You may be wondering what the merry hell this image is. Well, it’s a photo of me that’s been through Google’s ‘Deep Dream’ neural network, which is a piece of software designed to find patterns in images. If you use Google Image Search to find photos of cats, then Deep Dream can help Google identify photos containing cats without any additional context (i.e. some text next to the photo on a web page saying ‘this is a photo of my cat’). If you use Google Photos, then Deep Dream allows you to search your own photos in the same way.
The image I’ve included shows what Deep Dream sees when it analyses the photo of me. It seems to have identified my face quite well, although it reckons there’s a second pair of eyes on my forehead which is odd. There’s also a small dog growing out of my right shoulder, and it also reckons my left jacket pocket is a dog as well. And there appears to be another dog further down my jacket too.
Google made Deep Dream open source on Github, and so a number of online generator tools have popped up. I used this one which took several days, but this one seems better. So you too can see how many dogs Google thinks are in your profile photo. And it might also explain why, if you search your own photos on Google Photos for dog pictures, you may get some irrelevant results.