Amazon's range of smart speakers and their artificial intelligence assistant Alexa have proved to be a huge sales hit.
"We have thousands of engineers inside Amazon adding to [its] capability every day and then another tens of thousands of developers adding to the skills," he tells the BBC.
"The thing I am sure of is that this time next year she will be significantly more intelligent than she is now, and that sometime in the future we will hit our goal of reinventing the Star Trek computer."
It's a lofty goal, especially since any attempt to go beyond commanding a weather update or asking for the lights to be switched on is currently asking for trouble.
Try to have anything close to a normal conversation with Alexa and it tells you it doesn't understand or cannot help.
But though it may not always be obvious, the firm says rapid progress is being made.
Which will prevail, Alexa, Cortana, or Siri?
(Score: 5, Informative) by Virindi on Sunday July 30 2017, @12:41PM (16 children)
There is a lot of hype around these sort of "assistant" devices. However, the results are still comically poor and rely entirely on a legion of developers considering every possible thing you might ask of the device.
One of my family members recently outfitted their whole house with Echo devices. On a recent visit, I witnessed one of them ask Alexa about 20 times to do something, with suitably comical fail results. People are happy to overlook the fact that it is still crap because they have bought into the marketing telling them it is awesome. That is unlikely to last.
Star Trek computer it ain't; it is only incrementally better than the voice control of old. Except now it records everything for permanent storage and analysis by advertisers, spies, insurance companies, etc, etc, etc.
(Score: 4, Interesting) by fyngyrz on Sunday July 30 2017, @02:59PM (12 children)
That's correct. "AI", it is not.
It's a big increment, though. The ability to correctly translate what is said is extremely good. What's missing is the next step: understanding general human speech so that further progress on the request can be made. Alexa, thus far, has demonstrated very close to nothing along these lines. It is, just as you say, dependent upon looking up the translated speech among canned strings provided by developers. That will have to change before this tech moves much further along.
No, really, it doesn't. I (and many others) have carefully watched the network traffic from the device. Until it hears (or thinks it hears) the trigger word (presently, user's preset choice of Alexa, Computer, Echo, or Amazon), it doesn't send along what it hears. It does connect to check for updates to its software, and so the potential is there for this to change without warning, but so far, it hasn't. Frankly, I think it unlikely, as that would result in serious negative publicity for Amazon. As opposed to the low-level rumor-mongering that goes on now.
On the other hand, the cellphone most of us carry can both hear what you say and send it along to both the network and the phone company; so it's a little late to worry about hardware capabilities. It's all about what the device actually does, not what the paranoid think it might be doing.
As for the usability at present, there are things the Echo does well, and things it doesn't. At $50 a pop for the Echo dot, given what it can do now, I find it well worth the candle. We use various Echo models around the home to control lights and similar non-critical household devices, play music, tell us the temperature / weather, time, set alarms, timers and reminders, and order / re-order things from Amazon. All of these things it does very well, and in doing so is both hugely convenient and frankly, fun. Our home is a large open plan, and the light switches can be fifty feet away. That's the kind of thing that can make usability of devices like these very high.
The most important step will be when these devices can recognize speech locally. Then a WAN connection won't be required for many useful operations; and that's when worries about security will settle down.
Devices like Mycroft [mycroft.ai] coming from the open source community will keep the pressure on the commercial devices like the Echo.
(Score: 0) by Anonymous Coward on Sunday July 30 2017, @03:17PM (3 children)
The thing I worry about is the ability of Amazon to perform a software change on the device. You say that you "carefully watched" the network traffic and found that data was only transmitted when a trigger word occurred. That may be true right now, on your particular device. However, at any time, Amazon can perform a very quick and very simple software update on your device which could either change/add the trigger word(s) or simply have the device send all audio data through. I'm sure the TOS grants them permission to do so. When Bezos - or any other employee of Amazon - publicly state that the device doesn't or won't spy on you, it is merely a fluff statement: the TOS still says they can.
(Score: 2) by fyngyrz on Sunday July 30 2017, @03:28PM (2 children)
Yes, I specifically called out that possibility. I also pointed out that the cellphones many of us carry can also be compromised in the same manner (except more thoroughly, as they not only have voice acquisition, but also multiple cameras, email account access, location tracking, and so on.)
There are many potential threats. I choose to worry about the ones that are actually threats now. The Echo isn't.
(Score: 0, Touché) by Anonymous Coward on Sunday July 30 2017, @03:37PM (1 child)
So you surround yourself with a bunch of proprietary technology that doesn't respect your freedoms and become dependent upon it to some degree, and then just wait until you're abused (abuse besides denying you your software freedoms, which is a form of abuse in and of itself) before maybe doing something about it? That seems very short-sighted. Before you know it, you'll have created a proprietary fortress that could turn against you at any moment in ways that are difficult to spot.
(Score: 2) by fyngyrz on Monday July 31 2017, @05:04AM
I don't subscribe to this viewpoint at all. Actually, I think it's kind of hilarious.
My "software freedom" is embodied in the software I write. Which I have been doing for well over 40 years. I'm perfectly happy to write what I need, or to buy someone else's solutions if and when they're selling things I want. YMMV, and that's fine. I'm not feeling the least discommoded by Amazon's business model WRT the Echo ATM. If they change it in the future, I'll give it further consideration. Right now, it's fine.
See, this is just the kind of hyperbolic nonsense that gives me that feeling of amusement. Fortress, eh? You know how strong that "fortress" is? I reach out, pull the power cords, and throw these things in the trash. End of "fortress." And that, BTW, is what it likely to keep Amazon in line, if they aren't inclined to otherwise: if they abuse their customers, their customers can just walk away.
In any case, there's no threat, and there's no ability for anyone to force me to use the tech if it became a threat (of which there has been no actual sign, unfounded speculation such as yours notwithstanding.)
(Score: 0, Insightful) by Anonymous Coward on Sunday July 30 2017, @03:40PM (2 children)
You cannot dismiss this as mere paranoia, especially not at this point in time. The NSA is conducting mass surveillance on the populace and countless government agencies (besides them and including them) are looking for any and all ways to break any security anyone might have. It is foolish in the extreme to not take the future into account when selecting what technologies you choose to surround yourself with. For this reason, I do not have a cellphone, and will not have one.
(Score: 0) by Anonymous Coward on Sunday July 30 2017, @03:58PM
They're talking about me behind my back, aren't they?
(Score: 2) by fyngyrz on Monday July 31 2017, @04:55AM
Amazon is not the NSA, and the NSA isn't likely to be collecting information on demand out of the Echo inasmuch as it doesn't send anything along unless you use the wake word.
So yes, it's paranoia. Pure and simple.
It's worth keeping an eye on (and I do – my network is monitored and logged, I know what's going on within it, and specifically so WRT the various Echos) but until information starts flowing out of here without my say-so and explicit agreement, there's no call to disadvantage myself in the interim.
Oh, please. If the Echos get out of hand, they get unplugged and trashed. Problem solved. Same for any other such device, up to and including my cellphone. These are here to serve me, not the other way around – the very second I decide they're not doing that, they're history. You're asserting I should inconvenience myself right now because of something you think might happen in the future. That makes no actual sense. It is, in your own words, foolish. There's no present threat, and if, and it's a big if, the issue should actually arise, it is a matter of no more than a few seconds before it is 100% taken care of.
(Score: 3, Interesting) by sjames on Sunday July 30 2017, @07:22PM (2 children)
That's funny you should mention it. It looks like Google has gotten the shaft again. The Google assistant on my Android phone understands my speech just fine when I take mobile data and WiFi both offline. It's at least as useful as Siri. But for some reason, nobody even gives it a thought.
(Score: 2) by fyngyrz on Monday July 31 2017, @04:44AM (1 child)
That's very interesting. I wonder if Google Home can do this as well.
I don't give it a thought – have not up till now, anyway – because Google doesn't offer what the Echo does. Amazon's connection to its product and media offerings is a heck of a come-on, not to mention the large ecosystem of Echo-friendly devices.
Also - while Amazon has never directly screwed me personally, Google has. So they're not on the top of my "hey, let's try that" list.
(Score: 2) by sjames on Monday July 31 2017, @07:06AM
Google has it's drawbacks and limitations, for sure. But it does indicate where the state of the art is now.
I find it surprising the feature goes practically unmentioned in Android advertising. Especially when every other commercial was for Siri.
(Score: 1) by Virindi on Sunday July 30 2017, @10:27PM (1 child)
I was talking about the commands you send it. Which provide them a lot of useful information!
Now, you can say it is the same as using a spyware'd app, or an OS with piles of builtin "telemetry", or a search engine that does everything it can to track you, but many of us avoid those things as well.
(Score: 2) by fyngyrz on Monday July 31 2017, @04:40AM
Honestly, I am utterly unconcerned about Amazon and partners knowing what songs I listen to, streams I watch, and when I turn my devices on and off, or when I ask for time and weather.
Useful, in this case, means I get better music suggestions, playlists, viewing suggestions, less spam in my product suggestions, etc. It's exactly, IMHO, what that information is both good for and should be used for.
Should it become an issue, these devices can be trashed even faster than they were invited into the LAN. But it's not an issue. At all. It is something to be aware of as a possibility, and I am. However, short of that possibility being realized – and there is zero sign of that – I'm unwilling to cut off my nose to spite my face. If folks want to not have convenience now because they think that someday in the future Amazon may go too far, that's fine too. Everyone has choices. And consequences. :)
(Score: 2) by arslan on Monday July 31 2017, @03:37AM (2 children)
True. However that legion of developers is democratized. Anyone can create "apps" (called skills) for Alexa. If the rise of the app store is any good indicator, the next person that makes a bucket of gold from creating the next killer app/skill will herald another app building frenzy.
Of course like the app store, it'll end up with millions of shit apps but that is a problem for the owner to decide what "skill" their Alexa learns.
(Score: 2) by GreatAuntAnesthesia on Monday July 31 2017, @10:42AM (1 child)
Oh boy, I can't wait. "Alexa, play a fart sound."
(Score: 2) by arslan on Tuesday August 01 2017, @10:52PM
Oh boy, I can't wait. "Alexa, play a fart sound."
https://www.amazon.com/Bloatware-4AFart/dp/B019NU7L5O [amazon.com]
3.5/5 stars!
(Score: 1) by redbear762 on Sunday July 30 2017, @02:08PM (1 child)
I won't have one since a recent article made it obvious that these devices are intrusive - http://fox61.com/2017/07/11/alexa-automatically-calls-911-after-amazon-echo-overhears-domestic-violence-attack/. [fox61.com]
(Score: 3, Informative) by fyngyrz on Sunday July 30 2017, @02:37PM
Yeah, except the article was wrong [cnn.com] - the echo did not make any such call, which requires connection to / communications with the telephone system.
(Score: 0) by Anonymous Coward on Sunday July 30 2017, @04:30PM (1 child)
not the devices, but the girls and women
Alexa is a real name. Real humans have it.
(Score: 2) by arslan on Tuesday August 01 2017, @10:54PM
... except changing the name of this Alexa is a lot easier than going to your local magistrate (or whatever equivalent in your country).
(Score: 2, Interesting) by Anonymous Coward on Sunday July 30 2017, @05:27PM
Google is the one major tech company who is not, in any overtly direct fashion, commercializing their AI development. I mean obviously their search, image recognition, and countless other things do utilize AI - but they're not trying to wrap up some facade of intelligence and market it as such along the lines of Alexa, Siri, or Cortana.
This is incredibly conspicuous given Google's incredibly active pursuit of R&D in AI. However, when you look at their [patents](https://www.google.com/patents/US8126832) I think things begin to clear up. Google has a large number of secretive projects in development, and it's quite obvious that their goal is not to compete - but to 'win.' In a way like they did with Google search. They were so far ahead of the competition technologically, that any notion of competition was really destroyed. And I think it's the same thing with AI. Companies like Amazon are trying to simulate artificial intelligence by throwing thousands of engineers at the problem and hoping what's produced, almost as a sort of semi-automated Mechanical Turk [wikipedia.org] (another Amazon product). By contrast, I think Google's goal is to give small teams all the resources in the world and work on genuinely developing artificial intelligence capable of acting with a degree of understanding beyond word association. And I suspect they are also likely to give it a shell that is actually capable of 'physicalizing' its understanding and capabilities. I think that their goal is not to compete, but to revolutionize the market. And, for better or for worse, I think they will succeed.
(Score: 4, Insightful) by requerdanos on Sunday July 30 2017, @09:57PM
Amazon's Dave says:
These are separate issues, examined below.
Soylent's whomever says:
Google not a contender? Neither something like Mycroft [mycroft.ai]? Seriously? I know with the sidebar polls, one can't complain about the limited number of choices... But suggesting that one of either Alexa/Amazon, Cortana/Microsoft, or Siri/Apple will prevail in voice recognition assistant technology given huge competition isn't even something to take seriously.
That said, let's go back to Amazon's claims of increased intelligence and reaching star-trek level.
Increased intelligence is certainly possible; I say this with things in mind like...
- IBM's Watson's performance [ted.com] on the Jeopardy game show
- IBM's Deep Blue [wikipedia.org] beating Kasparov in Chess (20 years ago)
- AlphaGo beating human masters [scientificamerican.com] at the game of Go
Experts in the AI fields continue daily to make progress and advance the arts of machine learning and machine intelligence.
None of these players, certainly not Amazon, will reach the goal of parity with the "Star Trek Computer" standard. The simple reason for this is that that standard is the opposite and antithesis of their goal.
Permit me to tell a story. Many years ago, I was excited with X-10 automation. This isn't the pop-up ad X-10 phase you may be thinking of, but the much older X-10 devices sold at local Radio Shack and similar stores in the 70s and 80s. Going from manual everything to automated control was pretty exciting. But X-10 wasn't the real source of the excitement; it was the potential it suggested. X-10 has some problems (commands take a long time to transmit; there's no way to query a device to find out whether it is turned on; there is no way to know if a command succeeded or failed) but I was excited about where home automation was clearly going.
It didn't go there. X-10 happens within your house. You send commands to, or with, your own controller or your own computer, and your own controller/computer sends commands to X-10 modules to make them turn on or off, dim or brighten, etc.
"The computer" in Star Trek receives voice commands and monitors systems and data in a single craft and does the work right there onboard.
There is a real parallel here that Hey Google/Siri/Alexa/Mycroft/Cortana are completely outside of.
What disappointed me about the "advancements" in home automation was that they left your home: Imagine if "the computer" in Star Trek listened, encoded what it heard in super-latamatrix crystal form or whatever, and sent that stuff back to an approved planet of the Federation for processing before anything could happen! Total no-go. Yet that is Dave's goal. Not making your own computer, starship, or device intelligent so it can do what you want (and make no mistake, that should be our goal), but making systems and institutions by which the wishes and desires of the galaxy are submitted to him and his counterparts to be filled or not, just as they see fit.
So when he says "we will hit our goal of reinventing the Star Trek computer" you should see huge red flags and do well to remember that that isn't his goal *at* *all*. I don't say he's evil--he's probably not evil at all (or only a tiny bit). I don't say the work he's doing isn't important--it's tremendously important. But at the same time, accepting (even benevolent) corporate monopoly on voice-recognition is a bad, bad idea.
I remember 15 or 20 years ago setting up a computer that ran Dragon Naturally Speaking to interpret commands, parsed them in a program I wrote in (I think) Visual Basic, and then acted on them.
"Computer: A 4 off please." -- click --.
Very satisfying, and no institutional nor corporate gatekeeper involved. That way lies madness and danger, sure, but more importantly, the experts and appropriate programmers to do the job aren't even working on voice recognition as they probably should be because culturally, we by and large accept that only a few certain companies should be doing it and we should send them all our data to be voice-recognized. This is harmful to society, and will continue to be so until it is corrected. Machine learning and intelligence keep advancing, but software to do voice recognition isn't evolving nearly so quickly--and the technology is clearly there, because a few certain companies are doing it all the time.
I don't know how we got here from such promising beginnings, but I do know that the solution is not to participate in the alexa-siri-hey google-cortana-mycroft system of lining up and asking them to pretty-please do for us what, for example, Dragon Naturally Speaking was doing 20 years ago on individual computers in individual homes and offices.
This seems to have developed into a rant, which I will now terminate. Peace.