Why AI Advisors are The Future of Your Electronic Devices

We like product leaks as much as the next person, but what we don't like is the blurry, indistinguishable photos that go with them. Is that an iPhone, or a TV remote? Perhaps a chalkboard slate with the Apple logo drawn on? There's really no way to know for sure.

The obvious solution is also the simplest: learn to take better pictures. Automatic modes on photo and video cameras can help, but you could argue they do more harm than good. These cameras may take better pictures, but they're not making us better shooters — there's no feedback to show us how. 

That sounds like a job for artificial intelligence to us.
 
Modern cameras do one of two things: correct your errors automatically, giving you little chance to learn from your mistakes, or simply display raw data that's hard to interpret. There's no middle ground, no actionable steps for teaching or improving your skills. What's worse is that those automatic modes aren't the final word on good photography or video — they can get things wrong too.

But then there's an application called NudgeCam. Created by researchers at a Fuji-Xerox research lab in Palo Alto, NudgeCam is an Android app that actively teaches you to take better photos and videos, all in real-time. Your phone's gyroscope is used to ensure images stay level, while heuristics and algorithms alert you of unsuitable brightness or colour levels. 

But perhaps the most impressive feature is NudgeCam's innovative use of facial recognition technology. Not only can the application detect faces, but it tells you when they've been improperly framed, or even cut off completely, taking the rule of thirds to a whole new level.

What's important here is that the application provides feedback, both audibly and visually, enabling users to learn from their mistakes — and hopefully, take better pictures or videos as a result. It's a much more effective approach than trying to make sense of a histogram or zebra stripes. While those tools are great for more advanced users, they provide little in the way of valuable data for new users to photo or videography. 

That being said, nothing about NudgeCam is novel. It combines hardware and software that's existed for years, but in a way that the users can interact with. And it's clear that other companies have begun to notice this as well. Nikon introduced a Guide mode with its entry-level D3000 DSLR last year that does much as the name implies. Things like aperture and exposure are explained during the shooting process, so that new shooters have the potential to improve. It's not quite as real-time as NudgeCam's implementation, but it's still a step in the right direction.

However, this is the sort of thing that could have applications far beyond digital cameras, for any scenario in which unpredictable human input is involved. Motion games on the Wii, Xbox or Playstation 3 could benefit greatly from having artificial intelligence guide players through incorrect gestures or moves — a frustrating experience when a certain motions just don't seem to work as planned. Meanwhile, Skype, or other audio-recording applications, could not only detect incorrect audio levels, but poor audio quality or sound as well, and provide suggestions for potential fixes.

And while some have already taken the first step in implementing JARVIS-style systems for commands and instructions in their homes, what's missing is feedback. That interaction, complete with real-time learning and adaptability, is where consumer electronics are heading. And if our phones and cameras have the voice of Paul Bettany to go with it, that's just icing on the cake.      
 
Image via Flickr user iii_kkk5.