Well, it sort of works for some limited applications. It doesn’t work well enough for any kind of mission-critical application, or even as an aid for literaure review. That is in part because it is not really true natural language processor; it can do some limited parsing of grammar but you essentially have to speak to it as you would a small child, and even then it makes more errors than a child of comparable vocabulary does because it does not process language like a human brain does. And it is incapable of synthesizing new language which is a key facet in developing a true grasp of nuance and idiom. Siri and Alexa are basically sophisticated pattern matching algorithms which are basically a search engine combined with a recorded-voice backed NLP library. (Many people apparently don’t realize it but the voice of Siri is a voice actress who worked for months to record the dictionary of words and phonemes that make it sound more natural than synthisized speech, the best of which still suffers from the aural equivalent of “Uncanny Valley” syndrome.)
I’ll talk to a computer when it can follow instructions without repetition and respond at a level of communication better than a five year old. Otherwise, it is isn’t worth the effort.
On a computer, I vastly prefer mouse and keyboard.
On my phone, I find it’s much easier and faster for some things to give Android voice commands like “Navigate to big bill’s BBQ” or “Wake me at 8:00 AM.”
I find it useful in very limited circumstances with my phone, like in the car, where it saves looking at the screen. I rarely use GPS but turn by turn directions are useful when I do. Voice activation for phone calls is useful even though I don’t make many calls from the car. “Ok Google” starting voice recognition is off by default. I’ll turn it on for long trips. Typical shorter local trips don’t involve any phone use. Even when they do there’s lot of lights that let me hit voice assistant without it on.
Turning off Cortana is part of maiming the Windows 10 phone home features out fo the box for me. I don’t have one of the home voice assistants. The utility for me just isn’t enough for even the below cost pricing that Amazon and Google were using to push them. That’s before we even get into my privacy concerns about pushing even more data to either of them than they already have. By default, none of my devices are listening to me.
I went with no, since that is the general case. I do sometimes put up with text to speech in limited circumstances, but I do not prefer it. And I just cannot find a purpose for those assistants.
My dad has found a couple uses: He uses Alexa once a day to check the weather before going out. I’m not sure what he uses for speech-to-text, which he uses instead of typing out texts when his hands are busy. He finds it useless for GPS, though that may be an app issue.
That said, not even he uses the always on version. It’s the one that you press a button to activate. None of us want a device that is always listening. And the issue with them being activated by videos or TV shows seems like something they should fix.
Get into a Subaru and see how many voice commands the Bluetooth can execute (correctly). Or try to reset your workstation password at my old workplace using the voice-recognition ID. For all the time and money dumped into it, the tech is still critically unreliable. I’m open to giving the internet interfaces (Siri, Alexa, the other chicks) a try every now and then, but not without the keyboard I will eventually need nearby. And no way do I trust it enough to drive my car, for instance.
The iPhone’s voice-to-text voicemail is really good. It even gets strings of numbers correct. The few errors it makes are understandable and sometimes funny. I had a voicemail that read, “Alaska next time I see you” which was supposed to be “I’ll ask ya next time I see ya”.
My voice-to-text made a funny recently as well. My grandson needed some sports equipment and my daughter wanted my advice before she bought. I texted her that I would meet them at Dick’s Sporting Goods. The VtT rendered it as “I’ll meet you at ****s Sporting Goods”. It thinks I’m dirty…
