Heath Whaley was running late to the airport, as usual. Zipping through nudie scanners at airport security with minutes to spare, he made a quick stop at the restroom, and began searching in his bag for his boarding pass to check his flight departure info and gate number. But it was nowhere to be found. Vanished.
"I couldn't remember what flight number I had, and didn't have time to dig through emails to find the flight number," Whaley recalls. He pulled out his iPhone and tapped on Tempo, a smart calendar app he uses. After you grant Tempo access to your email and calendars, the app searches for all the tidbits of schedule-related information you have stored in your accounts, gathering it together and presenting it cleanly inside individual calendar events. All Whaley had to do was tap the entry for the day's travels, and the flight numbers and gate information were right there, saving him from potentially missing his flight.
Tempo is but one of a growing number of "smart" apps that use artificial intelligence algorithms in order to give you a more efficient and more personalized mobile experience.
For those of us like Whaley, whose lives are generally a bit chaotic and disorganized, an app like Tempo can be a life-saver -- as long as we're willing to sacrifice some of our privacy. You see, in order for an app to gain the intelligence required to become a sort of personal assistant, it needs to know a lot about you. Who your friends are, what's on your schedule, where you work, where you live, and what subjects interest you the most. You have to hand over a few passwords and fill in some private details. But once you've relinquished that data, you can get a vastly richer and more valuable experience from your consumer device. It's almost like your phone becomes sentient.
And -- the hallmark of any AI-based app -- the more you use it, the better it gets.
Take Tempo, for example. For an event like “Meeting with Robert at Mexico Au Parc,” the app would bundle together Robert's contact information, any email correspondence about the meeting and any attached documents, the location information for the restaurant Mexico Au Parc, and directions on how to get there. It would package all of this into one entry within the app, along with buttons to easily send a message or email Robert if you wind up running late to the meeting. The benefits are clear. With a "dumb" calendar app, you’d need to scour multiple apps to get all that information -- the calendar, email, messages, contacts, maps and directions. Here, it’s all in one place.
Tempo’s not alone in the smart calendar space. Other players include Sunrise and the latest entrant, Any.DO’s CAL app.
But that’s just one segment of the AI app sphere.
Google Now, built directly into newer Android devices and available to iOS users via the Google Search app, is a like-minded app that aggregates all sorts of information relevant to your needs and interests. It uses all the data accessible through your Google account and from the sensors in your smartphone (you give it permission to see both), and it provides all sorts of helpful guidance, showing you weather in your area, sports scores, package tracking info, breaking news stories, and traffic alerts for your daily commute. Everything is presented cleanly is a colorful interface that resembles a stack of cards. Google Now can also surface travel information, like boarding passes, and it can help you find hotel reservations, calculate foreign currency transactions, and even help with on-the-spot language translations.
AI is also prevalent in a number of event discovery and activity recommendations apps, like Weotta. This app tracks affinities between a user's interests, upcoming events, top places in the area, and the interests of their friends. It combines this information with other things, like natural language phrase extraction and user location, to deliver timely, relevant, and socially engaging suggestions of things to do.
An app called Triposo uses machine learning techniques to help predict what you’d like to do when you’re traveling -- like a smart travel guide.
“We're trying to approach travel guides from an algorithmic, Google-like perspective,” Triposo co-founder and COO Richard Osinga tells Wired. Triposo uses what it calls an "opinion mining" algorithm. The company analyzes the natural language used in online reviews to determine whether people who have posted about a particular place liked it, and what exactly they liked about it. This helps the app suggest places for very specific qualities -- like a restaurant that has spectacular Bolognese, or a hotel that is especially clean.
It also uses the time of day, your GPS location, and local weather to suggest things to see and do while you’re traveling. That, paired with analysis of the behaviors and opinions of other users, lets Triposo figure out what activities people are most likely to be interested in at a certain time -- you’re probably not looking for a history museum at 2am in in Paris -- and how far they are willing to travel to do that. This means you can nix all the planning you’d normally stress about before a vacation, and be confident that you’ll find unique, interesting attractions no matter what part of town you’re wandering around.
But the ability for the apps on our smartphones do this sort of AI and on-the-fly processing wasn’t feasible even just a few years ago.
According to Tempo CEO Raj Singh, there are four factors making today’s smart apps possible. First, storage and computation is cheaper than ever before. Lower server costs and cloud computing make including AI an affordable option for startups. There’s also greater openness and portability when it comes to data, thanks to sign-in technologies like OAuth and more open APIs for developers to use. Indeed, for Weotta, significant enhancements to MongoDB, Redis, and Facebook Open Graph are what make it possible for the app's developers to come up with increasingly better recommendations at a fraction of the computational overhead it would have taken in the past.
Another important factor: Many app users have accepted the tradeoff that they can get a better user experience in exchange for a little less privacy. The idea of "handing over the keys" isn't as scary -- a few years ago, linking your Facebook, Gmail, Twitter, and LinkedIn accounts to a single app would have been unheard of. Lastly, the improved mobile ecosystem in general makes AI apps possible. That is, the numerous (and more advanced) sensors embedded in each device paired with a super-fast, persistent 4G LTE or Wi-Fi data connection allow apps to gather a wealth of information and ping their own servers, pushing relevant information to your phone's screen in a snap.
But this is just the beginning of artificial intelligence infiltrating our mobile lifestyle. It can also be used to enhance other experiences in the real world. Bay area startup Anki, which debuted with a small amount of fanfare at WWDC in June, has shown us a glimpse of this future: How machine learning algorithms can go beyond just a computing device to actually power and give personality to toys. In Anki's case, the company has made toy race cars that know when to pass opponents, when to speed up, and when to hit the brakes to avoid crashing out of a race.
As for today’s apps, the AI-based experiences aren’t perfect yet. If you’ve got multiple contacts with the same name, Tempo may initially get confused with which details to pull up. Or if your friends are interested in sports and you’re not, you may have to toss a bunch of recommendations in Weotta until it learns you don’t care about going to the next baseball game. But by and large, AI apps are blossoming. And with their growth, the stress and time involved in planning events, travel, and even your day-to-day errands are diminishing.
