The good side of humans listening to your Alexa records

In the past few months, a host of details about how companies handle voice recordings has caused a lot of ink to flow. First, Amazon hired thousands of contractors to review Alexa orders. Apple and Google were doing the same thing with their voice assistants. Then, more recently, we learned that Facebook and Microsoft had humans that transcribed the messages and phone calls of their users. All this has been a nightmare for those who care about their privacy.

None of these companies have been open about the fact that they have hired people to listen to and transcribe what users have said on their computer. Unless you have mastered the basics of machine learning and natural language processing, you probably did not realize that human revision is still an essential part of voice software development. It is. Amazon, Apple, Facebook, Google and Microsoft all justified their hope of listening to the voice recording of their users by explaining that it was essential to improve the artificially intelligent software that makes their products work. It is very frustrating that these companies have overshadowed the fact that humans are listening. We can only wonder what is happening behind the scenes with these gadgets and data storage services.

While the situation as a whole seems disappointing, consider the good side: if humans are needed for these invasive machines to continue to work, we at least know that the robot lords have not yet taken the relay. We even have a better understanding of what we can do to ensure our safety, thanks to the human subcontractors who have whistled and revealed their existence to the world. And there is still time to master the technology with politics and pressure before the situation escalates.

This summer's torrent of revelations on voice assistants and voice user interfaces sheds light on how this technology works. When Amazon released Echo in 2014, it was unclear exactly what the device's range of still-connected microphones actually consisted of. People feared that Echos would record everything, and privacy advocates have long argued that these so-called smart speakers are just futuristic forms of wiretaps. So far, they are not exactly wrong, which is why it is so important to learn more about the workings of technology.

We get a clearer picture of how all these voice-activated products work, usually because some scandals force companies to reveal details. As at the time, an echo misinterpreted some words, recorded the entire conversation of a couple and sent the recording to one of their friends. At the very least, Amazon had to explain how wake-up words worked and alert the public that its technology was not always working as expected. Google had a similar confrontation with the reality of faulty software, when a Google Home Mini accidentally registered the life of a technology blogger. These examples do not even include the hackers and insane attacks that security researchers have developed.

Now, thanks to this new series of scandals, we know that human beings are involved in the backstage of voice assistants and voice-activated software. We also know a little more about its imperfection. In some cases, it seems that humans are working much more in this area than technology companies and their marketing teams would have you believe. When Google introduced Duplex, a new service that uses human voice-like robots to make restaurant reservations, users soon realized that real humans sometimes made these calls, with the software slowly improving. Google wanted to make the world believe that it had invented a robot that could talk on the phone and look human, but that dream had not come true – at least not quite.

Knowing that human beings are retranscribing and examining their speeches reveals an equally failed aspiration. It would have been amazing, five years ago, if Amazon had invented the software of a speaker capable of only listening to you when you wanted it and interpreting your stock orders. That's what Echo and other smart speakers have promised to do from the beginning. Anyone who has one of these things knows that he is flawed and often has trouble understanding human language. If you review your voice commands (which is easier to do with some scandals), you will see that voice assistants accidentally record things all the time. Humans have tried to make things more efficient behind the scenes, and companies like Amazon have probably hidden this fact from the public because it would have revealed how flawless the machines are.

It still fears for privacy. It's scary that a human has heard everything you said to a voice assistant. For the moment, some of the companies in question have suspended the process of human revision, presumably until the bad press of all these recent reports dissipates. The human process will probably restart because it needs to improve the voice software. At least, more people will realize that they do not necessarily speak to a computer when they talk to a voice assistant. If the invasion of privacy sounds too scary, do not use voice assistant.

But what will be more frightening is when machines can train themselves. Isaac Asimov must turn in his grave at this thought. For the moment, if you think from the right point of view, it's encouraging to see that humans are still part of the equation. It is interesting that some of these human beings have the audacity to alert the press about their existence and to let the world know how the technology companies that use them hide the truth from users. Computers would never do such a thing.