Say, “Alexa, do I need an umbrella?” A female voice with rounded, vaguely Canadian vowels responds, “No rain is expected in San Francisco today.”
There’s no gadget dream greater than this: Your wish is my command.
My electronic genie lives inside a Pringles-can-sized $200 speaker called the Amazon Echo. Just call her name—actually, Echo’s wake-up word—and this virtual assistant scours the Internet for an answer to whatever you ask. At least, she tries.
Alexa is Amazon.com ’s answer to Apple’s Siri, Googke Now and Microsoft's Cortana, which all come standard with different smartphones. She’s the latest sign that big tech firms believe our future involves talking to computers that can talk back.
But spend a little more time with Alexa, and the dream of a virtual assistant in a bottle begins to fizzle. “Alexa, can you give me a recipe for chocolate chip cookies?” I asked. “Sorry, I didn’t understand the question I heard,” she said. “How long will it take to drive to the office?” She couldn’t answer that, either.
In fact, what Alexa can’t answer far outweighs what she can, so her novelty wears off fast.
The Amazon Echo has a few good ideas about how voice control might be useful in our homes. But I can’t recommend the Echo to more than the curious. (Amazon makes customers join a waitlist to buy an Echo, and doesn’t let owners post reviews.)
Echo’s website says the device “continually learns and adds more functionality over time.”
Using Alexa mostly left me wondering: In 2015, what’s holding back all virtual assistants from being more useful? And does Amazon actually have a shot at making one any better?
Despite a few improvements since Siri made its debut in 2011, all virtual assistants feel stuck in that trough between what we can imagine and what technology can actually achieve.
Context is hard
Voice recognition isn’t the challenge—that’s the easy part now. The problem is understanding what we mean, the context of our requests.
“Virtual assistants today do not have effective memory,” says Norman Winarsky, vice president of ventures and a co-founder of SRI International, which developed Siri and sold it to Apple. Each time you talk with one, it’s like you’re meeting them for the first time, he says.
One of his favorite virtual assistant tests is to say, “I would like to see ‘Her’ tonight.” Can the machine understand whether he means the movie or a person—and if a person, which one? (In my test, Siri automatically assumed “Her” was a movie; the other systems just got confused.)
Virtual assistants have taken a few baby steps in understanding context. You can now ask Siri, “Any good sushi places around here?” and she will use your location to make recommendations. Then if you change your mind and ask, “What about Mexican?” she remembers that you just asked about restaurants.
Cortana helpfully asks for more information about the people and places that matter to you. You can tell Cortana to “remind me the next time my dad calls to tell him about the trip,” and it will know which call to flag.
Ultimately, what’s holding back the Echo is that Alexa doesn’t have access to my contacts, calendar and email, or where I spend my time.
Short of skills
Alexa stands out from the rest because she isn’t just a feature on a phone. She lives in your living room or kitchen, so everyone in the family can use her, hands-free. The Echo is always on and waiting for someone to call out, “Alexa.” (If you live with a real Alexa, you can change the wake-up word to “Amazon.”)
In my house, the Echo had one skill I loved: It is a voice-activated DJ. When I say “Alexa, play Beyoncé,” it starts a playlist of songs from my own Amazon collection. It can also play from iHeartRadio, TuneIn and Amazon’s Prime music service. Say, “Alexa, louder” and she pumps up the volume. Say, “Alexa, stop,” and it’s off. I’m more convinced than ever that voice, also a feature on speakers like the Aether Cone, is how we’ll soon control music, TV and a whole lot of other stuff around the house.
Voice commands aside, as a $200 connected speaker—it’s $100 for Amazon Prime members for a limited time—the Echo is just OK. You can connect a phone, tablet or computer to the Echo via Bluetooth to play music from Pandora, iTunes or Spotify. The Echo sounds fuller than the $130 battery-powered Jambox Mini, but fills the room far less than a $200 Sonos Play:1 speaker.
The Echo’s other impressive skill is hearing you, even from across the room. It feels snappier than other virtual assistants, though every once in a while it just wouldn’t catch my request. Amazon ships it with a remote control that you can speak into, but that defeats the whole purpose of the hands-free Echo.
The bigger problem is that Alexa quickly runs out of skills. She can tell you the weather, play a news summary, read brief Wikipedia and dictionary entries, make a to-do list and tell you the time in Dubai. In the kitchen, she can set a timer while you cook, and do quick conversions. But she can’t read you a recipe or answer questions about ingredient substitutions.
But even tasks you’d think Amazon would be great at seem to elude Alexa. It can build a shopping list, which you can see in the companion app for Apple and Android phones, but can’t follow through and complete any purchases on anything but music Amazon.com. She can’t read a book aloud…or even recommend one. (Neither can Siri, Google Now nor Cortana.)
She also doesn’t connect with any other apps, let alone any connected home products like the Nest thermostat.
Amazon says Echo is open to outside developers, but Cortana can already control smart home systems like Insteon, and Siri will soon star as the master controller of Apple’s HomeKit, which will connect to lots of different smart home devices.
Amazon is just getting started with the Echo—it even rolled out an update with improved features last Friday. But given the challenges, it’s clear that Alexa has a ways to go to live up to her potential.