New Watson-Style AI Called Viv Seeks To Be the First 'Global Brain'

paysonwelch sends this report from Wired on the next generation of consumer AI: Google Now has a huge knowledge graph—you can ask questions like "Where was Abraham Lincoln born?" And it can name the city. You can also say, "What is the population?" of a city and it’ll bring up a chart and answer. But you cannot say, "What is the population of the city where Abraham Lincoln was born?" The system may have the data for both these components, but it has no ability to put them together, either to answer a query or to make a smart suggestion. Like Siri, it can’t do anything that coders haven’t explicitly programmed it to do. Viv breaks through those constraints by generating its own code on the fly, no programmers required. Take a complicated command like "Give me a flight to Dallas with a seat that Shaq could fit in." Viv will parse the sentence and then it will perform its best trick: automatically generating a quick, efficient program to link third-party sources of information together—say, Kayak, SeatGuru, and the NBA media guide—so it can identify available flights with lots of legroom.

Re:P vs. NP

[Anonymous Coward]

Actually, once you're willing to give up 100% perfection, NP-complete problems tend to be astonishing easy (computation-wise, developing the algorithms is still hard). Either you can find an acceptable approximation algorithm, or much more importantly, most instances of NP-complete problems are "easy" instances. The worst case for our, say, SAT solvers is still as bad as "exponential time" makes you expect, but that worse case is actually very rare. This observation is why SAT/SMT solvers have gotten a lot better in the past decade or so. Also, the entire discipline of machine learning can be seen as figuring out how to get computers to make WAGs that tend to be right. It's certainly not human-level at many important tasks, but it's improving, and simply adding more data is surprisingly effective at improving machine learning performance (which is why Google Now/Siri/etc. are run by large companies with lots of users and therefore lots of data).

Re:This is important

[BitZtream]

Yea, its important ... because they've just realized they need to do multi-part/nested queries.

Its not really impressive, its a 'no shit sherlock', and I'm blown away that google can't do this already.

Watson can.

The important part is that someone just realized they need to do one query, look at the type answer and then use that to generate a new query.

Well, okay, its not really important or even new ... as I said, Watson can do it and has been able to for years.

Google Now Does Understand Context

[Forthan Red]

... to a limited degree. While you can't ask the Lincoln question in a single statement, you can ask, "Where was Lincoln born?" then when it replies "Hodgenville, KY", you can then say "What is its population?", or "Show it on a map" and it will know from context that the "its" you're referring to, is Lincoln's birthplace.