September 15
Intel’s CTO Justin Rattner described a coming era of gadgets able to act like smart assistants in his keynote at the Intel Developer Forum in downtown San Francisco today, and also spent a fair amount of it under surveillance himself. It seems that gadget snooping is something we’ll have to get used to if we want our devices to be smarter.
Throughout the presentation small sensor packs tucked inside Rattner’s socks tracked the movement of his feet, technology designed to detect when an elderly person is at risk of a fall. When he grabbed a remote control to talk about TVs that know who you are accelerometers in the controller identified him from the unique way he held it.
“Future devices will constantly learn about you,” said Rattner, “your habits, how you go about your life, your friends. They’ll know where you’re going, they’ll anticipate, they’ll know your likes and dislikes.”
Rattner presented several demos to illustrate how information collected and shared by your phone and other mobile gadgets can be used make that happen. One was a working version of the “shadow avatars” system I blogged about previously–animated characters that mirror the current activity of your contacts so you know at a glance that a friend is, say, walking and listening to music.
Another was a travel app developed together with guidebook publisher Fodor’s. Shown running on a small tablet PC, the app uses your current location and knowledge of your past preferences to make intelligent suggestions of things to do, for example picking restaurants within your price range that serving your kind of cuisine. When switched to “wander mode” the app automatically suggests nearby attractions, also based on what it knows about you. Everything you do is automatically logged and pushed to a blog–it could easily be a social network like Facebook–summarizing where you went and the photos you took. In user trials in New York testers reportedly said they would pay $20 or more for such an app, significantly more than the few dollars that mobile travel apps today command today.
All this is made possible by algorithms that crunch data from both sensors and information drawn from calendars and the software active on a device. To be useful those algorithms will have to be constantly at work to diagnose what you’re doing, squirreling away data in the cloud so it can be used to inform future actions or shared with others as appropriate. That will make demands on the power of gadgets, Rattner conceded in his Q&A after the talk, particularly an applications that processes audio to work out if you’re in a meeting, or one that works with images.
That suggests it would have to be integrated fairly deeply into both a devices operating system and hardware, although Rattner was vague about just how Intel will bring it to market. “I think it can really be available across the full range of operating systems,” he said, dodging a question about whether it would be integrated into Intel’s Meego mobile operating system. “The business decision about this technology has not been made.”
Another unknown is how users may feel about introducing a whole new level of sharing. “If you think identity theft is a problem today imagine when you’re whole context is readily stored on the web,” said Rattner. Although it was emphasized that users will be able to control what of their context is shared, and with whom, the dizzying complexity of the vision Intel unveiled suggests there may be some consternation from users when it comes to pass.