If you need to book a flight, answer emails, and pay bills, how would you do it? You could download three separate apps and learn to use them individually, or you could simply command your virtual assistant. Consumers want an on-demand app - one that has a conversational interface that is ready to carry out complex actions. That’s exactly where conversational interfaces are heading – by providing ease of use and more “human” interactivity.
Currently on the market vying to become consumers’ next favourite personal assistant include Facebook’s M, Apple’s Siri, Google Now, Amazon’s Alexa, Baidu’s DuEr, and Microsoft’s Cortana. However, these personal assistants are far from perfect. They can only complete simple queries, and often times are poor at transcribing speech.
What’s the Current State of Virtual Assistants?
Let’s take a closer look at what personal assistants can do at the moment. We’ll examine the six assistants mentioned earlier, and determine their utility.
Facebook M: This virtual assistant is text-based, and because it does not speak to you it is less personable than a chatbot. Facebook M has been advertised as an easy way to set reminders, find a good restaurant, and can even process refunds for you. Though often times it will rely on available online ratings and does not have an opinion of its own. Facebook M has been noted to get things done similar to humans, but lacks the ability to showcase human-like qualities. Currently Facebook has opened M’s AI platform as a bot engine for developers.
Apple’s Siri: Probably the most well-known voice driven assistant is Apple’s Siri. Siri can search the web, make calls, set alarms, change phone settings, play music, and the like. What makes Siri entertaining are the answers to carefully worded questions. However, most times when it isn’t sure what you are referring to it will bring up search results from the internet. Siri also cannot interact with third party apps, or pull information from your email. While Siri’s speech and “personality” is most human-like of all the assistants, it still lacks the ability to fulfill complicated tasks.
Google Now: This personal assistant is one you don’t have to talk to. What we mean by this is that Google Now will automatically pass relevant information to you through information cards. It anticipates the kind of information you will need based on data it has collected as you use your device. You can also choose to directly input your preferences as well. So you might get information related to the weather, or what transportation route to take home. There is a mic option that you can use if you choose to. So while Google Now aims to be helpful, like other assistants it lacks human-like qualities and the ability to carry out difficult tasks.
Microsoft’s Cortana: Cortana is a combination of a voice assistant and text analyzer – you can talk and type to it. Cortana can make calls, provide travel advice and weather reports, and search the web (using Bing). The biggest problem with Cortana however is the lack of consistency it provides, and answers aren’t as concise as ones provided by Google Now or Siri. Cortana is a hit or miss. So, we can see there’s definitely room for improvement in Cortana.
Click here for more details on Apple’s Siri, Google Now, and Microsoft’s Cortana.
Amazon Alexa: Alexa is a voice controlled system which acts as a smart home control allowing you to dim your lights, play music, and manage your smart home products. So we know this assistant is great at making daily tasks easier. You can also go voice shopping if you’re a prime member buying eligible prime items. However Alexa can only complete one command at a time, and background noise pose a problem while trying to interact with it.
Baidu’s DuEr: This voice activated assistant was created by the Chinese search engine Baidu. Baidu had developed Deep Speech 2 which proved to transcribe speech better than humans, but it could only complete simple queries. DuEr was then designed to help users complete more complicated tasks such as giving recommendations, make purchases online, place restaurant reservations, buy movie tickets, book flights, etc. However like the rest of the assistants, it seems that there’s a lack of human-like characteristics and ability to engage in meaningful conversations.
The overall consensus is that these personal assistants are capable of completing simple commands. It’s a very one-sided interaction where you instruct your personal assistant and it may or may not do what you asked. No rapport is established as assistants will forget what they’ve told you. But given the current context, which personal assistant looks most promising to you for future developments?
Key players in voice recognition and language processing such as Baidu, Google, Apple, Nuance, and Facebook are looking to progressively improve the way we access apps – making them “smarter”, more useful, and more human. Changing operational interfaces to conversational. One of the bigger challenges is teaching AI systems to understand and respond intelligently to more complicated phrases and demands. Baidu’s aim for DuEr is to eventually learn the ability to carry meaningful conversations while showcasing adaptability by incorporating changing information into the discussion. Andrew Ng, Baidu’s chief scientist says, “In the future, I would love for us to be able to talk to all of our devices and have them understand us”. This is the difference between getting a report and having a conversation – the information is the same but a conversation is more natural.
Some core functions we can expect to see in conversational systems in the future include improved speech recognition and the ability to carry out longer complex interactions, track information you’ve passed to them, and having more human-like engagement. As one of Gartner’s top strategic technology trends for 2017, we’ll continue to see further improvements soon.
At iTechtions, we’re excited to see the development of AI systems and how they can make our jobs easier in order to serve our clients better. What can improved conversational interfaces do for you?