Assistants voice : “We can not entrust our lives to machines “

Transformations. The wizards voice landed in France with Google Home, Echo (Amazon), and HomePod (Apple). In the United States, they would have already conquered 20 % of the market queries on the Internet. Orange and Deutsche Telekom have announced a joint product, Djingo, for this year.

Installed in a corner of the apartment, these assistants respond to your questions, allow you to launch the music, do your shopping, to inform you of the latest results of the football world Cup… Is to invent more uses essential, such as the fight against the loss of autonomy of elderly people.

These objects are the result of considerable technical progress. In the last century, research has started to resolve the question of the synthesis of speech from written texts. Then she is attacked in the speech recognition, a topic more complicated (it is easier to speak English than to understand a Scot or an Irishman…), because it is a matter of understanding the meaning of the text.

Markets ” acheulian ”

Progress in machine learning, coupled with the increase in computing power and the availability of massive data, have enabled important advances, which are applied in the search engines or automatic translation. But the problem is complex and the path is still long, which explains the current limitations of these assistants voice rather benêts, moreover copiously mocked on social networks.

Especially, the software that we have today require, for operation, massive volumes of conversations, and therefore of personal data, in particular to provide personalised services. In totalitarian countries, this represents considerable risk for the citizens. In Europe, where we are better protected with legislation such as the general regulation on data protection (RGPD), they are none the less intrusion unbearable…

Like this post? Please share to your friends:
Leave a Reply