I realized while playing that this is a terrible idea on Google's part. I mean sure, it's kinda fun...but it rewards ambiguity.
Take a look at these examples:
The first one. They almost look like atomic models. Of course, I can't really make out what it is, and a layperson wouldn't know they were atoms...so, of course the answer we both agreed on was 'balls'.
The bottom right one...it's probably a specific mineral. If I type "mineral" into google, and I get this pic, I'd be fine, but what is it really?
Some pictures, which I couldn't take a screenshot of (sorry)....the answer was something really simple that was in the picture. Like one a man, with a printing press, and a computer on a desk on the right side. The common answer was, "computer". If I typed "Computer" into google image search and got that image, I'd have cried. It was clearly not the focus of the image, but because it was simple and in the picture, that's what we agreed on.
I have a bad feeling about this idea...