At least once a week, one of my kids asks me how my day was and what I did at work or, more generally, what I do at work. My answer to them is generally along the lines of “I help my colleagues determine what personal information we should be collecting for apps, services and websites; how we should and shouldn’t be using that information; how we help people using those things understand what we’re doing and the choices available to them.” As someone who has been working in technology for more than 20 years and privacy for roughly 12, I was surprised when Alexa emerged as a concrete way for me to explain my work to my kids.
Last week, Rani Molla at Vox published an insightful article addressing the inconsistency of people claiming to care about privacy while continuing to purchase and use privacy-impacting devices with a particular focus on smart-home, internet-of-things devices. The article canvases the various reasons for this disconnect between individuals’ values and actions.
There’s growing criticism and fear of devices and smart assistants intruding into our homes, learning our voices, our preferences and our habits. We are concerned about devices listening in on and recording intimate moments within our otherwise private homes.
Although I am generally an early adopter of new technology, beyond asking Siri to play a particular song in the car, I haven’t been a heavy user of smart assistants. I’ve been a long-time Amazon customer and Prime member but never felt much pull toward an Echo (or Google Assistant–powered device or Homepod, for that matter). I am not generally looking to speak to my devices and, when I’ve tried, I’ve been less than impressed with their accuracy.
A few months ago, I won an Amazon Echo in a draw at a privacy conference (the irony). I figured it would sit on our kitchen counter and serve as an easy way for my family to enjoy music as we prepare dinner. And that’s mostly been the case, aside from a few of what David Letterman would call "Stupid Pet Tricks"— trivia quizzes, fart or vomit sounds for my 10-year-old, dog barking noises to amuse our puppy, and customized actions for Alexa to respond to phrases such as “Alexa, who do you love?” (her reply is, “you know there is only one,” at which point she plays Kesha’s song “Stephen,” much to the amusement of my wife and kids).
With the recent uproar about Amazon’s human review of certain Alexa recordings, I became aware of the Alexa privacy dashboard, available through the Amazon account page. I clicked over to take a look and saw months of mine and my family’s transcribed commands to Alexa. The more startling moment was clicking on one of the commands to hear my voice from two months ago asking Alexa if dogs can eat carrots (they can).
In privacy terms, was that "creepy?" A bit, but I’m a privacy and technology geek, so it wasn’t wildly surprising. I wanted to see how my family would react.
I gathered my family around our kitchen table and reminded them about how they ask me questions about my work. I told them that Amazon has people like me because they collect quite a bit of personal information, including information about how we use Alexa.
“You see,” I said, gesturing to the screen, “here are all of the requests we’ve made to Alexa.” They humored me by glancing at the screen and nodding. I clicked one of the links, and my 12-year-old’s eyebrows furrowed as she heard her voice asking Alexa to play AC/DC. Then my 8-year-old: “Can dogs eat banana skin?” Then my 10-year-old: “Start a 5-year timer” (she can’t).
They had not said anything inappropriate to Alexa, but they were all surprised and a bit unnerved about the trove of recordings Alexa was sitting on. I showed them how I could delete the recordings and reminded them that, as the account holder, I was able to review the recordings, and they should keep that in mind when they speak to Alexa. It’s just a matter of time before a parent stumbles upon a recording of their teenager asking “Alexa, how do I know if I’m pregnant?”
What transpired next was a high-level privacy review with my wife and kids as participants.
I explained that their reaction was what privacy professionals generally work to avoid — we don’t want surprises. Rather, we want people to understand at the outset what information is being collected, why, and the choices and controls available to them.
I pointed out that it was great to have a dashboard and how that helps with both transparency and control, but we also discussed how we could have avoided privacy surprises in the first place. Here were some of the suggestions we came up with together:
- Asking us at the outset if we want our voice recordings retained and, if so, for how long (or perhaps set a default retention period).
- Describing to people why voice recordings would be retained (e.g., does it help Alexa learn our specific requests? Does it help Alexa get better at understanding everyone?).
- Explaining the controls available to us if we change our mind or want to review our recordings.
The suggestions above aren’t intended as a comprehensive privacy critique of Alexa. On balance, we prefer the conveniences we get from Alexa more than we’re concerned over its intrusion into our household. And I wouldn’t be surprised to see some of our suggestions above work their way into the product as it evolves with the help of Amazon’s skilled privacy professionals.
As conversational and helpful as our smart assistants aim to become, what I appreciate most so far is the conversation that took place around my kitchen table. My kids finally understand what I do. Alexa, thank you.
Photo by Andres Urena on Unsplash