I dislike voice interfaces

Last year I bought a new car.  And I mean a new new car - a 2019 model year right.  My last two "new cars" were low-mileage "pre-owned" (i.e. "used") cars.  They were fine, but they didn't have any bells or whistles.  In fact, my last one didn't even have power locks or windows.

The new car has all that stuff, though.  And one of those bells and whistles is an entertainment center with a touch screen and support for Android Auto.  This was actually something I really wanted, as opposed to having just the integrated options.  My reasoning was that with Android Auto, I can keep the "brains" of the system in my phone, which can be upgraded or replaced, whereas integrated systems are likely to become outdated and maybe even discontinued before I get rid of the car.

The Reality

The reality of Android Auto, however, is not as great as I'd hoped it would be.  Much of the reason for this is that it's primarily driven by the voice-control features in Google Assistant.  There's some support for typing and menu-driven UI, but it's intentionally limited for "safety reasons."  For example, you can't type into Google Maps while you're driving, nor can you scroll beyond a certain limit in the Plex app, because paying too much attention to the screen is dangerous.

You may have noticed I put "safety reasons" in "sarcasm quotes".  That's because the voice control can sometimes be so frustrating that I find myself more distracted by it than if I could just search for what I needed from a menu.  I end up angry and yelling at the console or just picking up my phone and using it directly rather than the car's console interface.

Let me give you an example.  While driving with my son, he asked to listen to some music.  He wanted to listen to "My Songs Know What You Did in the Dark (Light Em Up)" by Fallout Boy.  So I said to Google, "Plan light 'em up by Fallout Boy."  Google replied, "OK, asking Amazon Music to play 'My Songs Know What You Did in the Dark (Light Em Up)' by Fallout Boy."

Great!  The music started playing and I heard, "Do you have the time, to listen to me whine."  I looked at the screen and Amazon Music was playing Green Day.  Why?  I have no idea.  So I tried again and asked Google, "Play My Songs Know What You Did in the Dark by Fallout Boy."  Once again, Google replied "OK, asking Amazon Music to play 'My Songs Know What You Did in the Dark (Light Em Up)' by Fallout Boy."  And this time, it played the right song.  It claimed to be asking Amazon the same thing both times, so why did one work and the other didn't?  Who knows?

This wouldn't be a big deal if it were an isolated incident, but it's actually pretty common when using Android Auto.  Sometimes it gives you what you ask for, sometimes it doesn't.  Sometimes is claims it's giving you what you asked for, but gives you something else.  And sometimes it just doesn't give you any response at all.

And it's not just Google - I have an Amazon Echo and Alexa does the same sort of thing.  It seems to be a bit more reliable than Google, but it still has problems.

When it works...

The thing is, voice interfaces are really great...when they work as you'd expect.  When you know what you want and the system correctly understands you and maps your request to the right thing, they're like the computers on Star Trek.  And to be fair, they do work well often enough to make them useful.  It's just that when they don't work, I find them unreasonably infuriating.

I think the reason for this is that the discoverability of voice interfaces is limited.  On the one hand, yes, you can just ask your device anything.  But the world of "things you can ask" is so huge that it's overwhelming - it's hard to tell where to start.  And when something as simple as "play X song" doesn't work, it's not obvious where to go next.  Try also including the artist?  Maybe a variation on the title?  All you can do is keep trying random things and hope that one of them works.  Sometimes you stumble on the right thing, and sometimes you don't and just give up in disgust.

It's kind of like trying to use a command-line interface, but with no help and no manual.  OK, so the one command you know didn't work.  What next?  Well, you can try some variations on the arguments you passed it.  Maybe sprinkle in a common option that might be applicable.  But ultimately, it's just blind guessing.

When you're using a graphical interface, you still end up guessing, but at least it's a bounded search space - you're limited to what's on the screen.  Also, the options usually have at least some description of what they do.

If nothing else, these issues with voice interfaces are a good way to relate to non-technical users.  You've seen non-technical users get confused, frustrated, and angry when they encounter a task that they don't know how to accomplish or a problem that they can't seem to fix?  Well, that's how I feel when Android Auto tells me one thing and then does another.  Maybe they'll fix that some time soon....

You can reply to this entry by leaving a comment below. This entry accepts Pingbacks from other blogs. You can follow comments on this entry by subscribing to the RSS feed.

Add your comments #

A comment body is required. No HTML code allowed. URLs starting with http:// or ftp:// will be automatically converted to hyperlinks.