Users aren't stupid. They use whatever input method makes sense, whether that's Siri, a touchscreen, or something else.

RTR2SCBE-615.jpg

Reuters

Now that Siri's been on the market for about five months, little bits of data are beginning to roll in that can give us a picture of whether people are finding her useful, and what sorts of things they are finding her useful for.

A new study from Parks Associates helps to flesh out that picture a bit. The data show that just over 60 percent of iPhone 4S owners are using Siri for some sort of task at least several times a week. Functions that people use their phones for regularly -- calling, texting -- are the most common functions for people to execute with Siri. And, generally speaking, the less common a function -- watching movies, scheduling meetings -- the fewer people are doing it with Siri.

But such trends are basically inevitable if you frame your survey questions as "Do you use Siri for X several times a week?" What's really interesting, and what we don't have good data on -- yet -- is whether Siri will be used relatively more for certain tasks than for others.

To try and understand a bit more about how people decide whether they want to use a voice-activated or touch-screen activated interface, I spoke with Alex Rudnicky, an expert in human-computer interaction at Carnegie Mellon University. The process he described is something like a marketplace, with voice and touchscreen interfaces competing for a user's attention: The user weighs a host of factors, some circumstantial (am I in a crowded bar that will make it hard for Siri to hear me?) and others about the inherent nature of the task (Siri may prove better at "find a cheap Chinese restaurant within a mile" than Yelp's interface, which requires you to sort through many options to nail that request). 

This description fits in well with some of the Parks Associates qualitative findings, in which many people told them they use Siri when driving or otherwise have their hands full, John Barrett of the Parks Associates told me. Calculating the costs and benefits of each, a user will go with whatever's easier, even if, Rudnicky says, they find the AI "annoying." (Of course, this raises the question of just what it means for AI to be "annoying." It seems that the robotic qualities people find annoying might cease to be so, if the AI were to just do it's job effectively and quickly.)

Beyond these sorts of functional concerns, users may also weigh cultural norms. A study done by some human-computer interaction experts and Carnegie Mellon and SUNY Buffalo found that people preferred to give feedback via text than vocally to a robot (pdf), perhaps because they became sheepish talking to a robot around strangers. But if you've ever taken a public bus or subway in recent years, you know how fragile such inhibitions are. Rudnicky said we can expect self-consciousness about talking to robots to melt away over the next few years, much as it has for talking on a cell phone.

Add to all of these factors the fact that Siri isn't competing against any old interface, but the touchscreen of an iPhone -- the iconic phone's haloed hardware -- and the interfaces of the best apps on the planet. All things considered, Siri's got a lot stacked up against her, which makes you wonder how high people's expectations must have been when you see headlines like this one, "People Are Barely Using The iPhone 4S's Assistant, Siri," plastered over reports of the Parks Associates' study.

Voice-activated interfaces including Siri aren't going to be the only way we interact with a device. What changes as these technologies improve and proliferate is the diversity of options. Where we once had only a keyboard, we then had a keyboard and a mouse. And then we had touchscreens. And now we have our voices. Not every interaction will be guided by speech, but the option is there, and on certain days, in certain car rides, for certain kinds of tasks, it will be the option we choose.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.