Statistical Analysis shows Echos process voice to serve ads
arxiv.org
arxiv.org
That’s a difference at least for my interpretation how “evil” company xyz is.
> In contrast, skills can only rely on persona’s email address, if allowed permission, IP address, if skills con- tact non-Amazon web services, and Amazon’s cookies, if Amazon collaborates with the skills, as unique identifiers to reach to personas. Though we allow skills to access email address, we do not log in to any online services (except for Amazon), thus skills cannot use email addresses to target personalized ads. Skills that contact non-Amazon web services and skills that collaborate with Amazon can still target ads to users. However, we note that only a handful (9) of skills contact few (12) advertising and tracking services (Table 1 and Figure 2), which cannot lead to mass targeting. Similarly, we note that none of the skills re-target ads to personas (Section 5.3), which implies that Amazon might not be engaging in data sharing partnerships with skills. Despite these observations, we still cannot rule out skills involvement in targeting of personalized ads.
While I don’t like the sharing of such data for ads, it’s a far cry from Alexa processing voice in the background with zero interaction.
The discussion in the paper is nuanced on that point and does not make that claim as far as I read it. Section 2.2 (page 2):
> The content of users’ speech can reveal sensitive information (e.g., private conversations) and the voice signals can be processed to infer potentially sensitive information about the user (e.g., age, gender, health [82]). Amazon aims to limit some of these privacy issues through its platform design choices [4]. Specifically, to avoid snooping on sensitive conversations, *voice input is only recorded when a user utters the wake word*, e.g., Alexa. Further, only processed transcriptions of voice input (not the audio data) is shared with third party skills, instead of the raw audio [32]. However, despite these design choices, prior research has also shown that smart speakers often misactivate and unintentionally record con- versations [59]. In fact, there have been several real-world instances where smart speakers recorded user conversations, without users ever uttering the wake word [63].
For me, it's this: "Your Echos are Heard"
So, the opening salvo. That's what gives me the impression they imply voice assistances are listening to everything.
I don't refer to voice commands or normal interaction as "echos" so the user of the word "echos" here implies something nefarious. Sure, it's the name of the product, but for me, it reads like something more.
Saying “process voice for ads” has subtle connotations in the current landscape of privacy discussions.
There is an argument that this is more privacy conscience than other ad platforms. One needs to say the word "Alexa" before Amazon will collect any potential targeting data. There is an active and distinct choice that must be made before every interaction. That isn't true for Google and Facebook. They will collect data in the background while you are doing other things. There is much less transparency in when and how they are collecting their targeting data and therefore we have much less agency in the process.
> We find that Amazon processes voice data to infer user interests and uses it to serve targeted ads on-platform
Amazon has publicly stated that it does not use voice data for targeted advertisements [83], [75].
https://www.nbcmiami.com/news/local/are-smart-speakers-plant...
They're using skill interaction, order history, listening history, etc to target ads.
This adorable little infographic on "the journey of a voice request" conveniently leaves out that it gets used for advertisement[2]. They have also made public statements that outright state that voice data doesn't get used for ad-targeting[3]
[1] https://www.amazon.com/Alexa-Privacy-Hub/b?ie=UTF8&node=1914...
[2] https://www.amazon.com/b/?node=23608618011
[3] In a statement, Amazon said the company took “privacy seriously” and did “not use customers’ voice recordings for targeted advertising.” https://www.nytimes.com/2018/03/31/business/media/amazon-goo...
“not use customers’ voice recordings for targeted advertising.”
I guess it depends on how one reads that quote. A trusting sort could read that to mean, we don't use anything we learn from voice recordings for targeted advertising.
A skeptic might read that quote and determine:
Well we generate metadata from the recording, and we then use the metedata for targeted advertising, but we don't use the actual recording for advertising.
Which makes sense, if I was to implement something like this, I wouldn't use the actual recording, I'd process the recording(which I have to do anyway to answer the request) and if I happen to save some useful for advertising data along the way, well, more $$'s for me!
Which one is true? I guess it mostly depends on how hungry Amazon is to make a buck and what they think they can get away with. As a privacy snob, I'd prefer the trusting version to be true.
The paper does talk about “voice data” which I think is a bit misleading. “Voice data” to me would imply an analysis of the sounds your voice makes directly for ad targeting purposes but it’s clear enough from the context what they actually mean.
Amazon does plenty of rubish things already, no need for us to make up extra things!
1. "Smart speaker vendors or third-parties may infer users’ sensitive physical (e.g., age, health) and psychological (e.g., mood, confidence) traits from their voice."
2. "The set of questions and commands issued to a smart speaker can reveal sensitive information about users’ states of mind, interests, and concerns."
They also mention that "smart speaker platforms host malicious third-party apps", and "record users’ private conversations without their knowledge", but that's mentioned as examples of prior research and thus seems to serve more as background than something this paper is trying to prove.
Point 2 is the one you're focusing on, and yeah, that's not surprising. You'd expect Amazon to build a profile on you based on the stuff you ask Echo to do (though the ethics of this certainly warrants discussion).
Point 1 would be the surprising thing, that smart speakers infer information about people from their voice, rather than from the commands themselves.
Their methodology seems to be to create multiple personas and compare the sorts of ads they get. In order to prove that information is inferred from traits of the voice rather than the words in the commands, they would need two personas which are identical in which commands they send but with different voices (female vs male voice, healthy vs smoker voice, something like that). From skimming section 3, it doesn't seem like they did that, so I'm forced to agree that the thing they prove in this paper (if their statistical methods are valid) is that Amazon builds an advertisement profile based on your interests as expressed in terms of which commands you're sending the device.
Imo the truth is much scarier: advertises know us better than we known ourselves. There's also the idea that we are so alike, advertises can generalize to great success.
I think you mean confirmation bias. Observer bias is similar, but relates to being non-blinded during a scientific study.
Baader-Meinhof phenomenon or Frequency illusion could be what we were seeking?
Maybe (just an hypothesis) there could be a set of other keywords also listened for, that once detected could start another more complex routine. Like that, you could limit the battery impact, and yet be able to listen to the users for advertising purpose.
Then again, I opened Instagram at a carwash once, and a few days later I got ads about car treatment products. I walked by an e-bike store the other day and stopped for about 15 seconds to look at the bikes being displayed, and a few days later Instagram started showing me ads for the brand. I thought Instagram or another Zuck-app was tracking my location, but I just checked and none of the Zuck-apps on my phone have location permission enabled.
Market theory says people will vote with their wallet, which is why I boycott all of them. But I'm under no delusion that my behavior will change anything. Voting with your wallet only works if someone is willing to offer what you want.
You are implying that all alternatives have to be "buying a smart home assistant". There is another alternative: Don't buy one. If all existing alternatives are bad, then not buying any is also a way to "vote by wallet". Then there will be a big market segment of not-having-bought-yets that can get tapped into by just coming with a privacy-preserving offering.
But I share your sentiment nonetheless, I have zero desire to become part of the group of persons that got used to voice control and would therefore miss it if it wasn't available (as long as I don't turn blind I guess)
For the blind however, I guess it's a huge win that ad-driven voice control gadgets exist and are available to them. I'm sure their options would be basically nonexistent if they were the entire market.
Siri Data and your requests are not used to build a marketing profile, and are never sold to anyone."
https://www.apple.com/legal/privacy/data/en/ask-siri-dictati...
It's about the interactions with Echos, and not as some other comments imply, about listening to general conversations passively. Although, the experiment setup might be used to study this in a future experiment.
- Gather data for ads
- Corner the "smart assistant" market... so you can gather more data for ads
I find it hard to believe that listening in on random conversations, collecting the audio stream, processing it, data mining it, accounting for substantial noise - is worth the effort for any of these companies.
Maybe for a smaller startup. When you're already a $1T company - this just doesn't seem like a good or valuable data source. Not to mention, it is an OBVIOUS gigantic hit to your trust.
- We can build Echos/etc. cheaply and make a profit
- We're working on speech recognition/AI anyway, so we have the data centers
- We can subsidize this work by selling ad data
On the other hand, maybe when you're that big you've already covered the low-hanging fruits (and so did your competitors) and now need even more?
It's so much easier to grow Amazon .1% than to discover a new $100M business.
“Sell my product to anyone that mentions fridge” Yanno?
But don't get me wrong, it is totally plausible that some eng team somewhere would still build it because it's a shiny new tech, it just doesn't seem like the kind of thing that has any real world value.
https://www.oracle.com/a/ocom/docs/corporate/acquisitions/gr...
In my understanding, showing ads for what people that are connected to you searched/bought/interacted, as in a simple network analysis, would be much cheaper and would give you very similar results.
It's not claiming Alexa is listening in while inactive. It's claiming that it's inferring user characteristics (age, health, etc) from the audio of voice commands, instead of just the post-transcription text.
Only that's not where it stops.
After the customer leaves the store one of you minions follows him to every other store he ever enters to record what he buys, or even looks at
That's why your analogy falls completely flat.
Nobody would complain if it only concerns your store and your sales. But people violently dislike you snooping on everything that your customer does.
Worse! He doesn't even have to be a customer of yours. You snoop anyway, even into his most private affairs.
Worse! You also share this information with credit bureaus, meaning the data is available to potential employers. Thus limiting your job options.
Looked at a hookah once because you were curious? Sorry, no pot-heads in my company!
In my store analogy, I’m trying to make discourse on this topic less black and white. To recognize that data collection is not some inherent evil. And sure, to point out that a store that tries to peep in your bedroom window should be called to task.
That's not really the argument your opponent has.
It's that pervasive data collection, even if it's 'only' outside your house, is quite bad.
> I’m trying to make discourse on this topic less black and white.
Then you need to rethink your framing, because right now you're trying to split it into "store tracks transactions by itself and uses that data itself" and "tries to peep in your bedroom window".
Everyone else is talking about shades of gray too. But they're pointing out that 90% of the store-tracking shades have significant negative consequences.
Those industries have essentially built a worldwide spying system that would make the NSA jealous without them even having to pay a dime and capitalism guarantees those systems will keep being maintained forever.
The need for consent would come if the assistant were to record the entire conversion you have with the customer, and send it to Facebook or Amazon in exchange for money.
You haven't read enough dystopic fiction.
But the main point with my comment was to re-frame what was advocated, from their fantasy of "one small store owner doing the best for their customers" into "a surveillance system where every corporation tracks your every move for profit".
Or the civil rights movement?
Or the labor movements?
A bit more than a decade ago, in a meeting at Lab126 where we were discussing a then secret future product with voice recognition -- in particular "keyword spotting", so that it could wake up on command -- I said something like: "I doubt this will work, to get high enough accuracy we'd have to stream data constantly for analysis, and consumers would have to expect this. No one would buy it."
I've since learned never to underestimate the naivete of others.
On several occasions I've seen ads appear based upon things being discussed, but never searched. Always with the realization I was near a smart device with a microphone (or an app with permissions)
I think in hindsight this will be similar to the realization that systems like XKeyscore and PRISM were not only technically possible but already deployed.
If i am not mistaken - this paper seems to look at explicit "Smart Speaker interactions" - not passive background listening which is what I believe you are alluding to.
Not arguing that I am saying that latter does/doesn't happen - but just that this paper is not proof of it. And there is a big difference.
So algorithms can percolate his interests to me after we meet and some time it would happen that we talked about it too.
Maybe not the best example (please chime in if you have a better one!), ha you ever shopped for a specific model of used car? Once you're on the hunt, suddenly you'll start seeing them everywhere when you're out and about, even though previously you didn't notice. There aren't more of that vehicle on the road than before, but now you're attuned to the pattern.
On top of that, it would be easily detected via networking monitoring.
Doesn't seem like a sufficient benefit when you consider the downside -- a device listening to your conversations and selling that data to ad agencies.
Just like anything, there’s a trade off and, for the case of convenience, I’m willing to make the trade off between convenience of talking to a device to do tasks for me and having an ad agency learn what music I listen to that Amazon is probably selling to them anyway since I use the Amazon Music service.
I get your skepticism and applaud it, but you have to remember 99% of the population doesn’t know, care at all, or is willing to make the trade off, whereas the 1% are some of those folks amassed here on HN.
YMMV.
But.. there is hope as jokes along the lines 'Wiretap, what is the time?' seem to have become more common.
So. For now, I'm fairly convinced the data collected are useless/meaningless and I'm happy to go about my day w/o much worry. If I suddenly start seeing much better targeted recommendations, I'll rethink my strategy.
And heavens know how much better we will get at it. I am retard at analytics and the things I saw are already absolutely amazing. Just because ads can't seem to show you relevant information is not a good indication Amazon can't infer you like midget porn.
Can't speak for G's offering, but currently Rest of Ear of Sauron Alexa and chipper, but particularly thick, Siri either don't work very well or fall into the easy to live without novelty category for us.
[1] Particularly around Xmas, lots of switchs which are almost always buried somewhere behind the tree ...
It is different because it's always available without looking for your phone on one site and on the other it actually removes the need to have your phone with you all the time.
It also feels much more natural.
Independent of the benefits, I have a smartphone with a microphone, webcam, Laptops etc. The sentiment that a hardware verified device starts to listen in after specific keywords is for me less of an issue than all the other devices I just mentioned.
I really think that natural voice input will be the future. I want a personal assistant in my flat. And if Elon musk now starts the robot war, those robots will be controlled by voice. I'm pretty sure about it. I also saw plenty of ml research which indicates that.
Edit: and yes voice is so much more practical when you cook or in the shower.
Aside from annoying suggestions like "By the way, you can ask me who the most famous person in the world is".
However, I still feel weird with these devices in the room. I have decided I am not going to trust them, I won't ever trust them, and I'm blaming them until they show us the code.
I know I'm a doomer and a fatalist, but holy crap we are sleep-walking into dystopia. Your average HNer might be a little more attuned to the possibilities for abuse here, but the billions of clueless customers and hundreds of billions of dollars to be made will just keep tempting these behemoths further over the line and renormalize all of society around pervasive surveillance capitalism.
The entire psychological context of life has changed, and I'm not super OK with it, TBH.