For many products, you want to do at least some research before ordering. For products you order regularly, reordering is usually only a couple of clicks/taps.
I think the goal from Amazon's point of view is to be the middle man in as many transactions as possible and to abstract away the payment from the acquisition of the product or service. The more you can abstract the payment, the more money people will spend. Like with credit cards versus cash:
http://www.nytimes.com/2016/03/27/your-money/credit-cards-en...
https://www.psychologytoday.com/blog/ulterior-motives/201001...
https://www.psychologytoday.com/blog/retail-therapy/201306/w...
I WANT the check out process, I WANT to verify my address, payment method, I want to confirm my order qty, shipping costs, and other items.
I am baffled by this trend to simply speak into the air and have a computer order anything with out limit based on what the computer thinks I said.
By removing all friction and letting people order something exactly when they think of it, I suspect they will make many more sales.
I'd expect a tiny reduction in difficulty of ordering something (especially while making it harder to be sure what something you're ordering) wouldn't help much. What I would expect to matter, is how easy the entire process is compared to at other stores.
Unless they're just counting on preying on poorly-thought-out impulse buys?
Other than that is pretty much novelty like you said.
(Both deliver products similar to peapod)
If I could use voice to add to my list and/or order that wouldn't happen.
how many doll houses is a person willing to return in order to make their soap purchase easier?
It's also a great way to practice your speaking ability. Really trains your enunciation.
Amazon needs to develop a biz model for skills that makes sense though.
It's definitely meant more as a replacement for a shopping list than it is a way to, say, do your Christmas shopping.
Is that a realistic scenario for the majority though?
Aren't most people going to either make a mental note and add it to the shopping list later, or just wash/wipe hands and add a note to the list on the fridge blotter/whiteboard etc.
It'll be a sad reflection on the evolution of cognitive thinking if the people believe that they have to drop everything, clean up and reorder cornflour the instant they run out.
> grabbing your phone, loading the app, searching for the product, and hitting "buy".
Considering that, as well as the other option of having a magical listening tube in the house, living in a small English village, I have the 'luxury' of grabbing my coat and walking about 400M to the local store, and apart from the minor health benefit, I could also stretch the operation into a dog walking session, bump into and talk to some friends on the way and get some fresh air. There's also the local pub (http://the-quaffer.blogspot.co.uk/2012/02/murrell-arms-barnh...). Alexa, pull me a pint of London Pride...
Damn, we really are living in the future the 1950s imagined we would.
For repeat orders, for small orders, for ordering a taxi, for ordering a repair man, for ordering specific things("i want the same earphones as my friend"), a voice chat may be useful. People don't want to research everything.
Or what if Amazon build a voice personality that people really like, and by extention, really enjoy getting help from when buying stuff ?
Furthermore, it's a bet. It wasn't so clear, at the advent of mobile, that so much ecommerce will happen there, instead of the big screen. but it did. People changed their habits. It's also not clear with voice, but it's hard to tell.
I'd probably use it for common household stuff if I didn't just buy that at the store. But I know a lot of people who buy cleaning supplies, cooking supplies etc on Amazon.
I don't think it's a blunder to say (almost) any sentence ... we do still have freedom of speech. To think that these viewers would expect that no one ever utters the phrase again is ridiculous. Next they' expect there to be legislation prohibiting certain words on any broadcast. An as more voice controlled devices are created, there are bound to be conflicts.
If they'd written this article correctly, they'd have an auto-playing audio clip that says "Alexa, order me a sandwich". A child named "Ok Google" will be the next version of "Little Bobby Tables".
"This is not the first time an ill-conceived TV spot has caused havoc with voice-control systems."
There's nothing ill conceived about the TV spot.
No, the phrase itself is not a problem, but when advertising your own product you should know better.
This is what's unethical. If users don't understand how they may be charged, they ought not to be charged.
My previous android phone didn't do this. The side effect is it doesn't catch all my different inflections of "OK Google". When I first said it, I said it in a super dull, boring voice. Now that's the only way I can trigger it. However, I haven't looked into training it further.
I also don't know how big a problem any of this is. Sure, it could be an issue but the fact that there are news stories about individual cases implies it's not rampant.
I never remember to start off being angry at my phone, so it always takes me two tries.
If something has harmful or annoying consequences, especially to people you don't want to piss off, doing that thing by accident really is a blunder no matter how much legal right you may have to do it.
(Also known as "freedom's just another word for nothing left to lose".)
You blame the person who built the thing to have harmful and annoying consequences. The one who made the blunder here is Amazon.
One side note is that I hope this doesn't become the "include no matter what" feature we've seen with IoT, etc. Imagine voice activation in an airliner cockpit - a terrorist walks to the forward lavatory and yells "dive and disable all controls". (yes ... it's very implausible but I predict there's at least one case where someone doesn't think through all the ramifications of their voice interface).
https://en.wikipedia.org/wiki/Speaker_recognition is hardly a new thing. Further, if Amazon get's 200,000 orders for a doll house in the same second should be able to detect that as a mistake.
Sorry no, in the US the 1st amendment protects you from government persecution, in other nations other laws protect your freedom of speech from government persecution
However this idea that "freedom of speech" is only about government is dangerous to the idea of freedom of speech. While protecting against government censorship is very important, protecting against other forms of censorship are also important.
Now let's say this is a biopic that happens to be playing on your TV? All true events, and if you enabled purchasing directly from your Echo without safeguards you'll get an attempt to purchase a dollhouse. None of this is fraud, none of this is intentional. These are all shortcomings with where we're at.
Now if I ran an advertisement that said "Alexa, order me a SuperafunTime My Brand Dollhouse"- I should be slapped with whatever the law allows.
At most there's some form of tenuous indirect profit, but seeing as there's a backlash I doubt doing this was their intent.
https://github.com/alexa/alexa-avs-sample-app https://github.com/Kitt-AI/snowboy
It allows the AI to hyper-specialize to recognizing that particular name, increasing reliability, and
use generic voice recognition, with the associated costs and privacy issues, only after activated.
Not 100% sure, but I think that activation recognition is even local. They only send your voice to mothership after activation.
Sensory includes a chart showing the CPU requirements for wake word activation and accuracy. [1]
The main reason custom names aren't allowed is accuracy and training time. Even with a well-trained model like Sensory's using 90 MIPS the wakeword isn't recognized 10.2% of the time and has false acceptance. Using a set wakeword could also allow for a FPGA or custom chip to have the wakeword model hard-coded for a much lower power hit for mobile applications. One of the other wakeword solutions, Kitt-AI Snowboy [2] allows for personal hotwords, but it takes some effort to make a personal training set and this isn't nearly as robust as what Amazon can throw at it.
[0] https://github.com/alexa/alexa-avs-sample-app [1] https://github.com/Sensory/alexa-rpi [2] https://github.com/Kitt-AI/snowboy
They blame (and punish) the kid that "explodes" and not the bullies taunting him/her for months. They blame the whistleblower. They blame the reporter and the symptom. They blame the reaction and not the cause. They overwhelmingly target the immediate thing/person that causes a disruption, and what is most surprising to me - even if the actual cause is apparent and not hard to find!!
But, the news story was about Alexa having issues. Which both makes it hilarious and ironic, and means that the news anchor should have had reason to have possible issues in mind at the time.
In their current state, voice assistants appear to be wide open for abuse. You think auto-play ads on sites are bad now? Wait until they start auto-ordering for you too.
Or maybe even stop their turn by turn navigation... So then they have to try and drive while restarting it. Maybe they don't even remember the address, so they have to open the app and go to recents or check their email(Not Sure if Siri handles that part too).
I wonder if they'd be prosecuted under the Computer Fraud and Abuse Act if they did that same thing here in America. Hacking people's phones from the radio it feels to me.
The answer is, depending on country.
In Germany, all children of any ages can buy anything with previous approval by the parents. From age 7 on products costing less than the monthly pocket money cna be bought with approval after buying, from age 14 on all products.
So, as a result, you might have a 4 year old child ordering something online, and have to honor the order (just as 4yos frequently are sent to fetch bread from the neighbourhood bakery).
Obviously, cultural values affect this heavily — while in Germany and Japan this is normal, in many areas of the US people below 16 are not expected to act on their own, or buy things, so it can be obviously different.
"Hey Siri" does a pretty decent job at having you quickly train your voice and only answer when it's you say the wake word.
Hell, it should also pay its customers for the hassle of finding themselves drafted into being beta testers.
Last night my wife and I spent about half an hour talking and SIRI activated twice--neither of us said anything to her and she doesn't even understand my wife's accented English very well. Casual voice recognition simply isn't up to the reliability to do automatic orders yet.
The risk of unintentional purchases seems much too high, and one doesn't lose much convenience with a quick confirmation.
You can also already restrict purchases. Mine only adds to the shopping cart, for example.
http://www.forbes.com/sites/briancaulfield/2012/01/12/hed-wh...
Edit: It is worth mentioning that Alexa being outside the TV does make a difference, as it's unlikely to have a directional microphone as would likely be installed in the TV.
Or, how about "Ok Google, how do I assassinate the president?"
I doubt that Amazon will release their voice recognition models and parameters though...
Simply talking to or about them in your living room can now be a problem ranging from mild annoyance to credit card purchase.
For example, could you list a uniquely named item on Amazon (perhaps as a Marketplace seller) and charge high restocking fees? Then instead of just trolling people for a laugh, your business model would basically be collecting restocking fees.
Not sure that was the ill-conceived bit.
The really useful next step will be voice independent language recognition with voice dependent command recognition. That and accent independent language recognition. That is one of the, if not the, next billion dollar acquisition by one of the big players.
by the way, how does Alexa knows that you said Alexa without sending anything to the cloud if it needs the cloud to decode any speech?
