First Experiences with Siri

Speech recognition has been around for decades.  Apple took a stab at it many years ago by asking volunteers come out to Apple HQ and read a canned piece of text into the computer with the idea that they would average the voice prints to help the machine figure out what was being said.  Volunteers got a shirt that said, "I helped Apple Wreck a Nice Beach!" which sort of illustrated the problem computers have with the whole thing.

For some time now, I've been saying that the next paradigm in computer interaction would be speech recognition.  But calling it speech recognition is a bit of a misnomer.  As I mentioned, computers have been able to recognize speech for some time.  What they haven't been able to do is recognize language.  If you haven't seen this video before you'll see what happens when you take this stuff for granted...



Compare this to where Apple is now with Siri:



But as Scot Forstall says, the real technology advancement is in the way I can phrase things any way I want and Siri understands.  Speech recognition is simply not going to be used if we have to memorize how to say what we want.  I have speech recognition in my Camry but I'll be damned if I can remember what the commands are and never get them correct.  It's like trying to get a two year old to make you a pot of coffee.  Siri does a fantastic job at overcoming these limitations but, as I'll explain, there are limits which is why Apple still has the technology in beta.

For instance, if you were my assistant and you had my phone and I asked you to "send Jesse a text message to tell him I love him" you'd send him a text message from me that said "I love you".  But If I ask Siri that same thing, it sends him a text message that says, "I love him" - a potentially relationship ending conversation may ensue.  It doesn't know that Jesse is a he and is the "him" I'm referring to.  If I slightly alter my wording to say, "Send Jesse a text message to tell him, 'I love you'" it does the right thing, but now I feel weird.  I feel like I'm telling him I love Siri.  The way to phrase it that both agrees with my sensibilities and conveys the right message is to say, "Send Jesse a text message that says 'I love you'".  But we're still a little too uncomfortably close to needing to say things a certain way.

Another example that threw Siri for a loop was when I asked it to remind me to do something "30 minutes after I get home".  It heard the 30 minutes and set the reminder for 30 minutes from NOW and placed it on my "Home" reminder list.  Many of the other issues I've had are likely caused by my helmet microphone which probably sounds too distorted compared to the iPhone's built in mic.  For example, I once asked her while riding, "Play my riding playlist".  It kept saying, "I'm sorry, Sean, I can't find a playlist called 'writing'".  Changing the title of the playlist helped a little.  When using the iPhone's mic, however, I'm very impressed.  The other day, I gave a command while holding the phone in front of me while there was a loud conversation going on around me and I was certain she'd mess it up but she got it right.  I can increase the odds of her hearing me by bringing the phone to my face.  She immediately activates and asks me what she can do for me.

The biggest problem I've had with Siri stems from me becoming reliant upon it.  Apple had it down a large part of the weekend for some reason.  I kept getting "I'm sorry, I'm having a difficult time reaching the network."  I missed it!  I can give her a request in one second that could take 30 seconds to a minute to do manually so I often rely on her.  Apple's got to make sure she stays up 24/7.  If you relied on an assistant in your office and she randomly disappeared for hours at a time, you'd eventually become disenchanted and fire her.  It's pretty frustrating to belt out a long request (especially in front of others) and have her come back with "Sorry, I can't do that right now" and you don't know whether to try again right away or do the task manually.  Millions of people purchased iPhone 4S's just to get access to her.  Beta or not, it needs to stay up.

I've also had some thoughts about security.  Who's actually deciding what to do when I say something?  I say "play my motorcycle playlist" and that audio is sent to Apple which interprets what I say, sends it back to Siri on my phone but do they also send the command to the phone to play the right playlist or does Siri now interpret the text received from Apple and determine what to do locally.  It matters.  Can you imagine what would happen if the wrong person could somehow get in the middle and figure out how to use Siri's hooks into your phone to make every iPhone in America call someone at the same time?  I suspect Apple is already on top of this because when I asked Siri to delete a contact, she says she's not allowed to delete things.  I'm sure that as the technology matures, we'll hear more about this.

Much has been made of Siri's personality.  I find her to be cordial but she gets frustrated with me.  She'll say my name at that point.  Something like, "Sean, I don't understand what you're saying."  Usually this is the third or fourth try and it's pretty rare.  But I can see things getting like this one day:

Attitude aside, I'm beginning to wish Siri was in more places.  I wish she was in my Mac.  I'd love to be able to say, "Open my Email, launch Skype, launch Filemaker, and open up the document I was writing last night" or "Reply to this message and tell him I said, 'This is fine but I'll need to make some edits to the schema which will affect our release date'".  I wish she was in my TV.  I wish she was in my car.  I wish my house had Siri.  How cool would it be to say, "Set the temperature to 68 and turn off all the lights when I leave" or "Don't water the back lawn for the next three days".  I promise you we're heading there.  If I know Apple, they're working on figuring out how to make Siri a central part of our lives outside of iOS devices and I think this is an exciting time to be alive.