Google has updated its Google Photos app on Android, which allows users to tell the search company about the contents of their images. By labeling these images, Google can improve its object authentication algorithms, making the resulting photos more useful. This is a good cycle of AI development with better data and more users by technology companies like Google.
This is not an unusual practice. Machine learning systems are not self-learning, and most of these applications must be taught using human-named data. This is the same reason why captchas ask you Identify cars and motorcycles in the pictures. You train AI to do this by identifying these objects.
The feature appears The most recent version Google Photos. Tap the search button in the app’s menu, scroll down, and you’ll see the “Help improve Google Photos” option. As Report 9to5Google, Click on it, you will be given four tasks: to describe your printing options for photos; The galleries or animations you like; Identify which photos are available for which holiday events (e.g. Christmas or Halloween); And identify the contents of the photos (“Name the most important things in this photo”).
Google explains one Help page About the feature: “It may take time to see the impact your contributions will have on your account, but your input will help improve existing features and create new ones; For example, advanced recommendations on which photos to print or the high quality works you want. You can delete your answers at any time. ”(To do so, tap the three-dot menu at the top right of the screen and press” Delete my answers “.) At the time of writing, it seems that the update is only available on Android, not iOS.
Although this seems like a new addition to the Google Photos app, the basic software is very old. The process is driven “Meeting hosted by Google, ”A crowded site that the company launched in 2016. It allows users to gamble on data-labeling, verify identities, identify the meaning of text snippets and earn points and badges by completing tasks (e.g., review positive or negative), transcript handwritten notes and other similar work. Be clear: However, users do not receive real rewards for their work beyond virtual pride from Google.
All of this is worth remembering when using Google’s Vissy Machine learning products: they are not half as good without helping humans teach.