Google Photos new optional feature, AI by labelling images

Google has updated its Google Photos app on Android with the latest option that allows users to determine the google about the contents of their pictures.

By naming these images, Google can update its object identification algorithms, which indeed make Photos more useful. It’s a virtuous cycle of AI development best deployed by tech Google which has many data and lots of users.

This isn’t an uncommon practice at all. Machine learning systems don’t just learn by themselves, and the large majority of these applications require to be developed using data labelled by humans.

It’s the same purpose that CAPTCHAs ask you to recognize cars and motorbikes in images. By identifying these objects, you’re training AI to do the same.

The feature looks in the most up-to-date version of Google Photos. Just click on the search button in the app’s menu, scroll down, and you’ll notice an option to “Help improve Google Photos.” As reported by 9to5Google, click on it, and you’ll be performed with four tasks: to define your printing choices for photos; your favoured collages or animations; to recognize which photos relate to which holiday events (e.g. Christmas or Halloween), and to recognize the contents of photos (“Name the most significant elements in this photo”).

As Google reveals on a help page about the feature: “It may need time to see the result your contributions have on your account, but your input will assist enhance existing features and build new ones; for example, updated suggestions on which photos to print or higher quality creations that you would want.

You can remove your answers at any time.” (To do so, tap the three-dot menu at the top right of the screen and click “Delete my answers.”) At the time of writing, it appears the update is accessible only on Android, not iOS.

Although this seems to be a new extension to the Google Photos app, the underlying software is much older. The method is powered by “Crowdsource by Google,” a crowdsourcing platform that the company launched in 2016.

It gamifies data-labelling, allowing users to earn points and badges by performing tasks like verifying landmarks, knowing the sentiment of text snippets is a review positive or negative, for example, transcribing handwritten notes, and another related job. Though: users don’t get any tangible rewards for their work behind virtual kudos from Google.

It’s worth getting all this when using Google’s whizzy machine learning products: they wouldn’t be divided as useful without humans helping guide them.