Upon first blush, Siri, the virtual assistant built into the iPhone 4S, is certainly impressive. You speak naturally, and Siri figures out what you’re trying to say, and then does just what you want it to. Well, most of the time, anyway.
The more you use Siri, though, the more obvious its shortcomings become. Here are a few improvements we’d like Apple to make in Siri 2.0—or whenever it decides to pull down that unsightly “beta” label.
Change Settings: Ask Siri to turn off Wi-Fi or turn on Bluetooth, and it simply won’t understand. Of course, asking Siri to, say, enable Airplane Mode could potentially prove troublesome, since the technology requires an Internet connection to work its magic—meaning you couldn’t necessarily use Siri to turn Airplane Mode off once it’s enabled. But navigating through the Settings app to toggle Bluetooth off or on requires a minimum of four taps—it seems like just the kind of task Siri could handle with aplomb, if only its developers would teach it how.
Tweet Maker: When you attempt to post an update to Twitter via Siri, the assistant knows what you want it to do, but still declines: “I can’t send Tweets for you. Sorry about that.” Given that iOS 5 includes systemwide Twitter support, and given that Siri already understands the instruction—even though it refuses to follow through—this seems like an easy improvement to make. We’d also be delighted if Siri could take a cue from Tweet Speaker and read tweets from your timeline aloud—sometimes you just want to catch up on Twitter while you’re driving.
Notification Integration: We have high hopes that Siri will one day support some sort of integration with non-Apple apps (see below). But Cupertino could enable some degree of such support without requiring third-party developers to write even a single new line of code, if only Siri could read incoming notifications when you asked it to do so. Perhaps a single instruction to Siri—such as ”Read new notifications as they arrive”—could let you enable the mode when you’re in the car; “Stop reading new notifications” could turn the feature off again. That way, you could keep your eyes on the road and know the moment it was your turn in Words With Friends.
Email Reader: Sure, Siri can compose emails for you, and even display your unread emails (or an email search). But try to get Siri to read you even just one offer from a hapless overseas banker looking to cash you in on an inheritance, and you’ll be in for disappointment. Given that Siri can already read text messages, it seems like it wouldn’t be a big leap to add emails. Understandably, if someone’s sent you a ten-page email, or one that’s heavily formatted in HTML, speech might not be an ideal medium for consuming a message, but perhaps over a certain length, Siri could give you the option to just hear a preview. And who knows, maybe some day Siri will be smart enough to just summarize the salient details. It’d also be nice if Siri could be instructed to flag emails and to display your flagged messages.
Read It All: Speaking of reading your own data aloud to you, it’d be great if Siri could read details from other stock Apple apps, like Notes and appointments in Calendar, when you asked it do so. Siri says “I can’t read your notes to you.” Can’t, Siri? Or won’t?
Better Reminders: Reminding yourself to do things using Siri is quick and easy—but it could be easier. For one thing, you can’t use Siri to change reminders once you’ve created them, nor can you mark them as complete. And if you want to add milk, orange juice, and bananas to your shopping list, you have to do so separately, confirming each one in turn. But Siri has shown itself able to distinguish multiple items in the case of recipients for text messages and emails, so why not here too? It would be far easier to say “Add milk and bananas and orange juice to my shopping list.” Well, at least for the human doing the asking.
Launching Apps: It’s easy to launch the apps in your iPhone’s dock or on your main home screen. Once you load up more than a screenful or two of apps, however, finding the right one becomes a bit of a slog—even if you rely on Spotlight, it still means tapping out the first few characters of the name of the app you’re after. Right now, if you instruct Siri to “Launch Photos,” it responds: “I’d like to, but I’m not allowed to.” If Siri’s overlords in Cupertino could instead grant that permission, launching tucked-away apps would quickly become as easy as saying “1, 2, 3.” Though we’re not sure what app that would launch.
Easier Transcription Correction: If Siri gets the text of your reminder wrong after you dictate it, you can simply say, “Change that to ‘Buy milk,’” and Siri makes the correction. But if you dictate an email of more than a few sentences, and you see that Siri understood the gist of it, but mangled a few words and phrases, re-reciting the entire message feels tiresome. Sure, you can tap into Siri’s transcriptions and make the necessary edits, but ideally, you’d tell Siri just what to fix: “Change ‘elephant shoe’ to ‘I’ll have that, too.’”
Automatic Punctuation Transcription: Look, we’re not saying every item on our Siri wishlist is a simple matter of Apple pressing a few buttons—sometimes it’s a matter of it pressing a lot of buttons. When we dictate our punctuation-filled missives, our voices still change with the punctation. That is, when we say something like, “Are you sure question mark?”, our voices go up for “sure” and “question mark” alike. Apple’s engineers should work to expand Siri’s transcription abilities, so that it can translate such pitch changes to question marks, short pauses to commas, and longer ones to periods—contextually, of course. We fully admit that this entry seems like something out of science fiction, but then so does much of Siri’s current functionality. And, hey, we’re not asking for the moon: For example, Siri can ignore semicolons, because most people do too.
Add and Edit Contacts: Ask Siri to add a number to a contact, or to create a brand new record in your iPhone’s address book, and it’ll balk: “Sorry, I’m not allowed to create contacts.” Most of the time, Siri can’t edit them either—although you can tweak relationship data for your own contact record by saying things like “Gregory House is my physician.” Those tasks can be cumbersome, since they often require so much tapping. Whereas simply saying “add email@example.com as Gregory House’s email address” takes just a second or two.
Default Address: Speaking of Contacts, the iPhone and Siri should provide a way to set an email address as a default for a given contact. Right now, when we say, “Make an appointment for 1 p.m. tomorrow with Jason Snell,” Siri inevitably asks which of Jason’s addresses to use. Invite multiple people to an event, and you’ll spend a while confirming each of their addresses, which somewhat lessens Siri’s magical nature. We need an option that says “Always use Jason’s work email address unless I specify otherwise.”
Safari Surfing: Siri can’t yet open an exact URL (“Go to www.macworld.com”), or launch URLs from saved bookmarks (“Go to Macworld.”). Instead, we have to tell Siri to search for Macworld, and then interact with our Web browser. For an intelligent agent that has the whole Internet at its disposal, that seems silly.
Third-Party Integration: We’ll admit, this request is a trickier one. For Apple to truly allow third-party developers to integrate their apps with full-blown Siri support won’t be simple. You’d need a system for determining which apps should handle specific requests. For example, if you said “Check me in here,” should Siri hand that request off to the Foursquare or Facebook app on your phone? And if you want to change which app handles such requests later, then what do you do? These are solveable problems, though it’s understandable that Apple will almost surely choose to tread lightly if it goes down this path. But you only need think of the convenience of asking Siri by how much the Red Sox are beating the Yankees to get the idea of how great this integration could be.
More Appless Third-Party Integration: Siri users already benefit from its integration with services like Yelp and Wolfram Alpha, even if they don’t have their apps installed. Surely there are more services Apple should add: OpenTable for making reservations at the restaurants Siri finds, a movie showtimes service, a TV programming guide, and, of course, integration with WebMD for all the medical needs that Dr. House just can’t cover—“Yes, Siri, I know it’s not lupus.”
Media Savvy: While Siri’s advantage over iOS’s earlier Voice Control feature is clear when it comes to playing back music—for example, Siri can summon a particular song, not just an album, artist, or playlist—there’s still room for improvement. For example, tell Siri to play a track and it will graciously oblige, but once that song’s over, the music will simply die. Instead, let us tell Siri to play a song and then shuffle music onwards from there. And while we’re at it, why doesn’t Siri understand a command like “Play the latest episode of the Sesame Street Podcast”? Is it making some sort of judgment about our podcast-listening habits?
TV Talk: We all enjoy screaming at our TV every once in a while, whether it be because we’re watching our favorite sports team do something stupid or because that idiot insists on checking out that mysterious noise in the basement. What if we could channel that vocalization into something productive? Like telling Siri to mute our Apple TV when the phone rings. Or using it to search for a title on Netflix or the iTunes Store without having to rely on slowly entering text with an Apple Remote. Or telling it to play the latest episode of a podcast. The iPhone already makes a great Apple TV accessory; letting Siri into the mix would make a good thing even better.
[Staff writer Lex Friedman and senior associate editor Dan Moren talk to Siri as often as they speak to each other.]