Go to Facebook and look at a photo containing a person or people that has been uploaded by you or one of your friends.
Right click on the photo and inspect it. Look for the image’s alt tag. Chances are you’ll see something like this:
Facebook is using a machine learning artificial intelligence to fill these out. This isn’t done by users and this isn’t done by employees of the company. It’s done by a machine.
Some thoughts on this:
Facebook is likely doing this mostly because they can. The stated reason is that the alt tags help the visually impaired. However, these tags aren’t written well enough or concisely enough to be of that much help. They’re also a bit too broad to help with image searching, unless at the granular level (and I’m not sure Facebook search can go that granular).
It’s likely that Facebook is doing this to collect data, test, and demonstrate a technology. Part marketing as it helps establish Facebook as a technology powerhouse and some part true utility.
This is fascinating as it means more of the Internet will soon (as this technology advances) be much more searchable than it is. Google is likely doing this to a reasonable extent for images on Google Image Search as well. Few images will be private, and most will be identifiable.
While yes, this means less privacy, it also positively means that it will be easier for those who deserve it to get attention.
This is an interesting sign of the future and is emblematic of the huge ways in which machine learning will be incorporated into search and marketing down the line. It’s a bit frightening, but I for one am welcoming of the automation.