Every since Captain Kirk uttered “Computer” and the USS Enterprise’s onboard AI woke ready to do his bidding, tech giants have been striving to develop practical voice-activated assistants to replace the keyboard. In the last couple of years the technology has come on leaps and bounds and interactions with other internet-enabled devices, enabling us to order groceries, dim the lights or find out the latest football scores with a couple of spoken words have emerged.
It’s no surprise then that the world’s biggest personal technology providers are vying for our voice commands to steer business their way. But people have raised concerns that devices in the house that are ‘always listening’ could be spying on them. News stories such as Samsung warning customers about discussing personal information in front of its smart television and Arkansas police demanding that Amazon release recordings from an Echo device that was present at the scene of a murder have helped stir misconceptions about how much our devices are listening in.
Are security concerns justified?
According to Alec Muffett, a freelance blogger, speaker, software engineer and computer/network security consultant who serves on the board of directors for the Open Rights Group, such fears are unfounded and are down to a basic misunderstanding of how the technology works.
If one treats voice command as a glorified keyboard for putting search terms into Google or Amazon or anything like that, what does it matter that it’s voice as opposed to all this other information which they’re already collecting about you?
“If one treats voice command as a glorified keyboard for putting search terms into Google or Amazon or anything like that, what does it matter that it’s voice as opposed to all this other information which they’re already collecting about you?” he asks.
If you use Google, and especially if you have an Android phone, you can get an insight into how much data is gathered on your activity via the Google Dashboard, for example. Similarly, consumers with Google Assistant enabled can access can review their voice command activity through the Google My Activity dashboard.
“You can go through your history, and there’s a transcription of what you asked, like ‘What’s the weather like in London today?’ and there’s a playback button next to it where you can hear your voice command. Your device records what you’re saying and uploads it to Google, because that’s part of their engineering and debugging process.”
However, the listening and recording doesn’t start until you say the ‘wake word’ relevant to that platform – ‘Siri’ or ‘OK Google’ for example. The way those commands are identified doesn’t require the constant listening some people fear, as that would be hideously inefficient. Instead it uses a similar pattern-matching trick to the Shazam tune identification app.
“Shazam doesn’t upload an audio clip because that would be really noisy. It analyses the frequency pattern of the sound – there’s some high frequencies here, low frequencies there, a pulsing backbeat– are there any songs with that fingerprint? Shazam can look up a fingerprint faster than it can match segments of audio,” explains Muffett.
So if people are worried that Apple, Amazon or Google are listening to them, it’s only because that is what they’re buying into when they trigger listening by saying the keyword, which is identified through Shazam-type fingerprint matching.
“If people are upset or concerned, is it in an informed way?” Asks Muffett. “Otherwise what they’re doing is essentially marching up and down and demanding the new looms at the mill are taken down because it will destroy work in the future; it’s that level of Luddism,” he warns.
Don’t wait for the law
While the consumer has a responsibility to understand how his or her data is collected and used, Scot Ganow, a US attorney at the law firm of Faruki, Ireland, Cox, Rhinehart & Dusing PLL in Dayton, Ohio, advises corporate clients on privacy and security law practice. He recently delivered a compelling TEDx Dayton talk on Humanity in Privacy.
“There’s a paradox that we have with privacy,” says Ganow. “We want the convenience, we want the technology, but we also want privacy, and Americans are negotiating this transaction every day. I think the biggest issue is, are they doing it knowingly, are they aware of everything their private data influences?”
Concerning stories such as Samsung’s snooping TV or the Amazon Echo potentially recording evidence of a crime, Ganow says, ”Any time a story like this breaks that tweaks people’s spooky button as to whether companies or the government should be doing this, you always hear the question ‘Surely there’s a law against this?’ Often there aren’t laws for a specific area, and I’d encourage authorities to be slow in making laws about new technology, because laws that are made quickly tend to be not very good law.
“The biggest impact you can have on your privacy is through the choices you make on a day-to-day basis. The law will be too slow, the technology will only be as good as you, and in the market place, let’s be clear, they want your data, and they want more of it.”
Ganow encourages his business clients and companies to build privacy into technology using the approach promoted by the Canadian movement Privacy By Design. This suggests that technology that uses personal data must be built with privacy at the forefront, and ultimately give the user clear choices, making it easier for them to say yes or no.
“Generally speaking the companies that make digital assistants, like Amazon, Google and Apple, build in privacy protection,” he says. “Siri doesn’t record and act on your commands unless you give it the keyword to do it. Part of Apple’s culture is a respect for privacy. We saw that in the US when Apple refused FBI requests to create software that would unlock an iPhone recovered from one of the shooters in the 2015 San Bernadino terrorist attack.”
While companies are doing their part to secure customer’s data, Ganow’s message to people using these devices is to educate themselves on the privacy and security functions of the product before turning it on and connecting it to the network, and exercise the options provided.
“As with any technology, it tends to blend into the background and people seem to forget that it’s on. There’s a very simple solution; unplug it when you no longer want to use it, when you go to bed at night, or if you have concerns. Make conscious choices as to when you’re going to use it and when you’re not.”
And, as the device itself may not be the weakest point in your data security, Ganow adds a final piece of advice. “As with all devices, make sure you’re implementing a secure wireless network within your house.”
Proactive personal assistants
Once we’re satisfied that our data is secure and being captured on our terms, can we be sure that the choices digital assistants make on our behalf are right for us? Ariel Ezrachi is the Slaughter and May Professor of Competition Law and a Fellow of Pembroke College, Oxford. Along with Maurice E Stucke, Professor of Law at the University of Tennessee and co-founder of The Konkurrenz Group, he wrote Virtual Competition, a book which examines whether the sophisticated algorithms and data-crunching that make browsing so convenient are also changing the nature of market competition.
He warns that as virtual assistants are being increasingly adopted by customers, this could generate risk as we trade convenience for competition and giving away more and different personal data.
“Personal digital assistants are alluring,” Ezrachi says. “They can read to our children, order beer and pizza, update us on traffic and news, and stump us with Stars Wars trivia. So we likely will trust them.
Our chosen personal helper will have unparalleled access to our information. Our assistant will become pro-active. Knowing what shows we watch, the stories we read, and the music and food we like, they will anticipate our needs
“Our chosen personal helper will have unparalleled access to our information. Our assistant will become pro-active. Knowing what shows we watch, the stories we read, and the music and food we like, they will anticipate our needs. Using our personal data, including our calendar, texts, e-mails, and geolocation data, our personal assistant may recognise a busier than usual day, and suggest a particular Chinese restaurant. Powered by AI, the helper will become an integral part of our life.
“In doing so, its gate-keeper power increases in controlling the information we receive. One concern is economic, namely its ability to engage in behaviour discrimination and foreclose rival products. But the larger concern is social and political, namely its ability to affect the marketplace of ideas, elections and our democracy.”
The nature of the voice interface itself may also mean we’re missing out.
“The moment you run a traditional query, if you’re unhappy with the results, you have the screen in front of you and it’s easy to navigate through other options,” Ezrachi says. “With voice activation people will rely much more on the first reply we get from the digital helper; it lends itself to a single recommendation or a very short list.”
Will our digital assistants be with us from cradle to grave?
Like it or not, digital assistants are here to stay, and for the next generation they could become as indispensible and ubiquitous as mobile phones are today.
“Mattel is now selling a baby digital virtual assistant called Aristotle,” says Ezrachi. “It can help purchase diapers, read bedtime stories, soothe infants back to sleep, and teach toddlers foreign words.
“For babies born in 2017, a digital assistant may become their lifelong companion, who will know more about each person than parents, siblings, or individuals themselves.”
For that to be an exciting rather than terrifying prospect requires consumers to educate themselves on the privacy and security functions of their device and how their data is captured and used today, so it can serve them better tomorrow.