Remember the outage that impacted a billion-plus Facebook, Instagram and WhatsApp users around the world earlier this week? It certainly was a trying time, but the important thing is that we all made it through to the other side seemingly unscathed.
And as everyone was swiping through their feeds filled with status updates and photos that wouldn’t render during the hours-long digital chaos, you might have noticed something interesting. The question mark-filled boxes where those selfie and vacation photos should be, instead, provided brief written descriptions of the picture contents.
While there is a relevant reason for the idea behind the text, the outage provided an in-depth look at just how closely Facebook’s artificial intelligence (AI) “sees” and tags every photo you upload. It also provides a clearer picture of the amount of data the social media giant continues to collect.
Facebook’s routine maintenance ‘triggered an issue’
The trouble all started around 8 a.m. Eastern on Wednesday, the day before the July 4th holiday. For a number of hours throughout the day, users reported photos on their Facebook news feed wouldn’t load properly, stories on Instagram weren’t showing up and various problems trying to send messages through WhatsApp.
Facebook blamed the outage on routine maintenance that “triggered an issue.” Hours later, the problem was resolved.
Earlier today, some people and businesses experienced trouble uploading or sending images, videos and other files on our apps and platforms. The issue has since been resolved and we should be back at 100% for everyone. We’re sorry for any inconvenience.
— Facebook (@facebook) July 4, 2019
This is one of several outages affecting Facebook and its other services since March, but this one was unique. While their teams worked to restore service in this instance, the outage offered a very interesting glimpse of how some of their most advanced technology works behind the scenes looking at your photos.
Image may contain: tree, nature, people smiling
It’s already a well-established fact that Facebook looks at what you post, whether it’s real human moderators or AI. Facial recognition has played a big role when it comes to photos over the past couple of years, which is evident when you’re prompted to tag someone the company’s vision systems recognize.
Facebook also uses AI to determine not just who, but to interpret what else is in the images you post. It’s part of their work in accessibility as a way for people with visual impairments to know what’s in a photo without seeing it.
So, on Wednesday when photos weren’t rendering, users were instead seeing blank boxes with descriptions such as Image may contain: Nature, people standing, sky.
Take a look at some of the real examples below.
No one likes being labeled
The way Facebook’s AI interprets images led to a lot of people taking notice, and even offense, to how their uploads are labeled. In its defense, I don’t think machine learning is inherently malicious right now but give it a couple of years and we’ll revisit.
However, let it serve as a stark reminder that Facebook not only collects data based on what you say or your social media likes and habits, its system also “reads” your photos. It helps Facebook paint a better portrait of who you are, regardless of whether it’s accurate or not.
Aside from learning something new, look at the bright side that the outage provided at least a temporary reprieve from the shackles of social media. It was an independence day of sorts, the day before Independence Day.