Psssst ... The Smart Home is listening

  Consumer electronics and home automation that respond to your voice—until recently, that sounded like science fiction, but it’s already a reality and is rapidly becoming a booming market. And in fact, the idea is quite cool. Instead of juggling a bunch of remote controls or searching through half a dozen apps to find the right one, you simply control your entire home technology with ordinary voice commands. Whether it’s dimming the lights, adjusting the heating, or just playing a particular song on the system—as long as the voice recognition understands the command and there’s a connection to the relevant technology, it’s all no problem. You can also easily retrieve and have information from the internet read aloud, such as the weather report for tomorrow or the traffic situation on your way home.  

The race is on

Amazon's Echo is currently the leader among smart home products—and is selling extremely well. Since last year, Amazon has been marketing its voice-controlled technology, called "Alexa," in the smart speaker Echo and its smaller version, Echo Dot. And apparently so successfully that in Germany, you first have to join a waiting list to eventually receive an invitation (!) to order one. The PC giant Lenovo (formerly known as IBM) just announced its own products at CES, which will also use Alexa. And there are very clear reasons why Amazon, in particular, is pushing into this market so aggressively. No other provider can link search, service, and sales as closely as the powerful online retailer, which is now on track to become Google's biggest competitor.  

Lenovo also trusts in Amazon's Alexa technology and has showcased its own smart products at CES   Google, of course, is not standing idly by and has recently introduced "Google Home," which takes a very similar approach. Apple seemed to have the edge for a while with Home Kit, but so far the system is little more than a nice idea. Currently, the Korean electronics giant LG and the German-Swiss company digitalSTROM are thinking at least one step ahead. Both have showcased voice-controlled robots in Las Vegas that are intended to support us as digital butlers in everyday life.  

Google simply calls its Alexa competitor "Home"  

High-tech at its finest

For many, it’s a strange idea: Thanks to modern voice recognition, smart robots like digitalSTROM’s "Pepper" will soon become everyday companions (Image: obs/digitalSTROM)   The technology behind the various offerings is indeed fascinating. Understanding natural language remains a gigantic challenge for any computer and has only become possible in its current form thanks to recent developments in artificial intelligence (AI). An AI system differs from a traditional computer mainly in that it is self-learning. It no longer needs human programmers to implement new voice commands. Instead, the system becomes a little bit smarter with each request and, unlike the unintentionally comical early days of voice recognition, can increasingly understand dialects, accents, and colloquial phrases.  

Voice recognition certainly has its quirks... not just for Scots!   This self-learning process requires considerable computing power and especially gigantic databases, which are growing at a rapid pace. Neither the storage capacity nor the necessary computing power is available on an affordable home device, so the voice inputs are sent over the internet to large servers, processed and interpreted there, and then sent back to the device at home. And that’s exactly where many see a risk.  

Attentive listeners

To react instantly to voice commands, these devices are constantly listening to their surroundings—Amazon’s Echo, for example, uses no fewer than seven microphones. When they recognize a preset keyword ("Alexa," "Okay, Google," "Hey, Siri" ...), they record the following voice command and send it to the server on the internet for analysis. So far, so good and helpful, but no one really knows what happens with the other audio recordings from the surroundings. Amazon, for example, states that the device internally stores the last 60 seconds of recording, which are then overwritten and thus permanently deleted. This relatively long recording period is said to be necessary to correctly interpret actual voice commands in context. Other manufacturers remain silent on such details but probably use similar methods and data.  

With no fewer than 7 microphones, Amazon's Echo is constantly listening for its keyword.   If it stayed at that, there would actually be little to criticize about Echo & co. It may seem a little strange to be constantly listened to by an electronic device. But if the data remains only on the device and is deleted again after 60 seconds, this intrusion into privacy is hardly worth mentioning. Every Facebook user, every smartphone owner voluntarily reveals much more about themselves and even explicitly allows providers to use this personal data for any kind of commercial purpose. But a slight discomfort remains. First of all, there’s a certain mistrust of the technology corporations that gnaws at the good feeling. Voice recognition, everyone agrees, will be one of the key technologies of the coming years and will open up a gigantic market. And the better a company’s own voice recognition works, the bigger the slice of the expected revenue pie the provider can claim. And since the quality of recognition depends largely on the learning abilities of the AI servers, it seems quite tempting to secretly use the constantly listening devices to build up one’s own voice database and thus gain an advantage over the competition. Data protection officers, hackers, and security experts around the world will be eagerly waiting to catch a manufacturer misusing audio data in this way. But as we can currently see from the Volkswagen example, in tough competition, moral and legal concerns sometimes get thrown overboard if the competitive pressure is high enough. So it’s at least not impossible that major providers like Apple, Amazon, or Google might one day succumb to this temptation. The terms and conditions you have to accept before using such a device, in any case, allow providers—as usual—an enormously far-reaching use of the data collected.  

  There is also at least theoretically the danger that an outsider could gain access to this data or even directly access the built-in microphones. To function, devices with voice technology logically require a two-way internet connection—they must be able to both send and receive data. Even if manufacturers credibly assure that they have taken every possible safety precaution, such protection can never be one hundred percent secure. So it’s not entirely impossible that the smart home speaker at home could be turned into a bug by a hacker. And surely intelligence agencies worldwide are already looking for ways to use these new technologies for large-scale eavesdropping. Are these all reasons for excessive paranoia? After all, we are nowadays constantly surrounded by devices like smartphones, tablets, or laptops that are much easier to hack, and when using them we voluntarily leave behind a vast data trail. But with voice recognition, and the always-on listening required for it by smart devices, this takes on an entirely new dimension. Imagine discussing dinner plans with your partner and suddenly being interrupted by your smart home system suggesting a new pizza delivery service nearby. Today, that’s a creepy idea, but in ten years it might seem completely normal. Because one thing has already been proven by Apple, Google, Amazon, Facebook, WhatsApp, Twitter, Instagram, and many others: Despite all initial concerns, people are ultimately happy to give away their personal data as long as the benefit for them is right. Smart home technology with real, intelligent voice recognition is coming, that’s not in question. But maybe now is exactly the right time to think about what freedoms you want to grant the developers of such technologies—and which ones you’d rather not.