Virtual language assistants such as Alexa and Siri are finding their place in more and more households. Android devices come with “Ok Google” by default; iOS devices come with Siri; Amazon devices come with Alexa; and Microsoft devices come with Cortana. Amazon, in particular, seems to be asserting itself alongside Google and Apple with its intelligent loudspeakers in the smart home.
What’s The Problem With Voice Assists?
I have to admit that voice control has its appeal. After all, in everyday life, we also communicate with other people via the instrument of spoken language. This also explains that something human is attributed to virtual language assistants like Alexa. For example, attributes such as humor, wit, and cheek were trained for Siri. For jokes like Alexa, play Despacito” and “Alexa, are you connected to the CIA?” It’s not bad.
I also see a severe area of application for people with physical disabilities. Beyond that, however, I only see convenience in the fields of application (e.g., playing podcasts or music, switching lights and devices on and off with programmed routines) at the expense of data protection. Above all, I am not concerned with preventing technical progress but with making it more data protection-friendly.
Privacy-Friendly Voice Assistants
So I was fascinated by the Snips.ai project. The company offered a platform for language assistants that could be operated locally and in a data-protection-friendly manner, for example, with a Raspberry Pi. Such a future for language assistants would be a dream. However, Sonos took over the company and dropped the local voice control option. Florian Quirin is currently working on an alternative with SEPIA and openHAB. The Mozilla Foundation works on a free language data set with the Common Voice project.
Speech Assistants: Speech In The Cloud
The providers, Amazon, Apple, and Google, just to name a few, rely on cloud-based solutions. The voice excerpts, which are spoken after the keywords “Alexa”, “Hey Siri”, and “Ok Google”, end up on servers in the USA. You will be analyzed there, and the program behind the language assistant will give you a suitable answer.
It is no surprise that the providers save the voice recording history. Instructions for Amazon Alexa show how to delete them. Much more information can be gained from spoken language than the written word. Are you happy or sad, healthy or sick? Data can be extracted from the spoken word and for whose (advertising-relevant) analysis Amazon has secured a patent. It is fitting that acquisitions such as PillPack and Health Navigator are gaining a foothold in the healthcare sector.
Visit The Alexa User Community
I do not want to violate anyone’s right to use such devices. At the same time, I feel uncomfortable when I visit a household with an Alexa device, for example. A friend invited me to his place. After a long time, we had a lively conversation until he told me that an Echo Dot was hidden in the corner. In response, I asked that we continue the conversation in another room. The Alexa device should only record the communication if the keyword “Alexa” is mentioned. That is not guaranteed, either.
A computer science student shared his clever solution on a train ride with me. He put a Raspberry Pi kill switch in front of the Alexa device. Placing cell phones aside would also be logical since language assistants such as “Ok Google” and “Siri” are also at work here. It has become common among friends to put their smartphones aside. After all, the person you are talking to is interesting, not the mobile phone. It’s generally an uneasy feeling knowing that conversations won’t stay private. This applies to cloud-based voice systems and to acquaintances who are too talkative.