Voice Assistants : Understanding technology before being afraid of it
Tribune by William Simonin, co-founder and CEO of Vivoka.
It took more than 30 years for voice assistants (since IBM in 1961) to get to know their first commercial applications. With its integration into smartphones and connected speakers, voice recognition has in recent years experienced an era of exponential growth (195 billion dollars in 2021).
From message dictation to an Internet search, via an e-commerce purchase or the launch of music, to continue to offer optimal interactivity, players must constantly extend their application ecosystem and offer new experiences.
Essential data to better serve the user.
Today, there is a significant global lack of clarity about the actual understanding of the voice assistant, how it works and how users’ personal data are used. And when you say unknown, you mean fear. These fears have been reinforced by multiple bugs in these tools: in 2017, a 6-year-old girl ordering a doll’s house and a box of cookies – without her parents’ knowledge – or in 2018, a couple discovering that their conversation was recorded and sent to a person in their phone book.
The user asks himself the question of the use of his data: are they stored, analysed, how and for what purposes ?
To provide a first element of answer, it should be known that speech recognition can use either the cloud or the embedded technologies in its operation. The first allows the AI to access a wealth of online information that can serve as a resource and feed many use cases. On the other hand, Internet dependency is inevitable. While the second is totally autonomous with regard to the Internet connection, but will see its field of action partially reduced. This dichotomy echoes certain directives of the GDPR (General Data Protection Regulations of May 2018). Indeed, in the latter is mentioned the “Private by design”: a conception of technologies designed to respect the user’s data. Logically, it is easy to think that only a tool without an Internet connection can be, because no information can pass through it without it. However, depending on the protocols in place and the ethics of the companies, secure cloud services can guarantee a total protection of the flow of personal data.
Indeed, data in the era of Big Data is sensitive for the user, but essential for IT companies. The technological tool behind the voice assistant needs to improve, to learn, to correct its errors in order to be able to assist the user in his daily life. This improvement, necessary to ensure the best user experience, is nevertheless very dependent on the “inputs”, i.e. the data that the system will be administered. This is all the more true when the objective is to customize the voice assistance with the user’s habits and preferences! It is for these last reasons that today’s voice solutions are based in particular on individual data. However, this does not justify that some actors keep the data and use them for purposes other than improving the experience.
The need to educate consumers about their rights.
Even if our words are vanishing, most voice requests are recorded in a cloud just like written requests made in a search engine.
Since the creation of the GDPR, it is important to note that users now have laws behind which to protect themselves when using their data. It is now essential to educate these users to choose whether or not they want their data to be used and to give them that choice.
In my opinion, it is the duty of the voice recognition market to educate and raise awareness on these aspects. Indeed, no one nowadays takes the time to read the T&Cs/TU of any product, so it is a moral duty to make this information accessible, without drowning it in a surplus of unnecessary information.
In this way, users will not only be able to know the actual use of their data, but also their different rights. For example, many people are unaware that on a simple request, we can ask for the total deletion of our conversations’ backups !
More generally, an education would allow everyone to understand how the voice assistant works from the first word spoken to the final use of the data and once with enough knowledge, everyone will have the choice of whether or not to use this or that assistant and above all be able to understand the problems and advantages associated with it.