“Computer, lights:” How Ars staffers actually use voice-interactive bots

Article intro image — Enlarge
Valentina Palladino

It's Labor Day—the day when, here in America, we get a little bit of a break from our daily work responsibilities. Most people are off work, students are out of school, and with any luck, we're kicking back with loved ones and enjoying beverages together.

That made us wonder: how do Ars staffers reduce their labor, day to day? Or put another way, do we trust those newfound voice-activated AI bots on our smartphones and smartspeakers?

I only use Siri on my iPhone occasionally ("timer, five minutes!"), and I like using it to read and respond to text messages while I'm on my bike, scurrying around town ("read last text message!"). But beyond that, my experience is that voicebots don't yet add much to the menial tasks of day-to-day computing. In the environments where I'd want a voicebot most—riding my bike on busy city streets—Siri often doesn't hear me correctly, so my texts get garbled. I can't tell you how many times I get the dreaded: "You'll have to unlock your iPhone first."

And no, I don't own a smartspeaker of any kind. But I'm still waiting for the day where I can speak, naturally, to a computer and have it just work flawlessly.

Echo, echo

We use an Amazon Echo in our living room and an Echo Dot in our bedroom, mostly for controlling the Hue smart light bulbs we have throughout both of those rooms. The kitchen, dining area, and living room are one combined room in our two-bedroom apartment, so it's convenient to tell Alexa "turn on the living room lights" when we walk into a dark apartment, hands full of groceries, shopping bags, and the like.

Aside from light control, we use Alexa to set timers and alarms. My boyfriend uses the Echo Dot in the bedroom as his "alarm clock" and occasionally as a speaker so he can listen to music while getting ready in the mornings. Connecting Spotify to our Echoes makes it easy for us to ask Alexa to play specific songs and playlists.

There are a handful of commands we use all the time with Alexa, and a few we probably would never use. For example, I'll probably never ask Alexa to place an Amazon order for me. I always go through my multiple wish and shopping lists before I place an order for an item I need, just to see if I can combine more items into one order.

Alexa rarely mistranslates anything we say—she's pretty good at understanding what we ask of her. However, there have been plenty of times that Alexa has heard us say her name when we haven't said her name at all. That typically happens when the TV is on, but once or twice she's piped up eerily when the apartment was otherwise silent.

— Valentina Palladino, Associate Reviewer

The Sonos One next to a Google Home.

Jeff Dunn
Such as making a video call, like these two who are in no way concerned about Alexa's privacy implications. [If you're seeing this image in the Ars Staff Voicebot article, this caption makes limited sense.]
New features for the Siri watch face.
The touch-sensitive pad (in the middle of the volume adjuster wheel) glows when Cortana is listening or talking.

Valentina Palladino
The Invoke costs $199, twice as much as the new Amazon Echo.

Valentina Palladino
The hardest part is figuring out where to put the Sonos' power brick.

“Eventually I got sick of it”

I have several Amazon Echo and Google Home devices, and I mainly use them to change my Nest thermostat temperature, add food items to my supermarket shopping list, control my Sonos speakers, and get quick updates such as the weather or sports scores. I used to play the daily Jeopardy! game on Amazon Echo, but the voice recognition didn't always work well, and I eventually got sick of it. I can change the channel on my FiOS set-top box using the Echo, but it can't turn the TV on or off; I usually just use the remote control for TV.

I mostly use Siri to set timers on my watch while I'm making food, or to set reminders.

— Jon Brodkin, Senior IT Reporter

Silliness abounds

The MacBook Air that Ars gave me has Siri, which lets me search my files and the Web. That's unquestionably dope, but no force on Earth can make me speak to it in a normal, non-stupid voice.

— Peter Opaskar, Line Editor

Sorry, Bezos

I'm not personally fond of Ars Technica endorsing anything other than "Don't use these things until companies become more responsible and offer no-cloud versions of their digital assistants."

I disconnected my Echo Dot after reporting on an accidentally forwarded voice-memo story. Shortly afterward, I realized that I had really only used mine to do three things: check the weather, set an alarm/timer, and play a terrestrial radio station. Other skills (particularly searching for and queueing up podcasts) were a monumental exercise in frustration, so I opted for simpler queries by default.

I don't need a voice assistant fully tied to the cloud and constantly listening to and analyzing my home audio when my searches are so limited and predictable. That's not enough practical payoff for how much I sacrifice as an Echo user in terms of privacy and security. I'm aware that the sheer use of a smartphone opens me up to serious privacy breaches and that my anti-Echo attitude might best be served by throwing all of my electronics in a landfill and living in a hole. But I still believe Echo's always-on demands and limited usefulness cross my "this is too much" line.

That being said, the value of a hands-free, voice-activated computer system cannot be overstated for many users, whether because they're physically limited or deal with hands-full situations on a regular basis (in a family, at a workplace, etc.). For those cases, I wish a product would emerge that valued audio-recognition privacy by limiting all audio analysis to local hardware—and only sent and received pre-approved search text. The result might be a slower or clunkier product, or it might require goofier phrases. But I see no value in asking what a puck-shaped robot thinks of Jeff Bezos or having it be so obsessed with my natural speech patterns that it goads me into asking for jokes or stories.

The audio-recognition devices should be the stupid, dehumanized slaves in this equation, not us. Our conversations and comments should not be freely processed for bigger companies' use.

— Sam Machkovech, Tech Culture Editor

Let's block ads! (Why?)