The company has recently conducted an experiment in which they wanted to see how machine learning could learn subjective concepts such as art, and thus they created a system in which AI would mimic the workflow of a professional photographer, and use landscape panoramas from Google Street View and edit them to make it look like a professional has done it.
As you can see in the image above, the top part of the image is the end result of the editing, while the bottom portion is the image taken from Street View. If you did not tell us that this was done by a computer, we would have easily believed that this is the work of a professional photographer.
What this means is that Google has managed to teach the machine what might be considered to be “good” by humans. Google even asked real photographers to judge the images, in which 40% of the ratings they receive were at “semi-pro” to “pro” levels.