on their own, and the vast majority of these applications must be taught using tagged data by humans. This is the same reason why CAPTCHAs ask you to identify cars and motorcycles in pictures. By identifying these objects, you train the AI ​​to do the same.

The feature appears in the most recent version of Google Photos. Just hit the search button on the app menu, scroll down and you’ll see an option for “Help improve Google Photos”. As reported by 9to5Google, click on it and you will be presented with four tasks: describe your printing preferences for photos; your favorite collages or animations; to identify which photos belong to which holiday events (eg Christmas or Halloween); and identify the content of the photos (“Name the most important elements of this photo”).

Screenshots: The Verge

As Google explains on a help page about the feature: “It may take a while to see the impact of your contributions on your account, but your contribution will help improve existing features and create new ones. . for example, improved suggestions on photos to print or higher quality designs you want. You can delete your responses at any time. “(To do this, tap the three-dot menu at the top right of the l ‘screen and hit “Delete my replies.”) At the time of writing, it looks like the update is only available on Android, not iOS.

While this appears to be a new addition to the Google Photos app, the underlying software is much older. The process is powered by “Crowdsource by Google,” a crowdsource platform that the company launched in 2016. It plays on data tagging, allowing users to earn points and badges by completing tasks such as checking landmarks, identifying sentiment in text snippets (is a positive or negative review, for example), transcribing handwritten notes, and other similar work. To be clear, though: users don’t get any real rewards for their work beyond virtual kudos from Google.

This is all to remember when using Google’s machine learning products: they wouldn’t be half as good without the help of humans.