Signs that Samsung Bixby 2.0 Will be Smarter

A better Bixby was definitely a focus of the Samsung Note 9 launch and device unpacking event in New York earlier this month. Samsung received a lot of criticism for the delayed launch and lackluster consumer response after Bixby’s initial rollout in 2017. Consumers saw it as less capable than Google Assistant. However, since February Samsung executives have been saying in effect, “just wait until Bixby 2.0.” Bixby’s improvements were overshadowed during the event particularly by the Galaxy Home smart speaker debut, but there were some notable features demonstrated. Ji Soo Yi, vice president of AI strategy showed off some of the voice assistant’s new capabilities. He began by saying:

From the beginning Bixby was designed to help you get things done.

Understanding and Maintaining Context

Ji began the demonstration saying, “I need a concert in Brooklyn over Labor Day weekend.” Bixby responded with suggested concerts in Brooklyn “during the Friday, August 31st weekend.” Bixby knew that Labor Day weekend was associated with specific calendar dates. Not mind-blowing, but a good first step. Ji commented, “Bixby already knows that Labor Day weekend begins August 31st so that he shows us concerts in that period of time. That’s good. Let’s try another date. How about the first weekend in October.” Bixby then responded, “I found these concerts in Brooklyn during the Friday October 5th weekend.” Ji adds, “So, Bixby remembered that we were talking about concerts in Brooklyn. That means that I don’t need to repeat myself. This is what I mean by more conversational.”

Bixby twice associated the correct dates with a request that didn’t mention a date other than the term “weekend” and the month of October. And, more importantly Bixby maintained state and context. Voicebot was first to report that Alexa began maintaining context in March of this year. Context maintenance is also available for for Google Assistant. Ji calls these elements conversational because the situations reflect how people speak. Your conversation partner is likely to remember topical context as you continue to interact. Context maintenance is another useful improvement for Bixby.

Context Related to Personalization

“When you talk to your close friends, they know who you are and what you like…For example, ‘I need good restaurants in New York.’ Bixby knows I like French food so it listed French restaurants at the top,” said Ji.  He continued, “Lets check it out. It looks great… I want to make a reservation…You can see Bixby has already filled in the number of people in my party and a reservation time based on my previous bookings. All I have to do is tap the reserve table button…That’s it. This is what I mean by more personal.”

We can see from these feature demonstration that Bixby is designed to be personalized based on past behavior. This is something that Amazon and Google only do loosely today. So, it may be that Bixby could have a short-term advantage in this area if it launches 2.0 soon and it actually works.

A Post App Interaction Model

Ji went on to show one more demonstration. “This is what I mean by more useful. ‘I need a ride to JFK,'” said Ji addressing Bixby. Bixby responded, “The fare for this ride will be 49 dollars, 11 cents. Ready to request ride?” “With just six words and an Uber account, I can order a car to the airport. I never have to open an app, explain myself or fill in the details. This is the power of conversational, personal and useful Bixby,” concluded Ji.

This ride ordering demonstration along with the restaurant booking are probably the most forward looking aspects of the demonstration. Semantic understanding of requests, context maintenance and context personalization are all important features and a significant step forward beyond what voice assistants have been able to do to date. However, they do not fundamentally change the structure of computing. They streamline human-computer interaction and make the process both faster and easier. Using voice to bypass opening a mobile app and executing a transaction is something entirely different from what occurs today.

It won’t be long before we look at mobile apps as a clumsy way to organize information and contain user experiences. Apps are silos. You must open them and then navigate to access information or conduct actions within a narrow domain. Ji was demonstrating how Bixby breaks down the barriers between silos. It flattens access to information and transactional capabilities because the functions are callable from anywhere the voice assistant can be employed.

Can Samsung Succeed with Bixby?

If you want to answer the question of whether Bixby can be a successful voice assistant, the answer will depend on your definition of success. I do believe Bixby could be useful and adopted by many consumers for specific use cases. It may also become the preferred choice of a small number of global consumers. However, I am not convinced Samsung can be a global leader in voice assistants. It is too reliant on Android for mobile which puts it at odds with and at a disadvantage to Google Assistant. Samsung’s broad portfolio of appliances may be a more promising asset, but it is hard to see those access points becoming more than narrowly focused on specialized tasks. The SmartThings ecosystem also has value, but it is not clear how that will ultimately lead to sustainable differentiation given the incentives of device makers to support the Amazon, Apple and Google platforms.

With that said, it is still early in the market’s development. Bixby appears to be heading in the right direction and already has visual capabilities that most of its rivals lack. Samsung’s real challenges in the voice assistant space are more structural and temporal than technological. It is not clear where Samsung can build its base from to overtake the early market leaders outside of its home base in South Korea. The company has had an impressive run in mobile, but voice will be different and even more fiercely competitive.

Follow @bretkinsella Follow @voicebotai