DIGITAL ATTITUDE: A new update of the Echo connected speaker from Amazon, allows it to “deduce” what will be the next question you are going to ask it.
For example, by asking her voice software Alexa to find out how long it will take to brew a tea, she will respond: “five minutes”, then initiate a continuation of the conversation by asking: “Do you want me to set a timer for five? minutes?”
The virtual assistant is equipped with tens of thousands of features activated by voice, skills or skills, developed by Amazon or third parties, which enrich their knowledge. Following the same principle as the applications that we download to our smartphones, allow us to access a great wealth of content.
Calling on Alexa to order a pizza, play music, consult your calendar or listen to the news, are among the skills current. But none of them, so far, have followed up on a search in attempting to preempt our next question.
According to Alexa engineers on theeur blog: “This improvement to the software is the continuation of our efforts to ensure that exchanges with the virtual assistant are as fluid as possible”. Chatting with Alexa should therefore be as natural as talking to a human being, and allow technology to anticipate the logical sequence of a request, giving rise to a real dialogue.
The company calls this process “inferring latent customer goals”. A system based on deep learning which analyzes previous requests made by a very large number of users. For now, this new capability is only available in the US and English.
Talking longer with Alexa could be fun, it is already the case, because she answers kindly and sometimes even with humor to questions, but it could also be annoying, if you only want a simple answer. .
Amazon’s ambition is to integrate Alexa into almost all the electrical devices of everyday life – televisions, fridges, toasters… in order to control them by voice. As long as they don’t all start questioning us.