Excitement, entertainment, and nonsense are the best words to describe our first feelings about the first contact we made with Alexa, Amazon‘s voice-controlled AI assistant that is integrated into Amazon Echo and Echo Dot.
After testing the most operative part of the device, it is the turn for evaluating Alexa’s communication and self-learning skills. How smart is Alexa? Does she learn from mistakes? Is the communication process natural and seamless?
Although 4 out of 10 Germans would use smart speakers, especially to control and monitor devices and appliances at home, with regard to virtual assistants the boundary line between fascination, indifference and usefulness still seems to be very fine. To what extent do Alexa’s skills reinforce the uptake of voice-controlled artificial intelligence as the main interface?
How smart is Alexa?
During this test, we asked ourselves several times to what extend Alexa is smart. We could not resist the temptation to ask her tricky questions such as: “Alexa, how old are you?” – I am available since 6.11.14, or “Alexa, are you silly?” – I am constantly learning and will be better next time.
Despite this willingness to learn we could observe that she makes the same mistakes over and over again. She doesn’t really have self-learning skills; instead, we learned how to formulate the question based on the previous experience, what is really disappointing.
In this sense, although she showed up her strengths and character, she gave us a glimpse of her limitations. These limitations should be refined with further upgrades because the gap between people and Alexa is still far from being narrowed.
Moreover, you can enhance Alexa’s capabilities by adding new skill apps, of which some require account linking by the way (annoying). The skills section is like an app store divided into different categories (such as newest arrivals, eating and drinking, lifestyle, local, music and audio, education, shopping, games, humour, transport or smart home, among others). In every skill you can find customer reviews and some guiding questions to take the most out of every skill. Developers are welcomed to create new skills, what has significantly increased the number of skills in the last months.
How effective is the communication?
Despite the fact that there is a guiding card inside the box with questions to try to ask Alexa and there is also a help section within the app menu, we realised that in the end we have to find our own way to understand each other by training and testing our wording, tone, pronunciation, and speed.
If you use very colloquial language, your pronunciation is not so good or your speech sounds plain, chances are that the answer you get back from Alexa is: “Sorry, I don’t have an answer for that question”. Consequently, we had to adapt our way of speaking in many cases, resulting in speaking unnaturally and inconvenient situations.
Although the speech recognition system is very promising, there is still a lot of room for improvement. For example: Alexa is confused with verbs that can have double meaning such as to play (sports or music). For instance: You can ask her about the result of a football match and she responds by saying that she cannot find anything in your music library.
To this respect, the conclusion is that the more we practise, the better understanding and the more effective our communication. (For us, not for her).
On the other hand, I would like to stress (another) negative aspect: Alexa is useful and appealing when she delivers short and specific messages. When the given information is too detailed, rich, and dense, the communication is pointless and endless, unless you have your smartphone within reach to be able to follow the whole history conversation.
The conclusion here is that to make effective use of the voice conversation, continuous support of a screen or display is more than necessary.
This first model of Amazon Echo has to be seen as the very beginning of a revolution that is yet to come. Thus, although voice-controlled interfaces can be very useful in situations where you have to pay attention to another important thing like driving, it seems like future voice-controlled virtual assistants will be enhanced by wider integrations, phone calls and even touch screens, which can facilitate and shorten the decision making and communication process and, consequently, convince more users of their usefulness.
Until all of this comes, I believe that Alexa came to market too early for users. Precisely, one of the first challenges manufacturers should face to increase consumer adoption is to turn robotised and cumbersome interactions into compelling experiences.
In line with this, more complex connections with smart hardware can already be established and the screen will have its own space within virtual assistants since it seems to be the fastest and most effective way for a machine to give us information. Baidu has already made the first move with Little Fish. Will Amazon be next?