Google's Artificial Intelligence Knows When a Stranger Is Looking At Your Phone

It's all about using AI to enhance the device's ability to identify the user and friends in order to more accurately increase monetization opportunities. JL

James Vincent reports in The Verge:

The system works in all sorts of different lighting conditions, and has a reaction time of just two milliseconds. With that speed, even a passing glance is likely to be spotted! Features like this may look trivial in isolation, but collectively they’re going to add up to a whole lot more. Machine learning and AI are already becoming the big differentiators between smartphones

Ever get the feeling someone is looking over your shoulder at your phone? Well, you might not have to worry about that in the future: Google’s researchers have developed an AI tool that can spot when someone is sneaking a peek at your screen.

The software — which was first spotted by Quartz — is called an “e-screen protector,” and is currently just in the research stage. (We don’t know if Google has plans to add it to future versions of Android, but we’ve reached out to check.) It’s quite simple: it uses your front-facing camera in combination with some face- and gaze-detection algorithms to identify anyone looking at your display.

In the demo video you can see it reacting almost instantaneously to a suspicious looky-loos, switching the user’s screen to catch the culprit red-faced, and tagging them with some Snapchat-style rainbow vomit for good effect. According to Quartz, the creators of the e-screen protector, Google researchers Hee Jung Ryu and Florian Schroff, say the system works in all sorts of different lighting conditions, and has a reaction time of just two milliseconds. With that speed, even a passing glance is likely to be spotted!

Although we can’t say whether or not Google will ever bring this feature to Android, it’s a perfect example of the sort of small software tweaks AI can offer mobile devices. Google is doing more and more on-device machine learning with the help of its TensorFlow Lite software, and has added simple tools that automatically detect numbers and addresses when you’re copy and pasting text, and suggest quick replies to your emails, for example.

Features like this may look trivial in isolation, but collectively they’re going to add up to a whole lot more. Machine learning and AI are already becoming the big differentiators between smartphones — so expect to see more like this in the near future.