Goodbye privacy, hello Alexa: Amazon Echo, the home robot who hears it all
theguardian.com
theguardian.com
The question is about our society, and government+corporate control. In a society where everyone accepts 5-10 different listening devices into their homes, a system is being built to listen to every thing we say and watch everything we do. How many different manufacturers built all the microphones and cameras that are in the same room as you and also connected to the internet? How many countries had their hands on those devices before they got to you?
The problem is with a big trend in society that we want to be listened to and watched, because it makes life more convenient on a personal level. Even HN users will happily let in multiple surveillance robots into their homes because they are comfortable with the personal decision.
But it's NOT a personal decision. When we have a society that so willingly accepts and even desires to live in a surveillance state, well, that's a problem. This is a direct path to a totalitarian surveillance state that would be much much more invasive than anything written about in 1984. We should be working to avoid that world, not invite it in as fast as we can open the door.
> When we have a society that so willingly accepts and even desires to live in a surveillance state, well, that's a problem. This is a direct path to a totalitarian surveillance state that would be much much more invasive than anything written about in 1984.
The cognitive dissonance in these two sentences is staggering.
Every time someone decides to trade some of their privacy for some convenience, they are taking away other people's privacy too, including those of us who do not share the view that it's worth it for the convenience.
Edit: I'm not allowed to answer any more questions right now, sorry to Pyxl101 and others. Hacker News tells me "you are submitting too fast, please slow down". Maybe I'll be able to come back later and answer your questions.
This is sort of a cheap shot but I am honestly curious, how do you reconcile this point of view with your goal of building an all seeing barometer?
It's a difficult line, of course. My goals are to reduce the effects of global climate change by better understanding the atmosphere. If we can provide more accurate wind speed information to wind farms, more accurate solar irradiation information to agriculture businesses, etc, then we can radically improve the world's preparedness for large scale global climate change.
For me, the issue here is deciding which problem is more important to work on. Mitigation of disastrous effects of climate change, or choose to not use any sensor data from smartphones due to my privacy concerns. So far I have aired on the side of climate change being a bigger problem, and that I have a helpful set of work that can make a difference, so I will do my work and hope that it helps.
I see that your company plans to "build[] an open platform for atmospheric data collection using smartphone sensors". It looks like you support Apple and advertise their products on your company's website: https://www.pressurenet.io/blog/iphone-6-has-barometer/
"We’re excited to start including iPhones in our study of the atmosphere!"
If you believe that "every purchase[] is a signal", then what signal are you sending about Apple iPhones?
I have not researched it in detail, but it seems like you're relying on data collection from large networks of smartphones in order to put together accurate weather forecasts. That's cool! But I find it hard to reconcile with the view that smartphone data collection (from people who volunteer to participate, or who use a product that has a collection feature) is invasive.
People in certain secure facilities or attorneys in many courtrooms weren't allowed to bring in phones with cameras because of respect for the privacy/security of the information and people within.
Now there are no phones without cameras and the insistence on privacy in many of these facilities has been relaxed to saying "please don't take pictures". Not because it's not important, but because of consumer decisions on the whole demanding cameras, to the point where they are in every viable option and it's not feasible to ban them. And privacy/security are eroded.
So when you make a consumer choice like this you're giving our society a tiny little push in that direction. You're pushing always-on always-recording voice monitoring software in the most intimate places of our lives to become the new norm.
If enough people put in listening devices that report to third parties in their homes, that can shift what is a reasonably expectation of privacy to where the government can put listening devices in homes without a warrant.
There was a good essay on this a while back by Alex Kozinski, the chief judge of the 9th Circuit Court of Appeals: http://www.stanfordlawreview.org/online/privacy-paradox/dead...
Spoiler warning: that contains a major spoiler for the terrific Isaac Asimov short story "The Dead Past".
The problem is there is no coherent programme on how to counter the privacy-encroachment trend. There is a bit of general dissatisfaction, some idealists urging us to repent and leave the clouds and a good amount of profiteers trying to benefit from that dissatisfaction. Meanwhile the forces behind this trend are organized and powerful.
The place to work on solving it (and, like all social issues, it won't be "solved," just "solved for now") is in the public and political spheres. We need to talk about this. We need to have the conversations that describe "what does privacy look like" and "how can we get the benefits from these gadgets while protecting our human rights".
Right now, no one's talking about a world in which you can have your cell phone listening to your conversations, while ensuring that private companies and government agencies can't make use of that willy-nilly. The place where those restrictions have to happen is in the laws we need to develop to keep them all in check.
We can't go back to a primitive, disconnected world, but right now, no one's leading us forward either.
IMO we need that, but we also can't rely on that. Things like the NSA have shown that the laws don't really matter. Yes, we should be striving for those laws, but we should also not depend on them, when they can so easily be ignored.
IMO, there sort of is: FOSS
I have my own "Alexa". It uses FOSS, and recognizes simple commands, like: lights off. Play/pause music. Add something to the grocery list.
Unfortunately, it's a lot less feature-full. I really want an Alexa (or the like), but I personally refuse to get one until I can run it on my own servers, and block it from getting on the internet. And unfortunately, that doesn't look like it's gong to happen anytime soon.
I think the big problem is the inability to run these systems without "big data" behind them, and I'm not sure what's the answer to that gap.
See also the http://www.voxforge.org effort to aggregate OSS acoustic models.
It depends on who you ask, some people may only be willing to run FOSS stuff |;)
But yeah, I've looked for a "offline: FOSS or not" solution, and haven't found much.
EDIT: I can't reply to you because thread is too deep I'm assuming, but: yeah, it looks like it does. Unfortunately I probably won't try this because it requires windows, and I'm not really wanting to set up a VM/Wine to accomplish this. But you are correct.
Unfortunately not a lot. That's why I said it was "sort of" and answer.
I've tried http://julius.osdn.jp/en_index.php and https://github.com/cmusphinx/pocketsphinx in the past, with very limited results. Though it's been a while and they may have gotten better.
True, if you are a geek, you can use open-source software to cobble together some resemblance of popular cloud services but under your own control. There are even detailed guides on how to do it (see also: https://github.com/sovereign/sovereign). But that's a lot of nuisance and in the end what you get is an inferior version.
Howso? I don't see it as any different, but instead of Microsoft, there are multiple organizations.
> FOSS is from another era. I'm not sure what you mean by this. FOSS is still current, and quite active today. In fact, I'd say even moreso.
> But that's a lot of nuisance and in the end what you get is an inferior version.
I agree. That's what I do now, but it's not nearly as good. But what I was implying was that we have the potential to create a much better version, that's trustworthy.
Is it though? To be honest, this article is pretty light on substance. I read through it all because the guy is a good writer, but the only revelation here is him realizing the microphone device he put in his kitchen is... gasp, listening to him.
If it turned out Amazon was actually recording and storing everything, that would be quite the revelation. If it turned out your data wasn't being deleted when Amazon said it was, that would be quite the revelation. Otherwise, there's just not a lot here other than a semi-humorous story about accidentally insulting a robot.
With respect to both, it's impossible to achieve it 100%. As the amount of applications running and the nuisance to which they run increase, it gets trivial to break either.
Developers have long agreed that security is an arms race you really can't win. It's easier to break into than it is to secure. What mostly stands in the way of security breaches is that it's generally frowned on to do and, failing that, is illegal.
The same might need to be true (or truer) for privacy. Privacy is harder to make people understand, though, given the muddiness of the area.
Privacy breaches will never be reined in, therefore. We need to deal with the new reality of ultimate transparency. This could be okay as long as everyone has the same access to that information.
Private information access inequality is the issue.
https://en.wikipedia.org/wiki/History_of_violence_against_LG...
It wasn't very long ago that the private lives of the LGBTQQ community risked physical violence.
Even today, people risk physical violence for their personal opinions: http://www.wnd.com/2015/11/black-lives-matter-critic-fired-f...
Privacy is the only tool that protects in a society where free speech is not respected. So no, you probably don't have any secrets worth protecting, but others do. It's these victims that matter- and you are turning a blind eye to the defenseless when you are saying your privacy isn't important.
That said, I believe that most HN are very privacy-aware. Indeed one of the reasons Echo wasn't a huge success is privacy invasion.
If you read the latest threads regarding Paris Attacks, you'll see that most HN users are ready to give up security for freedom (if there ever was such a trade-off, I'm not even sure anymore).
Most modern phones passively listen to your voice too, and people carry those with them everywhere. They are present in the most intimate situations, rather than (e.g.) just in your living room. So I don't think there's a reason for people to feel differently about Echo than about Cortana on Xbox, or "Ok Google", or any other voice activated interface.
Perhaps not evil but they are amoral.
If a course of action is not illegal and would increase revenue then a company really should be doing it. Particularly if their competitors are doing so.
That's not an objective fact; it's kind of condescending to present it that way. I would say that they corp's are amoral.
>they are not part of a bigger scheme that wants to enslave the world.
Corp's act in the self-interest of the directors without regard for whether their actions increase/decrease "world enslavement".
"We want Kickstarter to be similarly in sync with society. Earlier this year we became a Public Benefit Corporation. This means we are legally obligated to consider the impact of our decisions on society, not just our shareholders. Though we are still a for-profit company, as a PBC it’s very different from the expectation that for profit companies maximize shareholder value above all. It acknowledges and embraces that you are a part of a larger community."
I don't particularly care for the first one, I do very much for the second. These are separate concerns. If we can't have data-driven gadgets because the state may get that data, we're solving the wrong problem: we need to get rid of the state.
I used to agree, but don't we know for certain that this is false now? With the Snowden documents, we know that the USA government, along with others, knows how valuable these private companies' data stores are and will hack their way in so that they can get access (if they're refused when they first ask politely).
I no longer think there's much real difference at all between the private companies collecting the data and then actively/passively sending it to the government, vs. the government doing all the collecting and analyzing themselves.
And the point that I was trying to make is that even if the private companies can collect and analyze the data safely (they can't), that still sets up a society that is willing to be watched. That is the hardest part for totalitarian governments to overcome; if you hand the USA or China or Russian governments a silver platter of a willing population that will accept the surveillance from one entity, they will accept it from another. That's a dangerous place to be.
Not under the current rules where the data are treated as "business records" and subject not only to spying while en route, but also to administrative subpoena and all sorts of other "legal" and questionably legal collection methods employed by various divisions of the state.
>we need to get rid of the state.
For sure, the part that wants to collect all the data needs to go.
the people chose to carry smartphones with them all the time, everywhere. that decision was given up without a fight, in fact it was welcomed as the best thing that has happened in the technology revolution to date.
not only has the developed world already made this decision, the developing world continues to make this decision on a day to day basis, and their choice is clear; devices on everyone, everywhere.
My critique of this argument is that it has no bounds. Totalitarian surveillance states are very possible without modern (2000+) electronic devices and have existed for a good half century. To add to that, there are many other technologies that are essential to the kind of centralized bureaucracy that totalitarian states generally use. Without electronic communication, it would be extremely difficult to disseminate policy. Without electronic storage, it would be very hard to collate descriptions of suspects and centralize information. The parts of the Catholic church of the middle ages would very much have liked to run a totalitarian state - but it was beyond the technology of the time.
So, if you think we should oppose technology that allows states to survival and terrorize their citizens, there is a longer list to go after: police, vehicles, computers, electricity, etc. All of these technologies gave rise to fears that they would restrict freedom and those fears were ultimately realized. In many ways, we are far less free than people who lived in societies without electricity or cars or any of those things. However, we are also safer from many groups that used to use their freedom to take advantage of others (though not all, obviously).
I agree that this kind of electronic surveillance (or, more specifically, the ability for machines to collect and collate information) is transformative. It's similar to other technologies that changed things - the telegram, electricity, plumbing, etc. It will alter the balance of power and we, as citizens, should think about how we want this technology used and not used. However, I find the idea that this technology _in particular_ is any more or less totalitarian than ones that have come before it to be silly on its face.
Be that as it may, the FBI isn't above reaching back to the old ways. They run 15,000 paid informants nationwide, backed by the information dredged up by pervasive surveillance. Parallel construction is just the beginning of what that cyberspace/meatspace combination is capable of.
With the NSA and pervasive surveillance, it is possible to build an eternal, indestructible static system, governed by an elite, and with the appearance of representation.
Tell that to people who lived in eastern block countries.
I agree that this technology, like all communication technology, makes it easier for governments to repress their people. However, I don't think there's any reason to believe this is the straw that finally subjugates all mankind under tyrannical government or that this technology will be any different than previous advances. As I said before, this is only one of many technologies that are necessary to run the theoretical nightmare scenario you've outlined.
The main difference, from my point of view, is that that other technologies are older and the doomsday scenarios predicted with the rise of telephones, computers and other advances have not come to pass. At the same time, we've had plenty of time to appreciate the advantages of older technology and balance its use in society.
For the recent Paris terror attacks, at least some of the planning was done over text messages. The text messaging system is a better fit for the kind of dystopic observation you're describing: the infrastructure is more centralized, there is lots of legal precedent for observation, there are weak safeguards and no fancy voice recognition is necessary. Of course, the only way we know about the text messages is that they found a physical phone - because despite 20 years of commercial availability - such monitoring is not in place for text messages.
I would argue that it isn't convenience that attracts people, its narcissism. Society has been surrendering their privacy for years with Facebook. We are comfortable with it. We have seen a direct connection between sharing more (location, pictures, videos, etc) and getting more (food, taxi rides, sex)
So of course the next frontier is not having to share anything at all. The next frontier is listening and deciding. Its concierge services that bring you an array of options, specifically tailored to you. You are the center of their world, so as long as you keep making purchases and supplying them with information to better serve you.
Amazon and Google have both figured out that the current commodity of value isn't oil, and isn't energy in general, it is information. The people with more information have an advantage over the people with less information. That advantage can be tactical, commercial, or personal. So collecting information while being helpful gives you a much better asset than simply being helpful does.
We've read stories of cops misusing information databases to check up on their ex-spouses or future dates, Google SREs reading people's mail, and NSA analysts passing around "juicy" intercepts with prurient but no security value. We've got people on the other side of these piles of information and those people are fallible, and susceptible to bribes, and generally susceptible to poor judgement.
What is worse, is that the "service" that something like Alexa offers is pretty credibly offered in a modest cluster of modern machines with GPUs. I would expect that a 32 node Jetson cluster could provide reliable general speech recognition and execution. But the illusion is that you need a "big cloud" of machines to do this. You don't.
So I'm all in favor of these interesting technologies helping folks out, but I really wish we would step past 'cloud' and into a 'house cluster' which, like your furnace or HVAC system, does the work locally and protects your privacy. It is a company I have seriously considered starting twice now. Even in the US people can get multi-megabit connectivity to the Internet from their home, there is no reason any more why you shouldn't be able to access your pictures, your media, your email, and communities by talking to your server in your house. The challenge is turning it into an appliance. That is doable I believe.
This is the model, and it largely already exists.
If you're willing to give up the most candy and fluffy of high tech "features" (the ones that break all the time and are of questionable value anyway) you can have a very nice home control system that has no network connectivity and zero wireless emanations.
It's also more tasteful. LCD screens in the living area are the new formica.
Do you ask people with iphones or androids to turn them off when you are close because its listening for "hey siri" or "ok google" ?
If you asked me to turn mine off when visiting, I would happily oblige you- And then label you as a nutcase who has lined their walls with tinfoil.
That's an interesting response. So because someone does not like to have unknown third parties listen to everything that is said in a room they're right away tinfoil hatters?
(in case you skipped that the first time)
I'm surprised at how trusting people are when it comes to tech like this.
Regular, definitely-not-tinfoil-hat types that I know have little pieces of tape over the cameras in their laptops.
Edit: corrected link (was https://news.ycombinator.com/item?id=10605775)
The link was supposed to point to another comment in the same thread. I have corrected it.
Are you sure? Suppose, hypothetically, I was looking to circumvent this.
It would be extremely easy. I could monitor for key phrases (e.g. bomb, terror, NSA, freedom.) and then simply wait until the next "alexa" to send that recording to Amazon.
My personal issue with this is: That could be changed at anytime without any warning. Yes, I tend towards the tinfoil side of things, but the fact is: You can't really always know that, unless you are consistently monitoring the connection.
That is a nice way to say that your argument is unconvincing in that it depends on the ML classifier to recognize the 'wake' phrase reliably, and I have ample evidence that it has a number of false positives. And when it "wakes up" it is trying to figure out what you're asking it to do, and if you're not then you are perhaps in danger of it doing something you didn't intend. I eventually turned this feature off on my phone entirely, I could not make it reliable.
And I expect to hear interesting stories of that time when someone's phone called or did something unexpected, because the technology makes it inevitable.
That's the first step (we're trying to launch a product) and ultimately the goal is enabling customers to completely sever the Plum cloud dependency. Right now if you want 100% of the features (schedules, scenes, provisioning, what-not), you have to connect them to the internet, but we want users to have all of the features without any need for Plum's servers. As it stands, about 40% of the features do not require Plum's cloud and actually don't even use it if it's available, so we're working in that direction.
We've also talked about open sourcing a subset of the cloud infrastructure so that users could prop-up their own servers if they really wanted "remote control" of lights (which does matter to people).
Privacy sensitive IoT devices are certainly possible but until more people care about their privacy, it will unfortunately be very niche and a bit of an after-thought to go the extra mile to provide services that can be decoupled from the "smart devices" in the home. Particularly so, considering there's a land grab happening in the IoT space right now and corporations are trying to dominate the home IoT platform.
Humans, in a capitalist society, pretty reliably follow "ask for forgiveness, not permission". Which is interestingly at odds with the more fringe ethical philosophies of free software, privacy respecting software, and generally "ethical" software (that obviously needs explication but I won't go into it here).
Nest-like machine learning (the thinking on which predates Nest) can later be added in a cloudless way that starts to anticipate what lights to turn on together, what actions should trigger what scenes, etc.
Presumably Plum has similar ideas.
Home lighting which is activated by a motion sensor but doesn't blind you or awaken others but helps prevent you from stumbling through the home in the middle of the night.
Lighting which could be coordinated with media controls.
Simply more economical way to move or add an additional light switch than hiring an electrician to do it the old fashioned way.
As someone who runs his own home-automation server: Have you thought about something like a stand-alone server, that I'm able to turn off net-access for (open-source or not)? I'm currently looking for something like that, and I'm not aware of anything that exists.
https://youtube.com/tl880linux has demo videos.
But if you needed any access to the lights from outside the home, you would. In which case you can either colocate one or prop one up on AWS to serve as the remote command bus. Plum provides that right now but I want to, at some point, enable folks to do it themselves.
Or if you really want to, run the server in-house and port forward. That could work too.
Interesting. The pricepoint is a little high, but I didn't know that they were full linux-clusters. I run my own home-automation server in-house, so I could use this. I see that the android-client is in the works. Is that still the case?
I was taken aback briefly when I first considered it. I realized it must be recording the environment constantly, in order to scan for the code word. Upon hearing the code word, it wakes up and streams the last few seconds of audio remotely, where the real linguistic analysis happens with Siri/Cortana/Google Voice Search. [1]
The difference is my phone is with me virtually 24/7. It's rarely not in the same room. In my opinion, the cat was out of the bag a while ago.
I make a decision to trust the device manufacturers, who claim their devices only listen for the code word. As a scientist and engineer, I know what that means: it's recording everything, all the time, and probably running a simple linguistic analysis sufficient to recognize codewords only. Then on waking up, it streams that audio centrally, which is how it can respond to queries out of the blue like: "Hey Siri, what's the temperature?" Answer: "It's currently 41 degrees Fahrenheit". It must be recording and processing everything, though in a computationally inexpensive way, for the sake of battery life.
[1] Google Chrome apparently supported this on the desktop for a while too: http://thenextweb.com/google/2013/11/26/google-brings-ok-goo...
This is why we're "lucky" to have limited battery tech right now. What's to fear is when a minuscule device will get years and years of energy with new battery tech. This is where ubiquitous connected devices will ramp up and pose the real privacy threat.
I'm not so sure. Passive background network activity happens perpetually on Android and iOS devices now. I'm not saying it doesn't have an impact on battery life, I'm just saying that sending chunks of speech during this time wouldn't have significantly more impact than Gmail syncing in the background.
Related, my 'Hey Siri' is only active when the iPhone is plugged in.
I'm much more concerned about devices that are wired in permanently with power and network.
edit typo
With wired devices, I've comforted myself with the analysis that users could detect subterfuge by looking for unexpected network traffic. A device can't record my voice and transmit it centrally all the time without generating noticeable network usage. It could, however, record for a while and send in a burst later, perhaps concealed within other traffic like a weekly software update. Nevertheless, I think a close examination could detect foul play. I do wonder whether phones could conduct the same type of subterfuge:
Perhaps there ought to be a "phone surveillance challenge" that's kind of like the Underhanded C contest. Each year, contestants submit a phone app that runs in the background on a phone over a 48 hour contest period simulating normal usage. At several points during that period, an important conversation will be overheard. The app that does the best job recording the conversations and remaining undetected is the winner.
Naively recording all audio and storing it will probably drain the battery. It also needs to be compressed, though bit rates as low as 5 Kbps seem to be intelligible. If we record at 5 Kbps, then it will require only 108 megabytes to store 48 hours worth of audio. The compression might eat up CPU, however. Perhaps skilled contestants will find a way to do it cheaply.
If we're a smart contestant, we don't indiscriminately record and store all audio. Instead, we want our contest entry to record the human conversations. I'd hazard a guess that our highly skilled contestants can come up with cheap passive methods of detecting human voice, comparable to the ones that listen for "Ok Google". So our contest entry only wakes up when it hears human speech.
Our contest entry won't transmit anything over the airwaves regularly, since that uses battery. However, we know the phone will be charged at least once in 48 hours. We'll wait until it's charging to upload our recordings. Even better, we'll time our transmission burst to be right after some other network activity, like a daily software update. Or if the phone is plugged into a computer, we can compromise it as well and extract data that way.
Sounds like a fun contest.
We can emulate entire operating systems in a web browser now.
You can fit an entire routable, navigable map of the entire world on a mobile phone or usb stick. Every single named place and every single path or road between them on the planet on a phone, offline.
Why are we still thinking that we need the cloud to process speech? Why not do all the listening, all the private bits, locally, and optionally connect to specific services (Music, Shopping, hotels etc) as and when.
Why not?
EDIT: Oh, it looks like they have Pocket-sphinx and Julius now for internet-less systems. Interesting....
You also could not have control over the device and its user post purchase if the essential part of of the product is not under your direct control.
These are in my opinion two major forces that motivate the appearance of "cloud".
I might just direct a profanity-laced rant about surveillance at it, on the slim chance that there's actual humans listening on the other side.
...or not, because that could even be enough to arouse suspicion in the world today. This reminds me of a scene in a movie long ago...
But of course, there's also the concern that adversaries will intercept surveillance data.
Create a system that realizes the tangible value of privacy and people will be much less likely to give it away.
What does that look like? That's the billion dollar question. For starters, here's an essay by Albert Wagner of USV that sets the stage for a solution (in a very abstract way): http://continuations.com/post/131372549150/land-capital-atte...
The conscious trade-off in my family is that voice control is useful, natural, and fun versus a somewhat low probability that amazon/google/apple are recording everything we say. The latter would be as catastrophic for those companies as the VW emissions scandal. Which of course happened, so it's just a low probability, not zero. Companies do dumb things and break the law all the time. The answer is to make sure that sensible laws, vigorously enforced, are in place.
I don't see a huge difference between an Echo and the camera/microphone in my laptop - they both could be listening all the time, and there's a certain level of trust that smarter people than me have checked and found nothing. Can anyone wearing a tin-foil hat in this thread point to credible evidence otherwise? That would be enough to make me change my mind, but I doubt there's anything I could say that would change yours.
The smoke alarm and doorbells are exceptions (and I guess there are several other good exceptions: clothes washer, clothes dryer, dishwasher, coffee pot) but I think I'd prefer a purpose-built listener with a high pass or notch filter.
The Echo may have some unintentional funny failure modes, such as when a doorbell or other trigger sound plays on a TV commercial.
The Echo has a low power chip-mode which is able to do "basic" voice recognition and wake the device when the correct phase is spoken.
So the chips itself is interpreting everything you say looking for the correct wake-up phase, but the wrong ones aren't recorded, and most of the device's logic in in idle/sleep/hibernation.
Now, that being said, when the device is awoken anything you ask it IS transmitted back to Amazon HQ. For example "ok Alexa, what is the weather today in New York," the "ok Alexa" won't be transmitted (that just wakes the devices) but the "what is the weather in New York" is transmitted.
Most of these articles focus on the Amazon Echo "always listening" and while that is technically true, there is no known privacy implications to it (since it isn't recorded/transmitted).
I think the conversation needs to start with the assertion that "this is my data, and I'm letting Amazon use it".
The main topics I am surprised:
1. Why do I need a dedicated device if any phone can do this too?
2. I think Google is capable of doing way more than Amazon AI-wise since they have more NLP experience and more data about us and the world
3. This is another time that Bezos launches a questionable product which will probably fail (Amazon Fire), the one success in the hardware space of Amazon is the Kindle though
Yikes. IMO, the presence of something futuristic on popular TV does not establish it as something to naturally progress to. I look to (good) sci-fi for ideas and thought experiments and, often, cautionary tales.
I don't have a big problem with the privacy issue here, because Amazon's ecosystem is opt-in. But I hope that ethicist is more critical-thinking than the snippet in the interview paints him to be.
Your refrigerator does not need networking capability. Nor does you wash machine or your toaster or your tig welder.
In fact, I would argue that even your display device (your "smart tv") should not have a network connection, but I think I'm in the minority there. For those that are interested, I have very good luck using "commercial displays" like the ones you see in airport signage. They are dumb. They are also fantastic and well engineered displays.
Although you are correct, there's a lot of advantages that can be had with these devices being on a network.
I know this answer isn't for everyone, but I personally have a home-network dedicated to devices, and it's air-gapped from the internet. I don't really see any other solution to this (other than FOSS-everything), but I would love one.
Problem is that there exists people who try to convince us that we need a middleman between our networked devices and us - that we need to connect our devices not into our home network but with their servers.
The reality is that we do not need middlemen, the middlemen needs us such that they could steal the data, repackage it and sell it.
I will give one example that I have already given before.
Suppose you have an IoT device in your bedroom that uploads CO2 values (of course it may also do something useful too such us showing this data to you or regulating the ventilation).
Now the rate of CO2 production depends on bodily activities. More activity means more CO2 production. I suppose that you agree that having this information you more or less are capable to deduce when a couple has sex.
This means that we could calculate a metric how much sex is going on and relate it to specific user.
Suppose that we detect that rate of sex has gone down by 20 (a made up number dependant of our model). We can now sell this information.
This information is for example useful perhaps for psychologist who want to sell you counselling.
But it is even more interesting for divorce lawyers who now could prey on couples having period of difficulty in their sex lives.
Even more, this information could be made more valuable by some additional influence. For example feeding the user with articles related to marital happiness and sex life.
I leave the rest for you imagination.
I think that such devices are very usable, but the data should not leave our home.
1) I'm not a sadist. 2) Lowering the barrier of obtaining answers leads to a higher level of curiosity and greater yearning to learn. 3) It allows an easy way for my kids to be exposed to more types of music. 4) It helps with piano practice.
And seriously, lay off the personal attacks. Echo isn't a replacement for a parent. This isn't about sitting a kid in front of a TV, this is about a wonderful supplemental product which kids can use at an early age.
Edit: @tl, I'm rate limited, so I'll respond here. Hopefully you'll see it.
I'd probably ask them why they bought it and if I could try it out, or, if they could show me how they use it. It's an interesting interface, I'm just not sure how generally useful it is compared to more traditional interfaces (keyboard, mouse, touchscreen). I view it as a novelty "Trojan horse" of a sort. I'm sure Amazon is collecting specific metrics on users of their device and most users probably get some novelty from it before they unplug it and forget about it.