Amazon’s plan for Alexa
technologyreview.com
technologyreview.com
The last time I interacted with one was at a friend's place. We were talking over coffee and somehow Brad Pitt's age came up. Neither of us knew how old he was. My friend paused, and then yelled out across the room "OK Google, how old is Brad Pitt?" The answer came back almost immediately. "103 years old".
I personally fear privacy-related issues stemming from random people I may encounter online far more than I fear privacy-related issues related to NSA dragnet snooping or Google and Amazon monopolies. I have issues with them, and I'm aware of the risks, but in general I don't care much in a personal sense. I do care in a larger societal sense (these are all slippery slopes that are harmful to society), and I care about other people caring and believe their privacy wishes should be respected, but that doesn't change my personal behavior.
India is the world's fastest growing consumer of voice traffic. At this point 30% of all google searches in India are voice and growing.
The fact that we have over 20 mainstream languages contributes to this. a lot of people now write Indian vernacular in Latin script which is highly unstandardised in spelling.
Voice is far far superior.
in fact there is a cool Alexa-for-the-poor that works on phone calls in India (through a toll free number) powered by Google Assistant : https://www.indiatoday.in/technology/news/story/google-assis...
Voice is the most powerful UX that people in countries like India are using.
Billions of Chinese use voice as a main input for phones.
[1] In spite of my concerns, somewhat hypocritically, I have a mini in my bathroom which I received as a gift. I find it quite useful for listening to podcasts/music/audiobooks and switching between them purely by voice.
[2] From both personal usage, and usage at my friends' places, I find Google Homes as accurate as standard Google search, which is to say, very accurate the vast majority of the time.
Edit: this made me invest time into figuring out the exact command to navigate so thanks :)
What is amazing is how Amazon made an ecosystem around it to hook people to it - my parents were given a free years worth of amazon prime along with their broadband connection but they did not use it much. Then somebody gifted them a firestick and suddenly the prime subscription made sense because of prime videos. Now, they have an echo show 5 which is justifying the prime subscription even more because of prime music.
Indeed!
I have Alexa's and one Google home at home.
We now have cheap and good enough multiroom audio, voice controlled. It's used daily
We use very regular the weather forecast and timer and control our lights with it as well.
The future is here but it will probably go more in direction of offline assistance.
The integration in Philips hue, with more av receiver etc. Will just become even better.
I must be officially old, because I see no reason to give voice commands to things. If I did I'd just attach a microphone to the server I run at home and connect it to things.
It took me a fair bit of trial and error to work out the exact magic words to say to get google maps to launch navigation to a given address by car without requiring me to use the touch screen at all.
My natural inclination was to say "Hey Google, give me directions to 123 fake street" but this would bring up a route (or if you're especially unlucky a list of different places for you to choose from) but require you to press a "start" button with your finger, and often also require you to switch the navigation type from bus/walk/cycle to car manually too.
"Give me directions to 123 fake street by car" would skip the navigation type selection, but still require you to manually press a "start" button (various variations of "ok google, begin" or "ok google, start" etc didn't seem to do anything")
After a long period (months) of having to press a tiny button manually while driving to start navigation I finally discovered that if you say "Navigate to" rather than "Give me directions to" it would skip past the start button screen and jump straight to giving you directions.
So now it works great, but I hope I don't have to go through tedious trial and error with everything voice activated to work out the exact phrasing it wants.
All of these commands printed out on a list somewhere would probably be consistent, and make sense. But you don't get to see that list. Google and others want to present these systems as magical AI helpers that you just converse with, as if they had any more intelligence than a touch tone menu system circa 1990.
Knowing how these things work, it just strikes me as hubris. The UX isn't good because they can't admit that the whole system isn't as good as they say it is or want it to be, and this makes them opaque and frustrating to interact with.
What we have nowadays is more like a voice-input command line, with all of the stiltedness and attention to detail when crafting your input that requires.
Unfortunately people want to pretend that natural language parsing can be solved, which, at least, for English is impossible: https://en.wikipedia.org/wiki/Time_flies_like_an_arrow;_frui...
Do AI assistants actually make anyone's life easier? Serious question. I understand the novelty of shouting at your lightbulbs to turn them on and off, but it feels like that remains the epitome of the life-improvement that these systems can provide. I've never gotten a good recommendation from software outside of Pandora, which last I checked doesn't even use AI. I've never found a real use for Siri or Google Assistant. I briefly considered getting some Phillips Hues to go with an Alexa that I got for free, before stuffing the latter in a closet.
Please, for real, weigh in if voice assistants/AI suggestions have meaningfully improved your life. I'm genuinely curious.
The alternate (much more banal) explanation for the current prevalence is 1) a cynical data-grab on the part of companies, and 2) the devices are cheap and novel for consumers. In other words, it's not that they're creating such great value, but that companies have learned to embed their data-vacuums in what amount to toys.
I don’t subscribe to the just a data grab theory, that is incredibly cynical. Also, even a simple device is useful if it’s cheap enough (“Alexa play space unicorn” is really useful for us parents).
Still, that's about all of our progress in 50 years (in production ready, deployed systems). Estimating that we need another fifty years seems fair.
We have to account for accelerating progress, but I'm not sure that's enough to bring us much closer than 50 years away
It is handy when cooking, you can set named timers, ask for conversions with your hands full.
It is also fun at parties, guests can play whatever songs they want (there is a discount because I have the Amazon music just on that one device).
I listen to the radio on it, and I also get my news from it (I have it programmed to play the NPR news update and then the Fox News update. They usually cover the same stuff, but it is interesting to see how they emphasize different things.)
Alexa seems smarter than Siri or the "hey google" character if you have random questions and don't want to get out your phone or you already have too many tabs open.
I've played around with some of the skills, but really only use the timers, alarms, news, weather, music, streaming radio, and sometimes ask it questions.
Most new consumer technologies that end up in every home fundamentally change people's lives. I just don't see that happening here.
We talk about these companies destroying privacy, gaining power on the scale of nation-states, etc. And we put up with them on some things, because smartphones and social media are really useful and can be really important to living modern life. But I just don't get it with the assistants (and really, Amazon shopping too). We're giving up so much as a society for such a tiny modicum of extra convenience.
At the end of the day its just a fancy clock radio with okay sound that can play anything I want to hear without me getting up. But the "assistant" part of it seems like it is smarter than Siri or hey google.
Playing music, getting directions, and sending texts while driving the car.
Controlling music in my house. Its a lot faster to just ask for something than fiddling with my phone. The kids use it for this too (they also do: "Ok google, tell me a story").
Setting timers in the kitchen while I'm cooking. We have some physical timers but inevitably they're hidden somewhere and/or my hands are dirty when I need them.
Asking random questions: "Ok google, what city has the longest name?", "how many cups is 9 tablespoons?" etc. It can answer a shocking fraction of the questions I throw at it.
A smartphone is an complex always on device with its own internet connection, microphone, GPS and camera.
We are not trading low risk UI with high risk UI. We switch or extend it to nodevice/voice control.
And I personally find it very impressive and subtile.
It just makes sense to use it in your kitchen, bedroom or on the Couch.
It is the same with Philips hue. Do I need it? No. Is it worth to me? Every penny. Why? Because it's a completely new level. Now you can combine lamps which had not one switch like your ceiling lights with a standlamp in the corner and you put your ligjtswitches where you actually are using them.
That was never possible before.and it's great for creating lightsettings. I'm using different lightsettings constantly.
For example our light is much less bright when I go to the toilet at night and it is motion detection based which I really love for our floor.
You'll need to define that if you want a proper response. I suspect you don't really want a proper response and are just looking for a way to shit on people who have bought these devices.
Btw like much gadgets these things are for convenience so an example is when cooking and your hands are covered in food instead of setting a timer with your hands you can do it with your voice. Of course you can always just look at a clock and do it in your head manually. Is this a "meaningful" improvement in my life ... That's up for debated. I'd say the same thing though about mobile phones.
> Prasad’s ultimate vision is to make Alexa available and useful for everyone. Even in developing countries, he imagines cheaper versions that people can access on their smartphones. “To me we are on a journey of shifting the cognitive load on routine tasks,” he says. “I want Alexa to be a productivity enhancer … to be truly ubiquitous so that it works for everyone.”
My problem is that tech companies uttering those visions usually lie. Even in this case, there's set of business and technological decisions being made that go against the vision. As long as voice processing is done in the cloud vs. the device, as long as the device itself is tied to an account and a service (vs. just subscribing to services that provide actions which require those services), as long as it's done by amoral corporation with plenty of history of putting profit before humane behavior - Alexa will always come with privacy issues, will not be useful to everyone, and there will be business incentives to not make it a true productivity enhancer.
Should I ask if I visit, and how would you respond?
I treat the Echo the same way. I don't mind if friends have their own preferences when it comes to "smart" things, but I won't indulge their paranoia in my house.
These things have been out for years now, and I've yet to see any evidence that they're always listening or exfiltrating your conversations to HQ. Just vague anecdotes about people seeing ads after they discussed some product, or how their Alexa made a VoIP call by mistake (which, going back to the cell phone analogy, is the same as a butt-dial.)
If you got some particularly sensitive information to talk about, then let's just make sure she's muted for that conversation. I'll do that while you lock your phone to prevent the butt-dial.
https://www.independent.co.uk/life-style/gadgets-and-tech/ne...
https://www.forbes.com/sites/kateoflahertyuk/2019/04/12/amaz...
https://www.washingtonpost.com/gdpr-consent/?destination=%2f...
That said I agree its your home, and me as a potential visitor, will not expect you to disable/enable any devices on my account
Heck you get it even if you don’t buy it. My headphones were software updated with Google Assistant, plus there was the infamous Nest smoke alarm.
And real estate developers have been putting up apartment buildings for at least three years with Alexa built in to the apartments. It's only a matter of time before these "features" are in everyone's homes whether they want them or not.
I think the article misses the point here. Alexa is not supposed to summon you anywhere, it's supposed to be omnipresent like the "computer" in Star Trek. Being more proactive would just be similar to receiving smartphone push notifications. Maybe the real worry is about getting too many of those notifications from Alexa, which is obviously something Amazon needs to be careful with.
I also couldn't find any mention of Alexa being "among the world's greatest data collectors" in the referred infographic. I don't doubt that it collects a lot of data in terms of terabytes since it's all audio, but is there really so much information content in those commands people occasionally utter to it?
You forget that those devices are always on, always listening.
Some related links: https://www.iot-tests.org/2017/06/careless-whisper-does-amaz... https://arxiv.org/abs/1707.08696
At the same time I think people have become more privacy-aware in general and started to demand less bulk data collection. So they might accept certain risks of getting hacked but not necessarily submit to terms of service that allow bulk collection.
If it misheard you it can do bad things: https://www.theguardian.com/technology/2018/may/24/amazon-al...
When you make the argument that “no one said the keyword so it isn’t uploading!”, you aren’t considering that it’s listening for voices, your voice, your SO’s voice, how many unique voices, what time they are talking, what temperature it is, if there is a TV or music on, what station it’s on via pattern recognition, all manner of meta data that DOES NOT require a keyword or a large outgoing data stream.
It could store meta data for days and dump it with a single MQTT packet which it’s sending all the time, and they’re all encrypted.
We flat out have no idea what these things are doing on the data collection end that doesn’t have to do with someone saying “Alexa” first.
On the other hand, I still think people can analyze the hardware and software running on the devices. If there are "black boxes" doing something unknown, we should be worried, particularly if they are always powered on and have the processing capability to analyze audio content.
Still, the original claim in the article was that there is a lot of this metadata being uploaded right now. I don't think anybody has verified that claim, even if it could be true.
I mean sure Alexa skills are just Amazon letting other people figure out how to design the interface for interacting with Alexa until they lock that down. It also allows for wonderful ads for local news letting you know they have programmed an Alexa skill.
This is article is just description that as with every other large digital assistant team, it is much easier for them to learn how to design the user experience of the digital assistant by seeing what people do with it rather than designing literally anything.
I also tend to think that it is now becoming clear that the hard part of making a digital assistant is designed the user interface and interactions and not the AI part of it, not to discount the vast amount of work done by all of these companies.
I've often thought about what I would like a virtual assistant to do for me, and it's generally along the lines of "alert me if there's something I should know about", similar to how a real, live assistant might go about it. I don't want to know the weather parameters, only if they changed significantly from yesterday. And delivered at the correct moment for me to do something about it, not too early, not too late. Same for a commute, but for that, the assistant would have to know how I commute normally, what my schedule is for today, when I'm planning to leave. All of this would be a lot more accurate with omnipresence, as scary as that sounds. The value add is great enough that in the long run, implementation is almost guaranteed.
I don’t expect I’ll order items via Alexa anytime soon, but “she” gets probably 50 interactions a day. I’m fine with the trade off.
In capitalist society, if growth stalls, it leads to riots. The only way a capitalist society can present a face where some of the population feels like they are being taken care of is when the amount of wealth is increasing generally enough that the inequality isn't as sharply felt. Advertising, credit, and other behavioral nudges are accelerants of growth.
Of course, when things break down enough, Alexa will also be able to act as an arm of the state for gathering intelligence. I look forward with horror to the day when it is installed by default into housing.
I know this has nothing to do with the overall theme of the article, but this is misleading, Microsoft is the only company in that list with a market cap over 1 trillion. As of the time of writing it is ~1.1T. Apple, on the other hand, has a market share of ~1.16T, and they have made clear their stance (or lack thereof) in users' personal data. Just don't like the author's use of market caps to explain the power of "data" when their argument falls apart rather quickly.
I predict Amazon will make another attempt at releasing a phone within the next two years
If you want to check your phone, you can stand up and take the 60 seconds to do it.
This situation raises a whole host of issues, and I'm not claiming it'd be a good thing. It just seems to me like the obvious endgame for Amazon, and I think it perfectly fits the modern ethos of reducing friction and pursuing convenience at all costs, including losing our own agency.
Maybe I'm missing something, but what does the Cambridge Analytica scandal have to do with AI?
> Rohit Prasad, the scientist in charge of Alexa‘s development, recently gave MIT Technology Review’s Karen Hao one of the most terrifying interviews in modern journalism
Is the hyperbole really necessary?
> Rohit Prasad, Alexa’s head scientist, has now revealed further details about where Alexa is headed next. The crux of the plan is for the voice assistant to move from passive to proactive interactions.
So Alexa will make suggestions or remind you that you need to order toilet paper. Much like Google's Android greets me with the time it would take me to get to work every morning without any prompt.
Its really sad that the state of tech reporting is full of this hyperbole and fear mongering. You can have a real discussion about the dangerous of technology and erosion of privacy, but this isn't it.
Using a microphone, even with an offline ML model, to do things like play music at the right times and (almost certainly) show you ads or ask if you want to buy Amazon product is pretty off-putting. Google Maps and Siri Suggestions uses your phone's behavioral data and location history to figure out when you usually leave for work, but if you said "i need to get to work an hour earlier tomorrow" to someone, it would be weird if your phone or Alexa instantly said "do you want me to remind you to leave earlier?" I don't think it's as off-putting as the article makes it out to be, but it's still pretty creepy; we're almost certainly going to have boomer facebook groups boycotting Alexa for "always listening".
Then again I'm not likely to by a "smart"-speaker anytime soon.
I also missed all the bad things that happened because of Cambridge Analytica. Even if we take their extremely dubious claims about their effectiveness at face value them running amok resulted in politicians being better able to target ads. That's not exactly the stuff of nightmares.
Based on what he was caught saying on camera, in addition to his deputies, and the reporting around their political campaigns in Africa using propaganda that claimed AIDS would come back if so-and-so were elected, it's not a stretch to imagine those people doing everything the can, legally and illegally, to sway an election.
... never underestimate the impact of “just optimizing” something
Sure, they may not be able to deliver super effective results today, few startups in their early days do.
What they’ve identified, though, is that there is a vast market of incredible wealth willing to pay for services like this. CA may have shut down, but Nix turned around and founded another company that does the exact same thing[0].
Keep in mind that these guys also have access to incredible investment money, they don’t have to go out and raise seed rounds, series rounds, etc. These companies are being built in secret, and it’s incredibly short-sighted to write them off for their capabilities today, because if they succeed tomorrow, we’re in really big trouble.
0: https://www.newsweek.com/what-emerdata-scl-group-executives-...
Indeed. There are plenty of companies that would pay a LOT of money to use that data in any meaningful way. It doesn't happen, but people will convince themselves it's important data, enabling a host of dying properties to fool a handful of market actors into getting a profitable exit. The amount of user data leveraged from most acquisitions is basically 0, excepting a few mailings. Conversely, consumer reach (active users) matters a great deal.