Getty
Some readers might find this entry controversial and accuse me of being in the pay of the technological companies or some such nonsense (the day I find all that money they’ve supposedly been paying me, I’ll be a multi-millionaire), but there are some things I have to speak out about. For some time now, I’ve been reading about how domestic assistants like Amazon Echo or Google Home supposedly spy on their owners by recording our conversations, which are then sold on. Amazingly, many people seem to swallow this nonsense when they should know better.
The latest scare seems to have been sparked by a couple of news items: the first, in the United States, relates to a letter written in May by Chris Coons, a US Democratic senator, to Jeff Bezos, demanding information about recordings made by Amazon Echo, to which the company responded that they are stored and used by the company unless expressly deleted by the user. The second comes from a Belgian media outlet that had access to a leaked file with anonymized recordings from users, and was shocked to learn that Google has the same policy regarding Google Home recordings.
Firstly: these stories are not “discoveries” or “scandals” but instead are designed to grab the public’s attention. Anybody with a minimum knowledge of how these things work knows perfectly well that the recordings, intended (when the user turns the assistant on with a voice command) or failed (when the assistant mistakenly heard that command) are used not only to carry out the command in question, but also stored, unless expressly specified otherwise, so as to improve speech recognition technology. Can we prevent our requests being kept? Yes: both Amazon Echo and Google Home have a feature that allows you to deactivate recordings.
Secondly, nobody is forcing you to buy one of these devices, but if you decide to, you should understand that training the algorithms that drive them means that somebody has to listen to a recording, label it and categorize it, and then incorporates it an anonymous and extremely valuable database used to train speech recognition algorithms. Of course, no matter how anonymous the data base is, it’s still your voice and while you would probably be able to recognize it, the recording has no connection to the personal information in your account. Furthermore, if we all decide not to cooperate with the task of improving speech recognition algorithms, then the tech companies will take a lot longer to improve home assistants. But if it makes you feel better, then go ahead, nobody is stopping you. Is that what we all want?
That said, it makes no sense to be scandalized because a device designed to listen to what we say does precisely that. If you don’t want you home assistant to listen to what you say, then disconnect your microphones. But having a home assistant disconnected from its microphones kind of defeats the purpose of a voice-activated home assistant. So let’s be clear about these so-called news stories: home assistants are not there to spy on us, but to listen to what we ask them. Sure, the technology is still not failsafe and so they sometimes activate themselves mistakenly, but they aren’t spying on us. Similarly, how is Google or whoever supposed to train speech recognition algorithms without using as wide a sample of recordings as possible? Training algorithms requires understanding accents, idiomatic speech, or very varied ways of asking something.
Which is not to say that technology companies always get things right. If they allow these recordings to be leaked and, moreover, linked to the identity of the users — which has not been the case, at least for the moment — they should be punished, and the same applies if they ignore customers’ preferences and store their recordings. But being scandalized because our recordings are used for training is absurd. Nobody is spying on us. Our recordings are not being passed around among employees. Instead, they are listened to and labeled to try to understand them and to improve how the assistant interprets them. Oh, my God, Amazon and Google employees are listening to what I say at home! That’s right. That’s precisely how this technology works. There’s no other way to do it.
The sooner the media stop making up so-called revelations, the sooner we will begin to understand concepts as basic as data collection, labeling and algorithm training. Let’s get real here: nobody is interested in keeping a memento of your voice, nor passing it on to the media to embarrass you.
Over time, home assistants will develop sufficient capacity to accurately process and interpret most of what we ask them. But that will only happen when the media stop peddling conspiracy theories and help us to understand what companies do with the data we generate. Nobody is saying the tech giants shouldn’t be punished when they get things wrong, but in this case, there has been no wrongdoing: they have simply done the logical thing. From now on, can we have a little more thought and a little less scandalmongering?
