Three years ago, Jeff Bezos announced that drones are eventually going to deliver Amazon orders. In the past year, he brought out Amazon’s Alexa artificial intelligence service, which understands speech well enough that you can say, “Alexa, I really need a waffle cone maker,” and she’ll put one in your Amazon online shopping cart, even though nobody needs a waffle cone maker.
Both of these technologies—drones and cloud AI—are exciting today, yet still wobbly works in progress. But in coming years, Amazon or some other company is going to put them together. And that, finally, will evolve into a technology that could become as significant to humans as domesticated dogs.
Try Newsweek: Subscription offers
Right now, drones are more overhyped than a Kanye West fashion collection. At the end of August, the Federal Aviation Administration enacted rules so companies clearly know how to operate drones within the law. The FAA’s pro-innovation stance set off waves of drone exuberance, including predictions that 600,000 commercial drones would be operating in the U.S. within a year, creating 100,000 jobs and no doubt inspiring millions of PowerPoint slides proposing drone businesses.
We’re already seeing drones that shoot real estate video, pull advertising banners past crowds and scan crops for farmers. Sales of drones have doubled in the past year. Investment in drone startups has hit $1 billion.
Yet drones remain a fringe technology. They are tools for the specialist or toys for the man-child who wants to buzz the beach taking video of women in bikinis. Most of us rarely encounter a drone or have any practical idea of what we’d do with one. “There are still a lot of challenges we have to overcome as an industry to prove the value of drones, even outside the regulatory environment,” Gretchen West, a Silicon Valley drone consultant, told the Los Angeles Times.
The great limitation of drones is pretty obvious: They are pretty dumb.
Of course, pricey drones are loaded with computer chips, GPS and collision-avoidance software. But a person still needs to fly one, either by line of sight or by watching the video transmitted from the drone’s camera to a smartphone screen. The new FAA rules, in fact, dictate that a drone has to stay within the operator’s line of sight, because the agency considers autopiloted drones unproven and unpredictable. That constraint means a delivery drone probably can’t cart a printer ink cartridge—or an emergency order of cannabis—halfway across Denver.
A flock of pigeons flies with a prototype “parcelcopter” of German postal and logistics group Deutsche Post DHL in Bonn December 9, 2013. Though drones’ capabilities remain limited because of their need for human pilots, as automated drones controlled by AI are developed the number of practical for the burgeoning uses may expand greatly. Wolfgang Rattay/Reuters
But once drones come loaded with sophisticated AI and voice-recognition technology, humans won’t have to guide them. Such a drone could, in fact, operate a little more like your dog fetching the newspaper. (For millennial readers, a newspaper is news that was once printed on actual paper and thrown onto the lawn of every home in the neighborhood by a boy riding a bicycle. And yes, that does sound insane.)
Think for a second about how a dog does this task. First, you train it by showing it what you want it to do—no coding involved. After that, you’re able to say, “Rover, go get the newspaper.” It knows what a newspaper looks like and can find it, whether it has landed on the porch or in the birdbath. The dog picks up the paper and then knows what you look like and the layout of your house, so it can find you and drop the paper at your feet. That’s a complex series of events, and no drone today could do anything like it.
But add learning AI to a household drone, and imagine how it might affect the way we do things. Let’s say you name your drone Rover. You could tell it, “Rover, go pick up Daddy’s medication from CVS.” (If CVS is smart, it will have a drone takeout window tomorrow.) It would know where to go and how to get back to you. Or an AI drone could operate as a watchdog. Hear a noise outside, and you say, “Rover, go check it out.” It could zip around the perimeter and know that the person peeking in your window is a stranger, not your mother-in-law. It might even call 911. For that matter, Rover Drone could watch the house while you’re gone, a flying sentry. There are startups working on technology like this—still in stealth mode, so I’ve pledged not to say much more yet.
While ideas for dog-smart personal drones are whipping up tech entrepreneurs, the real action will be in commercial AI drones, which could affect just about every industry. If drones are going to deliver Amazon packages, Domino’s pizza, or rural mail for Canada’s postal service—and there are plans for all that—they will have to be able to navigate spaces and avoid people and recognize things even better than a dog might. Emergency workers could send a fleet of AI drones into a flood zone to autonomously search for people who need help. AI drones might zip around a construction site, bringing workers parts and tools that they ask for by voice. One futurist even imagines tiny AI drones working together to pollinate an orchard, doing the job of bees. The possibilities seem nearly endless.
The technology to make AI drones is emerging. Drones keep getting smarter. The biggest drone maker, China’s DJI, recently came out with its Phantom 4, which can recognize some objects and can be trained to follow a walking human on its own. (Like a dog!) One company, AirMap, is collecting data about every airspace on the planet, so any AI drone could reference that data and know where it can or can’t go for reasons such as privacy, security or safety. And since AI like Alexis or IBM’s Watson will reside in the cloud, drones with high-speed wireless connectivity will be able to get their smarts on the go from software that is constantly learning from every other drone using that same service. In other words, once drones start learning, they’ll learn fast—like a dog that could instantly know what every other dog already knows.
The only question is: Who will be the first to make this work for the mass market? Amazon? Google? Or maybe it will be a company in China or India, where there are fewer official restrictions on drones? Japan is ahead in developing empathetic AI robots that can bond with and help care for its aging population. Add some propellers, and you’d get a companion drone.
Dogs better watch out, or this “man’s best friend” thing will start sounding as dated as a paperboy.
