Guessing ages just got a tech upgrade. Microsoft's facial recognition robot attempts to correctly guess a person's age. It does this by studying an uploaded human face photo.
Microsoft's machine-learning team has built a robot at how-old.net that guesses the age and gender of a person in the photo. The photo must be less than 3 MB, and the robot can detect if the picture is the face of an animal or Muppet, but not a Barbie doll.
Joseph Sirosh, a Microsoft Vice President, said that the new tool is "really fun," according to PC Mag. He unveiled the new robot at a recent Microsoft BUILD conference.
Microsoft was initially hoping that "50 users" would visit the website. However, within a few hours tens of thousands of visitors put a strain on the site and servers.
To be fair, the accuracy of the site's estimate is based greatly on the picture that is submitted, such as the quality of lighting. A picture can be a stone's throw or a country mile away from a person's true age, due to various factors.
If the lighting in a picture is good, the robot often guesses the correct age within one or two years. Other times the Microsoft tool's result is off the grid, providing six ages simultaneously, for example.
The backend of the app is continuously searching for new photos that it can learn information from, according to Tech Crunch. So it is a work in progress on multiple levels.
The bottom line is that Microsoft's age-guessing robot has become a viral hit. And since Microsoft is "still improving" the feature, people should be more forgiving if its guess is decades off.