So, I was refreshing my twitter feed and a friend of mine was documenting a major milestone for him and his first new car. Basically, Yeri (that’s his name) was taking pictures of his dashboard.

I didn’t ask him why, but he was using Google goggles to take pictures of his dashboard. And, as it turns out, goggles labeled the pictures with the name of the car manufacturer!

Google goggles can even recognise what car maker the dashboard is from :o pic.twitter.com/ReAZNxbo7Z

— Yeri Tiete (@Tuinslak) August 21, 2013

First impression: nice job! Goggles is very intelligent, and keeps getting more comprehensive every day. Artificial Intelligence is capable of matching the colors and curves of a dashboard to a car manufacturer, no doubt. But if you’re familiar with Machine Learning, you should know you need at least a thousand pre-labeled dashboard pictures from different viewing perspectives in various lightning conditions to pull this off. A thousand pictures per car type…

30 seconds later I had the answer. It was so obvious: as Google has tons of services online, millions of users update their data every second. It turns out, for every type of car, dozens of speed junkies film their dashboard while accelerating from 0 to 60 mph / 100kph.

A simple search for “BMW 0-100” on YouTube:

Boom! There you have it: hundreds of thousands of frames to learn how to recognise the car manufacturer by the dashboard. No doubt Google is leveraging this infinite data source for this feature. Dashboard recognition is possibly less mind blowing than Face recognition. But hey, who knows what feature you’re about to discover next?

Anyway, what you should take away from this is: when some technology company changes its terms & conditions, you giving up your rights could be essential for innovation.

(For the record: Yeri is also an ex Google employee.)