We’re quickly becoming accustomed to having voice-activated assistants, but I think voice right now is still an inferior interface for human interaction.
Let’s talk about talking to our stuff.
You’ve probably been speaking to your electronic devices for years — I have, and not always with family-friendly language — but we’ve finally crossed that threshold where it’s not crazy to talk to equipment and expect a reasonable reply.
If you own an Amazon Echo device with Alexa, you know what I mean. I’ve resisted Alexa in my house because I’ve been uneasy about what information it’s sending back to Amazon. The recent spate of Echoes spontaneously laughing for no apparent reason hasn’t helped, either. (Amazon later determined it was mistaking some noise for “Alexa, laugh,” but that doesn’t crank down the creepiness factor.) However, many people have one or more Echoes in their houses, which is quickly normalizing the idea that it’s OK to talk out loud to devices.
I bring this up because I’ve been using an Apple HomePod for several weeks, and I agree with reviews by people who know more about audio quality than I do that it’s a speaker with great sound (especially from something so small, just 6.8 inches tall and 5.6 inches wide). The technology inside, which analyzes the area the HomePod is in and adjusts its output accordingly, is impressive and well executed.
It also has a feature that feels revolutionary and Apple-like in the same way AirPods work seamlessly among all of my devices: The microphones pick up spoken commands without you having to raise your voice, even when music or other noise fills the room.
What doesn’t feel revolutionary is the Siri integration. It’s the same Siri available on your other Apple devices, only in this case it’s limited because there’s no screen to fall back on when it encounters queries it can’t answer audibly (even those you would think could be read aloud, like what’s on your calendar for the day). Or you make what you think is a reasonable request and Siri misunderstands what you said. Or Siri plays a version of a song that’s a poor cover of the one you wanted. Or you get a different recording of the same song, by the same artist, than what’s in your library, something that happens to me regularly using Apple Music on my devices.
But I’m not here to just bash Siri. The most frustrating aspect is that we’re quickly becoming accustomed to having voice-activated assistants, but forgetting that they’re still relatively dumb.
Right now, I think voice is still an inferior interface for human interaction — which seems backward.
First, you need to keep a dictionary in your head of which commands and tasks will be recognized, and remember the syntax for making the request. The idea is that Siri and Alexa recognize “natural-language” queries, but they usually don’t parse well. I can be in the kitchen and ask Siri to add eggs to my grocery list, which works, but if I say, “Hey Siri, add eggs, milk, butter and bread to my grocery list,” I’ll get one to-do item labeled “eggs, milk, butter and bread.”
As another example, I do really like having HomePod as a point of contact for interacting with my HomeKit devices. Until my review unit arrived, it was really no trouble to raise my iPhone or Apple Watch, say “Hey Siri, turn on the office lights.” Now I only have to proclaim it to the empty living room (my daughter’s parakeet must think I’m nuts).
However, I’ve also had to train myself. I would say, “Hey Siri, turn off the office lights,” but HomePod would hear that as, “Hey Siri, turn off” and stop listening — which is all manner of frustrating. But if I reordered the command as, “Hey Siri, turn the office lights off,” it would work every time. Curiously, using the former construction recently has worked just fine, so perhaps Apple made some adjustment on its side.
The second reason voice stutters as an interface, at least with Siri, is that you have to keep in your head which Siri devices are capable of which tasks. For example, my iPhone can display which appointments are scheduled for later in the day, and can read them to me without requiring that I look at the screen, but HomePod can’t. I can control lights, switches and other HomeKit accessories using the HomePod, iPhone, iPad or Apple Watch, but not using Siri on my Mac or Apple TV.
Most Read Business Stories
Coincidentally, as I was writing this column, The Information published a lengthy article about why Siri has stumbled over the last seven years (subscription required), which boils down to several groups of services, some written in different programming languages, that are patched together to form what we on the outside view as “Siri.”
The upside to all this is that development on Siri (and Alexa) happens on the server side, so we’ll be able to take advantage of improvements over time. The question is how fast the technology will mature.