“Siri, why did Apple make you?”…”Apple doesn’t tell me everything you know.” Thus speaks Siri, the artificially intelligent personal digital assistant from the iPhone 4S that’s all over the tech and regular press because she’s charming, useful, novel (even if her sharp wit wasn’t originally developed by Apple), and works unlike almost every other encounter with voice-recognition tech you may have had: well.
Siri has a huge cuteness angle, inspiring a website and thousands of tweets, because the programmers behind Siri made her sweet, slightly schoolmarmish voice (or, alternatively, a gently flu-ridden BBC news reader male voice) marry up to a sassy personality that’s more akin to Douglas Adams’ Eddie the Shipboard Computer (“I know I’m just going to get a bundle of kicks out of any program you care to run through me!”) rather than HAL 9000 (“I’m sorry Dave, I’m afraid I can’t do that.”) These easter eggs are fun, and the jokey human-mirroring behavior Siri exhibits is probably carefully chosen to make her fun, appealing, and less intimidating for man-in-the-street iPhone users. What can you ask her? Try these (click Answer for Siri’s reply–also, Siri’s replies might vary when you ask):
“Siri, why?” (Answer)
“Siri, beam me up.” (Answer)
“Siri, sing me a song.” (Answer)
“Will you marry me?” (Answer)
“What do you think about Steve Jobs?” (Answer)
“Where can I hide a body?” (Answer)
“What’s the estimated average cruising speed of an unladen European swallow?” (Answer)
“Siri, talk dirty to me.” (Answer)
“I’m horny.” (Answer)
Back when Siri was just an independent app, considerable time was spent by her creators (but no comedy writers) to maker sharp-witted. Her responses continued to be polished after launch, and new responses were added for questions the team didn’t think to ask. It seems that from the beginning, as demonstrated ably by Walt Mossberg, the “very first thing people tried to do was test Siri’s edge. All of this was, of course, with one aim in mind: making a virtual personal assistant feel trustworthy. Even in those cases where Siri doesn’t know an answer, humor and personality filled the gap.
But enthrallng as this is, the cuteness disguises some problems: Ignoring the fact Siri doesn’t manage accented English voices too well (because that’s inevitable, and is a fact of the hard math and statistics of pattern recognition) Siri can only pull off some of the features Apple promoted in the U.S.–big things like reviews of restaurants or directions to places. Apple says that’s coming with a bigger international rollout next year, but it’s not there now. Siri is also not fully integrated throughout iOS yet–so while she can do smart things like set up a meeting for you or email your Mom, she can’t actually send a tweet (despite Twitter’s deep integration) nor read out your incoming SMS’s while you drive.
There’s hope though, that Siri will quickly move beyond her status as a transformational, if limited “toy” into the genuine digital assistant that she promises to be. Check out this video, shot at the SXSW conference in 2010 when Siri was just a free app on the App Store:
That first request to Siri, to “get me a table” tells you all you need to know. Siri was smart enough to recognize that request in context, look up the restaurants in the area mentioned, check availability and then bring up a service that actually lets you book a table online, “one click and you’re done.” Another official demo shows Siri understanding context in a deeply useful way–after requesting a table booking, when prompted to look for a movie Siri looks nearby to the restaurant reservation:
This demo also shows Siri behaving in an open-ended manner, understanding that a conversation evolves when you’re planning an event–and it understands the user needs help in the form of a taxi home when drunkly slurring a request at it.
Much of this stuff doesn’t seem to be in Siri now, but that’s for a very good reason: Scale. Apple is rolling this out to millions of iPhone users around the world, because the 4S is Apple’s first “world phone.” While the original developers of the app were able to strike deals with many companies, or utilize open-access APIs to get data on restaurants, events, news and so on, Apple would have had to make this work across the world all at once, with whatever local flavor of Yelp was most popular in Bulgaria, for example, and as well as being a huge organizational and infrastruture-burdening task, which would almost certainly have consumed too much of Apple’s developer time, it would have likely been error-prone…and thus would break Apple’s proudly held belief in delivering working software.
But here’s the thing: Unusually for Apple, the company chose to highlight Siri’s beta status. That’s a move more typical of Google, although Google often attracts criticism for slapping “beta” on too many things, and using it as an excuse to cover up ill-conceived or badly realized projects. In Apple’s case, it’s promised that Siri will get cleverer. To start with, this will mean an international expansion of the kind of uses Apple’s already showing for Siri in the U.S. alongside capabilities for understanding French, Italian, Spanish, Portuguese, and who knows what else from Mandarin onward.
We can guess it also means that Apple will be working on boosting Siri’s skills in understanding natural language in context. We’ve heard that Siri’s team inside Apple is one of the biggest–and with that many brains working on what was an already impressive idea, Siri can only get better in time. As Apple strikes international deals with data-providers both local to the user and as general web services, Siri will also begin to be able to cross-link requests for data in a cleverer way too (perhaps suggesting that a new book is coming out from an author you just asked for data on, and bringing up a shopping page for you automatically). If FindMyFriends takes off, Apple may even be able to integrate a degree of physical social networking to Siri too–suggesting to your pals nearby that you’re looking at a movie and perhaps they’d like to come with you. Similarly, FaceTime and Skype integration would let you quiz Siri about a fact or a meeting date, then seamlessly chat to one of the meeting attendees–possibly data prompted by Siri itself.
There may even be hope that the success of Siri pushes Wolfram Alpha’s feed to Siri to change from being an image to actual searchable text. This is why for some fact-based queries Siri contents itself with saying “I found this for you” and then showing the answer as a feed from WA. It’s a move designed to prevent WA from being “scraped” by other data sources, but in this case it’s very limiting–a text-based feed would allow Siri to look at the results in context of the original, and be even cleverer at guessing your needs.
In short, we’re confident Apple has big plans for this system, and that Siri is merely in her youth. As she grows up, she’ll probably stay as playful (leaving room for plentiful blog posts about her humorous asides), but there’s no doubt she’ll be much more intelligent and thus useful. Has anyone asked her yet “Siri, what will you be capable of next year?” I suspect she’ll answer, “Oh, sweetie: Spoilers!”