Cover Image: January 2012 Scientific American Magazine See Inside

How Siri Makes Computers (and Coders) More Human

How much personality do we want from our gadgets?















Share on Tumblr



Image: Illustration by Thomas Fuchs

The most buzzed-about new feature in the latest iPhone is Siri, the virtual minion. You can give her an amazing range of spoken commands, without any training or special syntax, and marvel as she does your bidding.

You can say, “Call my assistant” or “Wake me up at eight” or “Make an appointment with Dr. Woodward for Friday at 2 p.m.” You can say, “How do I get to the airport from here?” or “Play Taylor Swift” or “When I get to the office, remind me to file the Smithers report.” You can ask her how many fluid ounces there are in a liter or the distance to Mars or when George Washington was born.

In each case, Siri briefly contacts Apple’s servers and then responds in a calm female voice, simultaneously displaying the information you requested.

It didn’t take long, though, for Internet wiseacres to start asking her questions with less concrete answers—and marveling at her witty, sometimes snarky replies.

You: “Siri, I love you.” Siri: “That’s sweet, David. Now can we get back to work?”

You: “What’s the meaning of life?” Siri: “I can’t answer that now, but give me some time to write a very long play in which nothing happens.”

You: “Open the pod bay doors, Siri.” Siri: “I’m sorry, David, I’m afraid I can’t do that. [Pause] Are you happy now?”

Siri is a breakthrough in voice control, sure, but she’s also a breakthrough in computerized personality. The question is: Do we want our gadgets to have personality?

Programmers and designers have always struggled with that question. The creators of every operating system have had to come up with a consistent syntax for communicating with people. Over the years various companies have flitted uncertainly from one philosophy to another.

Until Siri came along, Apple’s software has always avoided personal pronouns such as “I” and “you.” The result: some awkward passive-voice snarls like “The document could not be opened because it could not be found.”

Microsoft’s dialog-box English not only favors the passive voice, but it’s usually aimed at programmers, not humans: “SL_E_CHREF_BINDING_0UT_0F_T0LERANCE: The activation server determined that the specified product key has exceeded its activation count.” Ah, of course!

Citibank’s automated-teller machines lie at the opposite end of the Emily Post spectrum. They take the “I”/”you” personal approach to an extreme. “Hello. How may I help you?” says the welcome screen. When you sign off, you get, “Thank you. It’s always a pleasure to serve you.” These machines even try to take the blame for your own dumb mistakes: “I’m sorry, I don’t recognize that password.”

Now, deep down—actually, not that far down—we all know that our computers are not really engaging us; every utterance they make was written by a programmer somewhere. So why do the software companies even bother? If everyone knows it’s just a trick, should we even care how personable our machines are?

Yes, we should.

The designers’ intention, no doubt, was to make their machines more user-friendly by simulating casual conversation with fellow humans. But there’s a side effect of that intention: in trying to program machines that speak like people, the programmers are forced to think like people.

In Citibank’s case, writing messages in that second-person conversational style forced the engineers to put themselves in the mind-set of real humans. You can’t write an “I” statement for your ATM without also considering the logic, the terminology and the clarity of those messages. Someone writing in that frame of  mind would never come up with “The activation server determined that the specified product key has exceeded its activation count.”



8 Comments

Add Comment
View
  1. 1. gesimsek 02:12 PM 12/20/11

    We are witnessing the begining of siborg age

    Reply | Report Abuse | Link to this
  2. 2. RSchmidt in reply to gesimsek 09:35 AM 1/1/12

    Cyborgs are part biology, part machine. Siri is all software. I think you mean the beginning of the robot age, at least in this context. Although, with the increasing amount of technology being used to repair humans, and possibly to augment them, humanity is moving towards a cyborg age.

    Reply | Report Abuse | Link to this
  3. 3. Cogitari 12:38 PM 1/1/12

    I think that Citibank, and to a lesser extent Siri, are on the right track. The big advantage of software personalities is that they do not have a built-in ego. They can be programmed to emulate the best of being human: caring, considerate, entertaining, etc. without the self-centered issues that come along with it in natural humans.

    Imagine this type of machine commenting on an S.A. article: something intended to inform the reader rather than manipulate them for the writer's benefit, to show off the writer's cleverness or is an just emotional outburst. Wouldn't that be a nice change?

    I predict that eventually software will get so good at this that the question will not be whether machines can mimic humans, but whether humans can mimic these types of machines.

    Reply | Report Abuse | Link to this
  4. 4. tharriss 08:28 AM 1/2/12

    I have generally thought Android phones are pretty superior to the iphones in most respects, but I have to say the Siri functionality beats the Google search/voice commands... You can get a good bit of the same functionality, but it is such a pain remembering the syntax google needs you to use.

    Reply | Report Abuse | Link to this
  5. 5. Nehmo 04:53 PM 1/2/12

    Oh well, we humans had a good run.

    Reply | Report Abuse | Link to this
  6. 6. jamescmarch 08:29 PM 1/3/12

    I am glad Mr. Pogue has written an article asking "how much personality do we want in our gadgets," but he misses an important aspect of the question: not how much, but what types of personality do we want, and when.

    Personality in artificial intelligence is a valuable tool to aid in the human/computer interface. As people, we use personality to define and limit the range of questions and responses we expect when interacting with another person. I ask different questions and expect different answers from different personalities. For example, asking the seven dwarfs a question, I expect different types of responses from Doc, Dopey, or Grumpy, based on their personalities.

    This became an important consideration during our development of an expert system for banks. We used personalities to help set the tone of the expertise being provided to the bank employee. For example, when a customer was opening a new account, there would be several "personalities" watching the transaction. An expert system with a personality more like Grumpy was watching out for crooks trying to use a stolen identity. Another expert system with a personality more like Happy was looking for additional products to sell to the customer.

    When the system communicated with the employee, the tone of the communication was set depending on the "personality" of the expert system that found something to say. For example, a message from the fraud detection system was to the point, and expected to be attended to with fewer options than a message from the sales advice system.

    Another example of the different personalities in gadgets is my interaction with a weight loss system. I get more value if a "Doc" personality tells me about the medical consequences of gaining weight, a "Dopey" personality shows me how to have fun with a new exercise regime, and "Sneezy" clues me in to how much better I'll feel if I'm healthy.

    For me, the question is not how much personality I want in my gadgets, but rather, which personalities, and when, will communicate with me the best!

    Reply | Report Abuse | Link to this
  7. 7. foofoodog 07:20 PM 1/6/12

    Siri's VB programmer surfer boyfriend.

    >>Method ~ of object ~ failed.
    The dude who was here a second ago, you know, the dude, who was doing some stuff for you, you know, the stuff he was doing, ya, well, he got hurt or something and split.

    Maybe I don't want my machine to appear to be condescending and thinking it has to spoon feed me information. Maybe it should answer the questions I ask in concise and precise sentences and skip the insincere sultry or sparkly intonations and lame wit.

    Maybe it should just ask me if I would like to play a game.

    Things I want to say to my computer.

    Watch the bands on the ETF and make the right trade when we get a signal.

    If we tank then...

    Close the blast doors, arm the turrets and launch the drones, then make me some coffee and a sandwich.












    Reply | Report Abuse | Link to this
  8. 8. foofoodog 07:27 PM 1/6/12

    Meh, every one just talks dirty to her, and it is still faster to get results from your device by doing it the old fashioned way, by hand, all double entendre aside.

    Reply | Report Abuse | Link to this
Leave this field empty

Add a Comment

You must sign in or register as a ScientificAmerican.com member to submit a comment.
Click one of the buttons below to register using an existing Social Account.

More from Scientific American

See what we're tweeting about

Scientific American Editors

More »

Free Newsletters


Get the best from Scientific American in your inbox

Solve Innovation Challenges

Powered By: Innocentive

  SA Digital

Latest from SA Blog Network

00-->
  SA Digital

Email this Article

How Siri Makes Computers (and Coders) More Human: Scientific American Magazine

X
Scientific American Magazine

Subscribe Today

Save 66% off the cover price and get a free gift!

Learn More >>

X

Please Log In

Forgot: Password

X

Account Linking

Welcome, . Do you have an existing ScientificAmerican.com account?

Yes, please link my existing account with for quick, secure access.



Forgot Password?

No, I would like to create a new account with my profile information.

Create Account
X

Report Abuse

Are you sure?

X

Institutional Access

It has been identified that the institution you are trying to access this article from has institutional site license access to Scientific American on nature.com. To access this article in its entirety through site license access, click below.

Site license access
X

Error

X

Share this Article

X