The discouraging news about Apple’s self-driving car project, code-named Titan, continued this week. The New York Times on Wednesday reported that the company has narrowed its focus from building an autonomous vehicle to building autonomous driving software for an employee shuttle. That shuttle won’t even be built by Apple, according to the Times’ anonymous sources: It will likely just be a commercial vehicle purchased from a major automaker.
This development will not shock those who have been following the embarrassing saga of Silicon Valley’s worst-kept secret. Titan, launched three years ago, has been on a downhill trajectory for at least a year. In October, Bloomberg reported that an Apple-made iCar was no longer in the works, and in June, CEO Tim Cook basically admitted as much. What’s new this week are the details about the self-driving employee shuttle, which underscores just how far behind Apple really is, how confused the company has become, and how it might find its way.
Historically, Apple has defined its products around specific pieces of hardware. Its greatest hits—the Mac, the iPod, the iPhone—were all physical devices thoughtfully designed to create an intuitive experience for the user. Those devices, in turn, gave rise to Apple’s most successful software and services, including iTunes and the App Store. It wasn’t crazy, then, to think that the iCar could be the next iPhone: the blockbuster product that sells by the hundreds of millions and transforms entire industries.
A driverless car, though, is about more than the car. It’s about artificial intelligence. And that’s where Apple ran into trouble.
Of all the new sectors Apple has entered under Cook’s leadership, A.I. software—the kind you need to build a self-driving vehicle—is the most important. It is likely to take over not only our cars, but our homes, our gadgets, and increasingly our jobs. But Apple is poorly positioned in the A.I. race. Whereas rival Google has always been about data and algorithms, Apple is a hardware company first. In the software realm, its strength lies in designing friendly user interfaces to go with its devices—not harvesting and processing the sort of gargantuan data sets that machine-learning algorithms rely on. It is further constrained in its A.I. efforts both by its strong stance on data privacy and its internal culture of secrecy, which is anathema to top researchers in the field.
That helps explain why Apple’s initial car plan involved designing and building a vehicle from the ground up. Google, Uber, and others have a long head start when it comes to the software, but very little experience designing and manufacturing machines that people want to buy. Perhaps Apple could gain an edge by marrying the software to beautiful hardware. Alas, Apple quickly realized that building cars is quite different from building computers, and it was at an insurmountable disadvantage in that realm, too. (Ultimately, it seems likely that our self-driving cars will be built by the same companies that build our current cars.) This leaves Apple in the same position it has been in for the past decade: as a maker of sleek personal computing devices, with the venerable iPhone as its flagship.
But the technology of the future will not revolve around discrete, self-contained gadgets that each work in their own special, clever way. Rather, it will center on the ethereal intelligences that float from one device to the next, animating each piece of hardware, and gathering data and refining understanding of you all the while. In the context of the car, the key to the future is not the machine itself—not the arrangement of the seats, the engine, and the wheels—but the machine-learning software that drives it.
What about in the context of personal computing, where Apple still reigns? The iPhone shows no signs of going away, and it’s possible to imagine that we’ll still be carrying some form of computing device in our pockets for decades to come. But the way we interact with our phones is already changing. Touchscreens and buttons are giving way to voice assistants such as Siri, Google, Alexa, and Cortana. Manual typing is being replaced, in some settings, with predictive typing and even “smart reply” features that automatically compose messages on our behalf. Passive web portals and apps are being supplanted by push notifications that make proactive suggestions.
All of these new features rely less on the sort of user interfaces that Apple is so good at building, and more on—you guessed it—machine-learning software. That offers an opening to just about every tech company that isn’t Apple to take over one element or another of our iPhones’ functionality.
Facebook’s main app already dominates much of the time we spend on our phones; its subsidiaries Instagram and WhatsApp are taking over the camera and messaging, respectively. Google beat Apple at maps, and its email, calendar, and browser are smarter, too. But the most existential threat to the iPhone’s central role in personal computing may have come from the unlikeliest of major Apple rivals: Amazon. Its Echo smart speakers, powered by Alexa, are showing the world that the phone may not be the ideal control center for the smart home after all. Why pull something out of your pocket and fiddle with buttons when you can just ask Alexa to turn on the lights, change the channel, tell you the news, order groceries, play music, or even make coffee?
The Echo—or Google’s rival, Home—can’t literally replace your iPhone, in the sense that you can’t carry it around in your pocket everywhere you go. But you can carry Alexa or Google Assistant anywhere: The software behind the Echo and the Home could theoretically serve as the brains behind any number of devices, including your phone. (Amazon is already offering Alexa on its Fire TV; Google is now offering its Assistant as a standalone app for the iPhone, as well as its own Android devices.) The more you interact with this software, the better it gets to know you, and the more incentive you have to use it as your primary portal to the digital world.
Apple can make its own version of the Echo, of course, and it announced on June 5 that it is doing just that. It will be called the HomePod, and it’s entirely possible that it will be a better piece of hardware than either the Echo or Google Home. But as with the self-driving car, Apple won’t be able to win on hardware and user interface alone. The smart speaker, like the vehicle, is just a dumb vessel for the artificially intelligent software that drives it.
Apple, of all companies, should have seen this coming. Siri was on the very leading edge of artificially intelligent assistants when Apple bought it back in 2010. But Apple didn’t realize what it had. The company saw Siri as a nifty feature to improve its flagship device, the iPhone. As such, it never made the kind of investment in Siri that would have been needed to turn it into something much bigger. But other companies saw Siri and thought: We can do that. And so they did. Each of the big five tech companies now has its own A.I. shop and its own A.I. assistant, and some of them have arguably surpassed Siri in capabilities.
More importantly, Apple’s rivals—Amazon in particular—thought more creatively about the type of devices an A.I. assistant might lend itself to. That is, they thought of the software’s capabilities first, then worked backward to the type of hardware that would be needed to maximize them. And so, while Apple was busy building tablets and watches that worked like iPhones, Amazon went out and built a brand-new type of device that worked very differently. Ironically, Amazon’s smart speaker has a better chance of being the “next iPhone”—in the sense of a gadget that changes personal computing—than anything Apple has done.
There are signs that Apple is finally starting to appreciate Siri’s true potential. Its AirPods, launched last year, represent a potentially transformative new interface, not only for the iPhone, but for any device that Siri can power. As I wrote when Apple announced them, everyone who dismissed the AirPods as Bluetooth headphones was missing the point: They’re actually a portable version of the Echo, a voice-powered smart speaker that you wear in your ear.
Apple has also given Siri a more central role on both the iPhone and the Mac. And, belatedly, it is getting serious about developing the world-class artificial intelligence needed for Siri to compete with Google, Alexa, Microsoft’s Cortana, and whatever it is Facebook’s A.I. wizards eventually cook up.
Apple’s future, in short, is not a device—not a small one like a phone or a big one like a car. It’s Siri. And it needs to be smart enough to follow you everywhere, just like your iPhone follows you everywhere today. The fact that Siri already lives on the world’s most popular phone should give it an advantage that no other company’s virtual assistant can match. It should already know you—your habits, your preferences—better than any other.
Which brings us back to the car. Because Apple’s way into the self-driving business isn’t through the engine, the steering wheel, or the computer vision system. It’s through Siri.