When I read this morning's Wired story about the advanced digital assistant that the developers of Siri are working on, it sounded very familiar.
According to Wired's account, Siri's developers are now working on a new company called Viv that has a similar albeit more advanced vision. They hope their service can parse complicated commands like "Give me a flight to Dallas with a seat that Shaq could fit in." Viv would look at data from a variety of sources like Kayak, SeatGuru, and NBA information to find the right answer.
[ InfoWorld dishes on must-have iPad office apps, essential Android productivity apps, and road warrior standbys. Start downloading! | Get the latest insight on the tech news that matters from InfoWorld's Tech Watch blog. ]
It seemed familiar because Microsoft talked about a very similar vision years ago. In late 2010, three years after Microsoft bought speech recognition company Tellme, executives described an ambitious vision for how it planned to use the technology. Microsoft was trying to get its voice recognition software used far and wide so that it could collect loads of information about how people were using it in order to continuously improve it.
Way back then, Microsoft said it was collecting 11 billion speech requests a year.
Some of the learning Microsoft wanted to do from gathering information about those requests was pretty rudimentary. For instance, when people used the voice recognition service on their Windows Phone 7 and repeated the request a few times, Microsoft would figure it wasn't offering the right response but it would also look at phone connectivity data, which might have affected the result.
But the idea was to use machine learning and natural learning to let people make requests in a normal, conversational language. Microsoft also wanted to be able to tap into a wide source of data to offer the best response.
The example Microsoft gave back then was that a user could ask something like: "Find somewhere for Bob and I to have dinner Tuesday night." The service would check your schedule as well as Bob's to find out where and when you'd both be available. It would also check what kind of food both people have eaten before and then return with a follow up query: Will you be eating in San Francisco and would you like sushi? From there, the service would make a reservation and put it on both calendars.
Microsoft wanted its technology to be widely used, on its own products like Windows Phone and Xbox, as well as third parties like Ford, which used the technology in its Sync product, and companies such as American Airlines that were using the technology in their customer service phone lines. The more widely it was used, the more data Microsoft could collect to improve the service and offer better answers.
Viv will need its technology to be similarly widely used in order for it to become as powerful as its creators hope. Plus, its technology will have to be as good as its creators say it is.
Microsoft's vision sounded pretty good but hasn't come to fruition yet, although Cortana -- the virtual assistant that rolled out in Windows Phone 8.1 -- is getting close. Cortana scans information sources like local email to help figure out your preferences and interests, then uses this information in its responses.
People have been dreaming of this ideal digital assistant since science fiction writers first imagined it. Clearly it's a hard problem to solve. But advances in machine learning and artificial intelligence might make it possible eventually.
This story, "Siri inventors share vision that sounds like Microsoft's from years ago" was originally published by CITEworld .