Gadgetory

google has a lot of products it also knows a lot about you Google now is a feature that has a potential to tie all that together in a way that's actually meaningful they invited us to their Mountain View campus of California to talk to people behind the project and see where it's going in the future we think the phone and more broadly kind of the computer that's with me whether it's a phone or surveys glasses or something else is is pretty unique because it has them basically the context of what I'm doing and we're on that so we've always had a search bar and across the very top on Android from the very beginning in jellybean we made a few pretty significant changes to the search experience on Android the first thing that we wanted to do in and did was to go much deeper in answering questions directions to the museum with the William Paley exhibition so it turns out it's at the DeYoung museum what's happening here is that there is a lot of information on the web that talks about this particular you know exhibition and then it obviously talks about where it is and that's the de Young Museum in San Francisco so when I say that Google crosses all these various different pieces of information in my question to give me an answer back which in this case is actually just a map we're obviously very excited about search on mobile devices on Android and about the idea of people being able to interact with their phone and in just a natural way and if you think about what that takes we feel like there's a few key elements that have to come together one obviously is voice recognition the second piece is once I have those words is natural language understanding and a third piece which is a fairly new thing for us at Google is understanding what are the basic kind of components and facts in the world and this is something we call our knowledge graph and it's really the first time in sort of computer science history that those three things are all available and not just available in some giant computation center but of the you know in a powerful device that's in my pocket so one of the most important things from Larry's founder letter from earlier this year is the part where he says he thinks computers should do all the hard work for you sort of getting all the hard stuff out of the way so you can enjoy life you know enjoy whatever use it they would like to do for example if you have a flight reservation you will have received that in your email and you know you get to the hotel you get to the restaurant whatever it's hard to find it so what Google now does is basically retrieve that information for you automatically from your Gmail and then display it as a card at the right point in time right so if if you have a hotel confirmation it'll display that you know probably when you land in that city I think putting together information that's mine whether it's on my phone or not and information that is or the world's is absolutely critically important people are going to expect that of course I can have a conversation with Google and of course Google is gonna access more than just the public information on the web we're both gonna know when my flight is whether my package has gotten here yet and you know where my wife is and how long it's gonna take her to get home this afternoon and so on and so on and so on because of course people knows that stuff we wanted to make voice search like 10 times better speech recognition is the process of turning sounds into meaning and between those two there's many levels of processing you want to turn sounds into what we go for news from the phonemes you want to turn them into words from the words you construct sentences and then once you have a sentence you have a transcript of what you're said you have to turn that into something that's meaningful for the computer to interpret so if you're doing the search query you're asking for pictures of cats the the computer has to understand that your doing an image search for the word cats we've actually been using neural networks for variety of tasks because if you have a very large amount of data and you can train big models then you just throw more data at the problem and you'll start to catch up pick up on those more subtle refinements like Oh someone's speaking with a you know an East Coast accent play the scene where George says it's not you it's me so here it gives me a link directly into YouTube which I can click on every back end of Google every different web service that's been developed over the last you know ten years or so it's part of this service what Google now is is precisely what Google is best at how do you do heavy amounts of computation a large amount of data to extract patterns into X you know into score things I think what we've tried to do is kind of stay true to the in some sense you could say the personality or the persona that Google has today it has that whimsy and that that fun aspect inside of kind of unexpected ways but that doesn't really try to pretend to be your human bunny what is an Android in science fiction a robot with a human appearance