Can Samsung's Bixby Smash Alexa?
SEOUL—In the battle of the virtual assistants, Amazon has friends and Google has search. Samsung has your TV and washing machine.
Samsung's Bixby AI system, coming on the Galaxy S8 phone which launches on March 29, starts out as voice commands for your phone. But its end goal is to tie together all of your home electronics—Samsung or not—with natural-language voice interfaces. The question is whether Samsung can spin its plans up quickly enough to chase Amazon and Google, or whether Bixby will languish on phones as the next S Voice.
We spoke with Samsung Mobile's CTO, Injong Rhee, at Samsung headquarters in Korea. Rhee, whose leonine curls make him look like a 1980's pop star on a reunion tour for screaming moms, has a big, long-term vision for Bixby, which is fully supported by Samsung mobile chief DJ Koh.
Bixby isn't named after the TV actor best known for playing Bruce Banner in the Incredible Hulk. Samsung just came up with a bunch of names that sounded good, and "Bixby was the No. 1 choice among the millennials," Rhee said.
Coincidentally, the Bixby Bridge by Big Sur is a California landmark, so the name "has the connotation of bridging machine and human."
It also helps that the "X" in the middle of Bixby makes it easy for computers to understand the word, and that it jibes with another Samsung product, Knox.
Bixby defaults to a female voice, and initially works in English and Korean; US Spanish will follow, then other languages. It may also have a male voice option, but that wasn't clear.
Speak to Me, Bones
Initially, Bixby is a voice interface for the built-in features on phones. It also does some computer-vision tricks, translating languages and identifying places in photos shot with your phone. But the full idea is an "intelligent user interface" to let you control all the features of home electronics.
"Right now our focus is on device interface and application controls," Rhee said. For internet-based queries, the next Galaxy phone will presumably also have the Google Assistant, but there was some back-and-forth about whether Bixby would be able to hand off web queries to Google Assistant in the future. It sounds like Samsung's working on it.
"It starts off with smartphones, but anywhere that has an internet connection and a microphone, Bixby can be used. It can be in any devices we produce: appliances, TVs, refrigerators, washing machines," Rhee said.
Before you wonder about washing machines, think about the complicated dials are on today's machines. Then think about just standing in front of it and commanding it to "wash a large load of lights on hot and tell me five minutes before it's done."
"I always go to my washing machine and say, 'Hmm, what do I do?' That's a limit of the interface," Rhee said.
Samsung envisions Bixby as working across networks of devices, probably linked through Samsung's SmartThings Hub (above). LG is trying to make a similar vision happen with Amazon's Alexa and its own SmartThinq hub.
That way, you can say, "'What's the email message I just received, and can you show it on my TV?' That capability might not be on the TV, but it may be in your smartphone, and you're doing it with the TV remote control," Rhee said.
Press the Button
In an era when we're getting nervous about always-listening devices, Samsung has a slightly retro solution: a button. By default, Bixby won't listen until you press the button.
A Bixby button could be a fingerprint sensor on the front of a washing machine, verifying a user's identity and waking up the voice interface. Or it could be on a remote control, the way it is with Amazon's voice-activated Fire TV.
"How can you make sure [the user] has control of the data? That's why the button is so important. We're not going to listen to the user unless the user does something with the hardware button," Rhee says. "Our plan is to have that button almost everywhere, including our appliances."
Samsung's possible great leap forward here—if it works—is in applying true, natural language voice interfaces to local commands on electronics. Alexa, Siri, Cortana, and their ilk still think like computers, looking for specific trigger words and say "I don't understand you" if they can't parse a sentence. Bixby is supposed to react in a more human manner, Rhee said.
"We know it's really cognitively challenging for people to understand exact commands," he said. So you could say, "show me a picture that I took in Paris. Turn that picture around. Share it with Michael," and Bixby would get as far as it can. If it can't figure out who Michael is, it'll show you a list of possible Michaels and transition to a more traditional touch-screen interface.
"For a fairly convoluted or complicated task, a user has the freedom of speaking it in any manner and stopping in any place," he said.
Samsung recently bought Viv Labs, an AI firm started by the creators of Siri, with natural language processing chops and an open third-party application platform. But the first version of Bixby won't integrate Viv's technology, Rhee said. Now we get to Samsung's greatest danger.
Version 1.0 Troubles
Bixby needs to take some inspiration from its non-namesake, because to battle Amazon's wonder woman, it's going to need to hulk out—fast.
While Amazon's Alexa has hundreds of third party "Skills," and both Google Assistant and Siri have a seemingly endless array of knowledge cached from the internet, Bixby must be tied into specific applications and devices.
And Bixby's initial feature set, while fine for a 1.0, is definitely a 1.0. While the full details haven't been released yet, I'm concerned that the first Galaxy S8 owners will feel it's too much like the long-unloved S Voice at start, and that assessment will stick in people's minds.
Therein lies the danger. Viv's natural language technology? Coming. Third-party app support? It's coming. Support for internet-based search queries, or smoothly handing off to Google Assistant for search queries? Coming. Those promises are all well and good, but Samsung is starting in third or fourth place, and can't afford to delay differentiating features.
Rhee wants Bixby to become a broad standard, on the back of Samsung's market dominance in phones. Samsung reps say that with the new leadership of Rhee and mobile CEO DJ Koh (who's been in the role for about a year), the company is supporting its services much better than it used to. Against the older story of Samsung's failed Milk Music service and struggling Gear third-party apps platform, they put Samsung's newer, successful Knox security system. Bixby is of the new era, they say.
Samsung's position is inherently difficult. No Android manufacturer in a Google Play-supporting country (in other words, not China) has developed services that have really driven sales. It's telling that the four big voice interface companies—Amazon, Apple, Google and Microsoft—all control their own OS platforms. While that's true of Samsung's appliances, on phones it has to deal with the push-and-pull of being dependent on Google for its software.
"There's a revolution in interface technologies, and the operating system becomes less relevant as time goes on," Rhee argued.
That said, if Samsung opens this up big, and fast, it may have a chance. "I don't see it as so difficult to have it on other devices, even competitors. That's a future we have to think about, it's not completely out of the question," he said. "It personally makes sense to me."
This article originally appeared on PCMag.com.