Follow Slashdot blog updates by subscribing to our blog RSS feed

 



Forgot your password?
typodupeerror
×
AI

Talking To Computers? 395

merlock18 writes "Is it un-natural to talk to a computer? After discussing the outcome of the Jeopardy game with some colleagues, they seem to think it is mildly 'scary' to talk to a computer and have it competently talk back. Is this what everyone thinks? I was thinking to myself how much I would like to be able to even tell my computer to open programs by telling it vocally. A simple idea that I am fairly surprised is not common. Am I a minority in this one? Do people just not like the idea of talking (without cursing) to a computer, let alone have it act or reply? Would anyone else be interested in building their own mini-Watson, or is this just scary?"
This discussion has been archived. No new comments can be posted.

Talking To Computers?

Comments Filter:
  • Privacy (Score:5, Funny)

    by JPLemme ( 106723 ) on Thursday February 24, 2011 @12:50AM (#35296840)
    I can't speak for anybody else, but a lot of the time I don't *want* people to overhear what I'm asking my computer to do...
    • by Hojima ( 1228978 )

      "Set reply mode: Samuel L Jackson"
      "ENGLISH MOTHERFUCKER, DO YOU SPEAK IT?"

      (Then find out how many 'whats' it takes for it go go skynet on you)

    • passwords (Score:4, Funny)

      by roman_mir ( 125474 ) on Thursday February 24, 2011 @05:38AM (#35297902) Homepage Journal

      Sitting at a Starbucks:

      -Computer, open my bank account.
      -Which one?
      -Bank of America
      -That's a stupid bank account to have, they are broke
      -Not as long as Bernanke keeps bailing them out.
      -Fine. But your dollars are crap.
      -Whatever. Open it.
      -It wants your password.o!
      -12345
      -So the combination is... one, two, three, four, five? That's the stupidest combination I've ever heard in my life! The kind of thing an idiot would have on his luggage!
      -Remind me to change the combination on my luggage. And what's the balance on the account?.
      -15 bucks
      -Yaho! I am gonna buy me a mouse and I'll make you shut up!

      ---

      A day later:

      -Computer, open my bank account
      -Same stupid account as yesterday?
      -Shut up and open it, and what's the balance?
      -Negative 1000
      -????!!!!

      • Re:passwords (Score:4, Interesting)

        by icebraining ( 1313345 ) on Thursday February 24, 2011 @08:22AM (#35298516) Homepage

        Reminds me of Phillip Dick's Ubik:

        The door refused to open. It said, "Five cents, please."
        He searched his pockets. No more coins; nothing. "I'll pay you tomorrow," he told the door. Again hetried the knob. Again it remained locked tight. "What I pay you," he informed it, "is in the nature of a gratuity; I don't have to pay you."
        "I think otherwise," the door said. "Look in the purchase contract you signed when you bought this conapt.
        "In his desk drawer he found the contract; since signing it he had found it necessary to refer to the document many times. Sure enough; payment to his door for opening and shutting constituted a mandatory fee. Not a tip.
        "You discover I'm right," the door said. It sounded smug.
        From the drawer beside the sink Joe Chip got a stainless steel knife; with it he began systematically to unscrew the bolt assembly of his apt's money-gulping door.
        "I'll sue you," the door said as the first screw fell out.
        Joe Chip said, "I've never been sued by a door. But I guess I can live through it."

    • by methano ( 519830 )
      It was my understanding that Watson didn't actually get the answers from reading or from listening to Alex, but was, instead, fed the answers via text file. So, although this is a great conversation to have, it shouldn't have started from Watson on Jeopardy.
  • by CohibaVancouver ( 864662 ) on Thursday February 24, 2011 @12:52AM (#35296844)
    Open the pod bay doors, Hal.
  • Not only privacy but the standard office would sound like a bar of a busy Friday night. Can you imagine loud howard dictating a document just over the cubicle wall?
    • Not if the voice interface was limited to certain features or functions, and always under user control. If you don't want to speak, just keep using the mouse. Speak only when it's easier than mousing or keyboarding.

      Loud Howard is another issue, altogether.

      • by Archangel Michael ( 180766 ) on Thursday February 24, 2011 @01:30AM (#35297064) Journal

        gawk; grep; unzip; touch; strip; init,
          uncompress, gasp; finger; find,
          route, whereis, which, mount; fsck; nice,
          more; yes; gasp; umount; head, halt,
          renice, restore, touch, whereis, which,
          route, mount,
          more, yes, gasp, umount, expand, ping,
          make clean; sleep

        • TNG Commands ... (Score:3, Insightful)

          by bronney ( 638318 )

          tea, earl grey, hot :D

          • Only in an American show would you have to specify HOT Earl Grey tea. After all, you wouldn't ask for "popsicle, lemon, cold"
            • They locked 'popsicle' out of the computer after Riker snuck up behind someone and yelled "Rectally!", just after someone got as far as "lemon".
        • by fishexe ( 168879 )

          gawk; grep; unzip; touch; strip; init, uncompress, gasp; finger; find, route, whereis, which, mount; fsck; nice, more; yes; gasp; umount; head, halt, renice, restore, touch, whereis, which, route, mount, more, yes, gasp, umount, expand, ping, make clean; sleep

          make clean before sleep? Too much work...

  • when swearing at them improves their performance.
    • They are dong that for years, probably you are dong it wrong, you need to tell them that you are about to pull the plug...
  • by Waccoon ( 1186667 ) on Thursday February 24, 2011 @12:54AM (#35296854)

    Just give it 30 years. Once it becomes publicly available, it only takes one generation for society to get used to new tech.

    Personally, I find it impressive but annoying. I'm already driven nuts by people talking on cell phones all day, and I don't want to hear and endless stream of command instructions, either.

    • by Skidborg ( 1585365 ) on Thursday February 24, 2011 @12:56AM (#35296864)
      Or the results of a recently fired employee raging through the office roaring "SELECT ALL! DELETE!"
      • by Imrik ( 148191 )

        Or someone making a Youtube video and trolling with it.

      • Smart move. Especially when you are going to be looking for new job soon. Kinda hard to explain to new employer that you will not do it in his company too...

      • by Kosi ( 589267 )

        First thought: I would have a special "you are fired" routine integrated into the system, that instantly revokes all access of the fired person. Except when that's me.

        To get a little more real here: When computers really learn to talk, you could use one to talk i. e. in your bosses voice to another computer, so authentication via voice only wouldn't make any sense. Also because of the privacy/annoyance matter, I think it much more likely that when we will control computers to a greater part via voice, it wi

    • by Kjella ( 173770 )

      Personally, I find it impressive but annoying. I'm already driven nuts by people talking on cell phones all day, and I don't want to hear and endless stream of command instructions, either

      I doubt you will, voice communication is much slower, error prone and most people would rather type all day than talk all day.

      I think the advantage is more if you can replace the whole system with a voice command like a train ticket "one adult from [station] to [station], please" and it'll pop up the (hopefully) right thing, if not you can try again or dig through the selection like today. Hell, I'd be pretty happy for an elevator that'd understand what floor I wanted to go to. I guess some of this exists b

      • I think you've hit on the core problem. Speech would be a very poor input method for our current computer usage models. However, speech could be useful in communicating our desires to a computer, which then carries them out in a much more autonomous fashion than today.
        • speech could be useful in communicating our desires to a computer, which then carries them out in a much more autonomous fashion than today.

          "KILL THEM!"

      • I'm surprised that you say that voice communication is much slower - if anything, I find it faster than most of the other methods available. Just see the amount of data that you can cram in a 10 minute discussion, compared to the endless roundtrips with e-mail and the tedious speed of IM.
        Of course, I am talking about natural voice communication, not about the very poor voice interfaces that we currently have with computers.
    • I hardly think it would take 30 years. Thanks to TV, movies, and books, practically every person alive today has probably been expecting to converse with computers since they were a kid.

      If someone invented a Jetsons-esque flying car tomorrow, I'm sure people would describe the experience as "scary". For about a week. Then they'd wonder how they ever lived without them.

    • Best "use case" for voice commands is not constant talking to computer and replacement for keyboard/mouse.

      Instead, think of possiblities where it makes actual sense, where people want to interact with computer, but walking to it and pressing buttons is not really good way, examples:

      "Next Slide", "Previous Slide" (yay for powerpoint meetings).

      "Next Song", "Next Radio Station" (I am at kitchen, cooking, dont want to listen to particular song, dont want to dirty the mouse).

      "Check Connection" (I am fiddling wit

      • "Next Slide", "Previous Slide" (yay for powerpoint meetings).

        I've given presentations like that, in a room where the computer had to be at the back, so someone else had to press the next slide button, and it's horrible. You quite often want to jump between slides in the middle of a sentence, so having to interrupt your communication with the audience to communicate with the computer is horrible. A big buzzword in HCI at the moment is 'multimodal interaction', and replacing keyboard with voice is just a sideways step in single-modal interaction - augmenting it with

    • Although Watson did not actually hear the announcer (he apparently did on the contestants, however), I think that voice control is only useful when it would be more efficient. Simply barking a bunch of low level commands into a computer (or programming with it!) would not be efficient, since you can probably type/mouse faster. Asking the computer a much higher level question, however, could be a massive time saver, such as: "Watson, what is the current status of Libya?" or "Watson, is there a drug which cou

  • Uncanny Valley? (Score:4, Interesting)

    by DeviantxOne ( 893930 ) on Thursday February 24, 2011 @12:56AM (#35296860)
    I wonder if there has been any research on the uncanny valley for speech...
    • autotune and automated call centres would show that it's not disturbing like it's visual counterpart, just grievously annoying.

  • Every mac OS since 10.0 has had speech recognition - I had some fun with it when it came out, but lost interest after a while. My disenchantment may have had something to do with having to vocalise (for all to hear) every command I made - and can you imagine the yammer of a roomful of computer operators? I'm looking forward to thought-recognition software.

    • Linux and even Windows have had voice recognition for a long time, too.
      Only last week, I enabled it for my dear aunt let's set so double the killer delete select all.

    • by Graff ( 532189 )

      Every mac OS since 10.0 has had speech recognition

      They had it far before that too, back in the MacOS 8 days I believe. It actually worked pretty well, although it was a bit iffy with certain types of background noise.

      These days it's a lot more tolerant of background noise, especially if you combine it with a decent noise-canceling microphone.

  • There is no new revelation about talking to machines, it has been going on since man looked outside of the sacred relationship for some semblance of reasoning about his existence and meaning in life. The Jeopardy 'thing' with IBM was just another step in the direction you can't deny, resistance is futile.
  • The first thing I do when a phone operator robot asks me to say "English" for English or "Espanol" for Espanol, I push all the buttons to see if I can get to a number-based menu, or at least hurt the robot's ears. Saying "English" and waiting for it to confirm that I said English is not faster or more convenient than hitting 1. It's not scary, but it's a computer, and I'm not going to pretend it's not.

    Saying, "Open a command prompt," is in no way more convenient, faster, or easier than slamming the mouse

    • Re:Annoying as hell (Score:5, Interesting)

      by thePowerOfGrayskull ( 905905 ) <[moc.liamg] [ta] [esidarap.cram]> on Thursday February 24, 2011 @01:23AM (#35297012) Homepage Journal
      Agreed; I can't stand the speech recognition on phone systems. For one thing, it universally sucks unless you're only using single words. (I recall calling .. Verizon was it? it: "So you're having a problem. Please tell me what kind of problem you're having." me: "Internet is not working". it: "Okay. Did you say 'Phone line repairs?' ")

      For another, it negates the only advantage (from a consumer perspective) of touchtone menu systems - the ability to quickly navigate when you know your choice ahead of time; or even when you hear it spoken without having to wait for the full menu of options. It seems that most systems allow touchtone interrupt, but don't allow voice interrupt, so if I press "5" for technical support it's fine - but I can't say "technical support" without being forced to listen to all the options.

    • The first thing I do when a phone operator robot asks me to say "English" for English or "Espanol" for Espanol, I push all the buttons to see if I can get to a number-based menu, or at least hurt the robot's ears. Saying "English" and waiting for it to confirm that I said English is not faster or more convenient than hitting 1. It's not scary, but it's a computer, and I'm not going to pretend it's not.

      Especially annoying to blind people! Could you imagine if it were possible to actually get the computer to do complex things with minimal effort by only using your voice!? Imagine if the computer could actually tell you what's going on with it's voice! The horror!

      Saying, "Open a command prompt," is in no way more convenient, faster, or easier than slamming the mouse to the lower left, clicking, and typing cmd.exe. Having it say, "OK, here's a command prompt," afterward would just be annoying.

      Maybe I'm just not picturing the right use case.

      Indeed.

      P.S. Vinux - Linux for visually impaired [vinux.org.uk], Blinux - Blind + Linux discussion group [counterpunch.org] & LinuxSpeaks [joekamphaus.net]

      See also: StarTrek TNG -- Talking to the computer midship instead of having to be at the damn terminal.

      • In the case of Star Trek, there are plenty of terminals, including pads they carry around. The only reason speech ended up being efficient at all was the natural language processing -- you could actually ask the computer a question, and as long as it was a question which could in principle be answered by a computer, the computer would do the work of translating it for you.

        Even so, once the computer found an "answer", often they'd page through it with touchscreens. Saying "next page" isn't that useful.

  • People talk to their pets all the time, and although most pets have just as much of a chance of understanding what's being said as most computers, that doesn't strike people as odd.
  • It seems to me that voice recognition is not the most efficient way to interact with a computer, especially when the user interface is well designed. For complicated tasks, and for interacting with computers where you may not have a normal desk or terminal, perhaps. As far as voice-to-text, if the recognition is accurate, it can possibly increase productivity depending on the person and their typing skills. On another not, however, this is a way for paralyzed individuals to interact with computers without t
    • by fishexe ( 168879 )

      It seems to me that voice recognition is not the most efficient way to interact with a computer, especially when the user interface is well designed. For complicated tasks, and for interacting with computers where you may not have a normal desk or terminal, perhaps.

      Efficient? No. Natural? Yes. I expect the generations growing up after speech recognition and natural-language parsing become both reliable and ubiquitous to use speech as their default interface. They will lose out on many of the efficiency gains a well-designed GUI or CLI will provide, but on the other hand the learning curve to use a computer will be almost flat.

  • Do people just not like the idea of talking (without cursing) to a computer,

    Why, would it be politically incorrect?

  • Teamspeak (Score:4, Insightful)

    by Charliemopps ( 1157495 ) on Thursday February 24, 2011 @01:09AM (#35296920)
    Speech recognition sucks and always will. I can hotkey my way to any program faster than I can say the name of it. Simply double clicking and icon is super easy. Why do I want to have to say "Computer! Open! Porn!!!" when I have a shortcut to all my porn on my desktop? it doesn't even make sense. And entering urls? It would take 10min just to get the url at the top of this article in.

    On a related note: I fucking hate teamspeak. If I wanted to talk to you retarded assholes I'd call one of those party lines. Fuck that, I want to play a video game. I don't want to talk to people. For whatever rudimentary communication I need I can type.
    • Does it? I almost never use the keyboard on my Vibrant to send text messages. I just speak into it and it does a surprisingly good job.

      Now that I'm so used to doing speech to text, using a mouse or a little keyboard on my htpc is incredibly annoying. I really should be able to talk into my remote and say stuff like "play this video." Or navigate by just saying "down, down, play."

      Speech for the PC could be similar. RIght now I'm doing this with my kinect, so we're halfway there.

    • Well, I hate teamspeak too, why force me to use proprietary junk when we have Mumble? But there are, in fact, games where stopping to type, even "rudimentary" things, is a competitive disadvantage, and strong communication is a massive competitive advantage. Natural Selection (not 2) [unknownworlds.com] finally made me get a headset for gaming.

  • Someday (Score:5, Interesting)

    by jmorris42 ( 1458 ) * <{jmorris} {at} {beau.org}> on Thursday February 24, 2011 @01:15AM (#35296946)

    Someday speech will be an important input method. But not any time soon.

    If you have to wear a microphone it isn't ready yet.

    If you have to use a PTT switch it isn't ready yet.

    If you have to repeat or cancel more than 1% of the things you say it isn't ready yet.

    If you have to spend as much time proofreading dictation it has taken down and correcting the mistakes, it isn't ready yet.

    If you have to speak in an unnatural way it isn't ready yet.

    If it won't work in almost any environment it isn't ready yet.

    • If it cannot tell whether you are talking to it or someone else, it isn't ready yet
      If it cannot ignore voices other than the logged in user, it isn't ready yet.

    • by glwtta ( 532858 )
      If you are ever in a situation where there are other people around, it will be totally fucking annoying.

      Notice that that's not a "isn't ready yet" problem, it will always be true.
  • Do people just not like the idea of talking (without cursing) to a computer...?

    I have a PC running Windows; cursing is unavoidable.
    [I spend the rest of my time talking with my stuffed animals.]

    Would anyone else be interested in building their own mini-Watson...?

    Would a Mini-Watson [wikipedia.org] be small and wear a monocle?

  • If it can talk, and you smack it, what happens?
  • Only if the voice input was limited enough, like text input, which we have had for 10+ years, really. I can remember running a Speech to Text program on my 486 that would try to read what I was saying and turn that into a text document.

    Google Voice does the same thing for voice mail.

    Both suffer from a huge problem: accuracy. A conversation with a human doesn't need every word to be perfectly accurate, but something more than a text message is going to fail if some part of the vocal command is incorrect.

    "Did

    • The beauty of the almighty command line, declarative programming language, and any interface with a well-defined command set is that you get exactly what you asked for, at the expense of modifying your thinking to be more exact than natural language. This is a feature. And it's one that I don't think would translate well to the spoken word. I can say

      for file in `ls -1 *.txt`; do echo something about $file; done;

      on a keyboard, but I can't even begin to thing how I could get my mouth to say that out lo
  • Issues (Score:4, Insightful)

    by dcollins ( 135727 ) on Thursday February 24, 2011 @01:33AM (#35297078) Homepage

    (a) Accuracy, (b) Efficiency, (c) Privacy, (d) Noise pollution.

  • by NoSig ( 1919688 ) on Thursday February 24, 2011 @01:34AM (#35297082)
    It's creepy if the computer is trying to pass itself off as a person, because fakeness in social interactions is creepy whether it's a Wallmart greeter or a computer program being fake. If the computer is plainly just presenting itself as a voice interface it won't feel creepy for very long if at all.
  • Much like the animation of human features, there's an uncanny valley in communication that can provoke a strong xenophobic response. If someone or something can respond to *some* conversation but not all conversation, it tickles something deep in my brain that produces an instinctive reflex of distrust and hostility. Watching Watson, I found that the way the interaction was framed as if it was natural conversation put me into this uncomfortable zone where I found myself thinking "KILL IT WITH FIRE" more t
  • I think people find it "creepy" because they've never done it. If it was implemented well on most computers, people would get so accustomed and welcome to it that it would be a huge step back for them to go back to manual input.

    • by xnpu ( 963139 )

      This, beyond anything else, was the purpose of this Jeopardy show off.

      Note that this machine's real purpose is medical diagnosis and that the healthcare bill puts a lot of pressure on lowering costs, including that of diagnosis.

  • by beadfulthings ( 975812 ) on Thursday February 24, 2011 @01:42AM (#35297132) Journal

    I, of course, am now officially older than dirt. A couple of years ago, when I finally got my iPhone, I got the Google search app of course. I used it, it worked, I liked it. When I put the damned phone down, I thought, "If somebody had handed me this when I was fourteen I would have thought it was a phony Hollywood prop." That was when I decided that computers should only be addressed by means of picking up the mouse, pressing one of its buttons, and speaking clearly and distinctly into it in a fake Scottish accent.

  • I thought the movie AI actually was pretty good at wondering about this very thought. If you haven't seen the movie, I thought it was very thought provoking on the idea about what the world might be like if computers ever became super advanced.

  • There are voice actions already, but it takes so much more computational power to make it really-really fast, recognize any accent (which google is having a very hard time with right now) take context into account, and be able to intelligently ask user for clarification. So, I guess in about 5-10 years it will get to the point of Star Trek, where you can address computer and not worry about speech pattern or performing deletion of all files when you said "delay all files".

    Generally voice interface is more

  • the phone based systems that use this suck and some times you need to go nuts on it just to get a real person.

  • Quantizing instructions into unambiguous, 100% clear communication is the pride and joy of the push-button. Vocal communication, especially natural language communication is fraught with ambiguity.

    Think about it this way. Listen to your own conversations with humans for the next week. Count how many times you or your listener asks "say that again", "can you repeat that", "what do you mean by that". Then watch how many keystrokes you miss in the same week.

    I'm cool with a button that launches a nuclear wa

  • Voice recognition has taken the wrong approach from the beginning. The computer listens all the time, then tries to decipher a command in the middle of all kinds of stuff. That's unrealistic. If you gave it a name, an unusual name phonetically (Like "Esmerelda") and only had it follow a command after it heard it's name, then turn off again, then I think we could have working voice recognition RIGHT NOW. "Esmerelda, check my email" "You have 7 new messages". "Esmerelda, play music" Then poof Audacious o

  • Back in the days (must be at least 10 years ago by now) I used the dragon engine from philips to control my pc. To be honest, it had a cool factor but got old soon. Not only was it quite inaccurate at the time, but I found it to be slow in comparison with just mouseclicks. Besides, at night when the family sleeps I don't want to be making to much sounds.
  • There have been good voice-based systems. Wildfire (audio demo) [virtuosity.com] was a really nice voice controlled phone system. The original version, from the late 1990s, used a lot of CPU time and Wildfire accounts were very expensive, about $5 to $10 a day. As CPU power got cheaper, the technology became more available, and Orange offered it on mobiles for a while. Then Microsoft bought the technology, Microsoft never did much with it, although part of it ended up in OnStar Virtual Advisor.

  • Opera already provides a speech interface. Mobile phones routinely provide voice dialling and similar functionality. This technology is already here. Neither is it reliable enough nor is it more efficient than just using a mouse/keyboard. I expect that the main driver behind this is accessibility rather than UX. As for the computer talking back, it's a lot easier to get the computer to say something rather than get it to process human speech input.
  • Eh, you do realize it wasn't actually listening right? The questions were entered through old-fashioned typed text (off screen obviously), not much different from how you would type something on Wolfram Alpha or even Google. The point of the demo was to show of it's capability to search and analyze a huge data set, not it's voice recognition or processing.

    Also keep in mind that the eventual purpose of this beast is to diagnose patients. The show is an important part in making the general public more comfort

  • You can tell a Mac to open programs and many other things by voice for well over 10 years now, and almost nobody uses it. You can tell an iPhone to make calls and many other things by voice for about 2 years now, and people use it only when forced, like when driving.

    I think one reason this kind of thing is unpopular is it makes us realize how stupid computers are. Not even as smart as a small child. You have to construct your sentences even more carefully.

    But I think the main reason it's unpopular is the co

  • On the contrary, I think it's the most natural way to interact with a computer. Back when science fiction authors first started trying to imagine what computers of the future would be like, they almost always imagined speech to be the interface. It was only experience, with the primitive 20th-century interfaces that we had, that habituated people to thinking that keyboards and mice were the way to interact with a computer. The command line and GUI are un-natural ways to communicate, which we've gotten us
  • My first experiences with speech recognition began back around 1990. A computer company I worked for was experimenting with voice control. After being misinterpreted repeatedly, the person trying to use it would inevitably adopt a tone that was either (A) pleading, or (B) infuriated. It was always entertaining to listen to, and my favourite joke of yelling "FORMAT C COLON! YES! YES!" over the cubicle wall never lost its hilarity for me.

    Then there was the submarine simulation game I purchased, which taught m

  • It's not scary because it's not really possible. Even Watson had the questions input before hand - and if a whole show that is dedicated to the process can't even produce a computer that can be communicated to in natural English, at the cost of hundreds of thousands of dollars (millions?), there's no way the average consumer is going to have a system capable of understanding them in a natural environment.

    We simply aren't there yet. Talking to a computer is at best a frustrating, inaccurate experience. Yo

  • A couple of years ago I thought about gluing a whole-house mic system to a text-to-speech engine, allowing me to do simple queries like 'define ', 'weather tomorrow', 'traffic on I-15'... Give the system a unique name ('squizzlesauce?') and make that the key for temporarily enabling speech recognition("squizzlesauce, what's the weather like tomorrow"). Whip up a few scripts to glue the voice-recognition engine to google, a TTS engine for parsing and speaking web results... Then I got a laptop with WiFi and

On the eighth day, God created FORTRAN.

Working...