https://gizmodo.com/amazon-is-uses-conversations-with-alexa-to-train-ai-1850873720
Not only will the tech giant's little assistant be listening to you, it'll sometimes use those conversations to train the company's newest product.
I'm never putting one of these in my home.
I work for Amazon.
This has been the case for many years. Amazon has used AI in Alexa and other services for many years as primary providers, and has told it's users it's used it's data for as long. We're talking from close to inception here, so 6-7 years, at least. Hell, LLM's aren't even new to most big tech companies!
I'm all for privacy, but if you want privacy then you probably shouldn't have a fucking tin can in your house that actions every conversation to a cloud service!
Considering I set up one of the content types that relates to wakeword and utterance text analysis for Alexa, I trust it completely.
But can I trust you? Are you willing to share the source code?
Edit: Tell me why I'm suppose to trust an internet rando?
There's this study for those interested in knowing more about how often these devices mistakenly record conversations:
https://moniotrlab.khoury.northeastern.edu/publications/smart-speakers-study-pets20/
Well,that's the thing with "news" right? Just scattered information without context for clicks. If people start connecting the dots and things make sense, most of the news become pretty uninteresting and would not evoke anger, prompting you to click and share.
Yeah, that's kinda the point. They literally tell you that your voice interactions are used to improve the service.
I will be the last person to not have a smart home. There will be a banner over the doorway: "Welcome to Stupid House".
There will be a small cover charge.
You can have a privacy-first smart home. I have. I run Home Assistant in a docker container. No external services/plugins. My smart doorbell streams to my local nvr. If my internet is down, everything keeps working. And it's not even that hard anymore. It's become a lot easier over the last 2-3 years. Still not for non-techie users, but a lot better.
That sounds pretty reasonable.
Edit: Still kind of want to call my place "Stupid House" for myriad other reasons
I'm not tech illiterate by any means, and everything after "home assistant" in that post is Greek to me
Docker is a way to run containers. Basically lightweight virtual servers. That makes it easy to run multiple servers on one machine. An NVR is a network video recorder. It's like a video security system like they use in stores where all cameras are viewed and recorded in a single place. I assume you know what a doorbell is 😄
Have any resources to get started with that? Been looking into security systems but don't fully trust nest/ring/simplisafe etc
Just start with a local Home Assistant on. Raspberry Pi and go from there: https://www.home-assistant.io/installation/
Teach me your ways please! Setting up a Home Assistant seems like such a daunting task. I'm stalling converting my devices into it. Any tips for a (home assistant) beginner?
I just followed the steps on their site. Containers give me cancer, so I did a real install on my home server.
Caveat: I am a professional software engineer (but I didn't really have to hack anything)
Start with a Raspberry Pi and just follow the docs: https://www.home-assistant.io/installation/
That's the easy way. I did it the hard way, but that's because I run on on a big dedicated home server together with a dozen other services.
I'm with you. I hate how they expect me to control everything from my phone or with voice commands. I'm fine walking to a light switch or walking to the thermostat.
There's a middle ground as well. I refuse to put Alexa or OK Google or whatever on any of my stuff, but I run home-assistant with zigbee smart devices. My entire setup runs completely cut off from the internet. I could in theory even air gap it, although that's a little overkill. It's a "smart" house, but one I'm 100% in control of.
Is that self hosted? I'd just about fuck with a FOSS self hosted smart home setup, but even then I could barely be arsed
Be careful running it in a Pi because it's a little heavy for that depending on how you configure it. A Pi model 4 is probably OK, but you wouldn't want to run it on a model 3 or something even older, and you're going to want to use one with at least 4GB of RAM.
I'll get a lot of hate for this but when you say pi you mean pi4. I kept seeing this HA on lemmy and tried it on a pi2. I don't know if it worked or not, it's a very bloated piece of software. After an hour of waiting I installed docker and the HA instance on my main server (which is ancient) in under a minute.
It's cool and all but my feit dimmers require some pcb work and flashing to be compatible so verify what devices you have before you hop in.
I used to have an automated building running on a bare 386 and a floppy drive. Hate on me all you want but sending simple commands like turn device on shouldn't require a giant software package but otherwise HA is neat, just a lot of overhead i can't exactly justify.
Worth trying out though.
I think reflow stole a lot of their code.
No hate from me.
Just about every project I've started with a pi has ended up working out a lot better as a vm on an x86 host. But lots of people seem to love them.
To be fair It has its uses i suppose. I've had one running pihole since the original pi came out. Used PI2s in the past for OSMC and, even better, ambilight.
I think now a cheap android TV box you can flash is probably better for a simple less than 5watt device.
Besides the HA test I've been trying to use one to be an openvpn TAP interface but it's been a fight and i think you just convinced me to do it in another docker instance on the server and save myself some headaches.
Yep, I run mine on an ODroid XU4, but you can run it on just about anything including a docker container on a generic Linux install.
I'm using z-wave stuff but similar setup. Home Assistant does reach out to the cloud for some things like weather forecast and Google calendar but otherwise it will operate 100% without internet if needed. I also have cameras that while they aren't air-gappend they are blocked from Internet access and can only talk to the NVR.
The last thing I want is to talk to a computer. Buttons are fine. The roboto phone customer service is bad enough.
I’m fine walking to a light switch or walking to the thermostat.
When the hallway light was left on again it's really damn nice to simply say "Turn Off Hallway Light" while staying under my nice warm covers. It's also pretty swank to have the garage lights turn on when the garage door opens then turn themselves back off 5 minutes after the garage door closes. Someone left their closet light on? No problemo, my automation catches that and shuts it off.
Window coverings like blinds and drapes? Yeah, those are opening and closing automatically based on the position of the sun, even when I'm not home to do it. Did it rain while I was at work? Automation keeps my sprinklers from running tonight.
All of that is being done by Home Assistant and absolutely no Internet is required to make it work.
it's really damn nice to simply say "Turn Off Hallway Light"
Can you use a custom wake word? The only reason why I'm still using Alexa is for the "computer" wake word.
HA doesn't have "Wake Word" yet. It supposed to be coming before the end of the year. Right now you have to PTT (Push To Talk) to make it listen and for the privacy minded this is actually better than an always listening device.
Still, a lot of people want WW support and it is on the road map.
When skynet comes online, I'll die quickly, being mopped to death. You'll have to struggle in the post apocalyptic hellscape where humans fight robots with A-10s for some reason.
I think most people, me included, underestimate the scale of the operation. When you hear "company will use private data to do X", you imagine what a reasonable person would do, like random sample a few conversations here and there. In reality they record everything permanently over months and years, far beyond what would be necessary to run the service.
It's kind of crazy how we get this level of surveillance while still having software that will lose your data if you don't hit Save often enough.
What's fucked up is if you try to regulate it and make these companies have data retention policies. It creates a giant moat around them where no newcomer can have a chance to compete.
That's because you are not enforcing data portability at the same time. Having studied and discussed the GDPR at length within tech circles, I became convinced that data portability is the ultimate right and the key to ensure continuing innovation
data portability is the ultimate right and the key to ensure continuing innovation
Interoperability in general is the solution to walled gardens and monopolies that harm competition, consumers, and innovation.
Yep, most of my interactions lately that are non trivial are just sigh-inducing. Smart assistant my ass.
The new Amazon AI is going to be remarkably foul-mouthed. Every time it screws up (and it screws up a lot) I have to curse at it to make it shut up so it can hear the command again.
I brings me joy when I tell her "Alexa, shut up you dumb bitch" and then she responds with that sad minor tone dejected sound.
We always knew that. What they don't tell you is your phone is also secretly listening. "Ok Google" <- turn that thing off too
So who thinks this conversation here on lemmy isn’t being used to train an AI? Maybe not right now but later?
Sure the relatively small size of lemmy means it might not be scooped up and trained on. But the point still stands. All that is publicly online is food for the big-corp AI builders. And while Alexa invading your home privacy is obviously a shitty thing, I’m not sure we’ve all thought through the new relationship between us, the internet and the big AIs.
Well I know I have no expectation of privacy here, but I'd rather open source LLMs train on my words along with proprietary ones, than some company hoarding information and selling it to each other.
The only thing an AI trained on Lemmy will ever be able to do is discuss the merits of socialism and talk about Linux lol
I love being able to dictate a grocery list but god damn is she stupid.
Good luck asking for cream cheese and chive crackers without ending up with cream cheese as one item and chive crackers as another. Or worse peanut butter and honey crackers as peanut butter and then honey crackers
The problem is that Alexa isn't actually parsing the meaning of the total phrase, she's taking each individual word as it comes. With that context, she would just as easily interpret your phrasing as "thing with thing on the side". You'd still get chive crackers, honey crackers, peanut butter, and cream cheese.
Edit: I thought about this a bit more, and it seems to me the only way Alexa could actually understand what you wanted is if you said "chive cream cheese crackers" or "peanut butter honey crackers". You have to implicitly make it one item and not a potential combination of multiple items.
Yeah I had the same realization as I was reading this:
you can not add Boolean terms and expect it to not separate it.
So run on nonsense sentences are your friend and will definitely make the training from you so much more useful.
An always on microphone connected to a company that is mostly known to exploit their customers and employees! Say it ain't so!
Yeah, I realized these things are terrible about a year ago. So, I hacked them into computer speakers using some cheap amps and a 12 volt power supply.
I'd love a citation on this, outside of wakeword usage (a local device waiting for "Alexa* to begin recording).
Source: their ass.
Alexa devices use an onboard DSP to detect the wakeword and maintain a rolling audio buffer. On a positive match, the DSP wakes the main CPU which combines the saved buffer and any following speech and uploads it to the cloud where Alexa lives so she can try to figure out what you meant.
No audio is uploaded without being triggered by a wakeword. Also, the "mute" button physically cuts power to the mic, and the indicator LED is hardwired to the power rail as a failsafe indicator.
Oh, I know. For reference, I work at Amazon, so always interested in where these stories come from 😂
That was more for everyone else's edification. I worked on a plethora of Amazon devices over the course of a decade.
Not going to get much out of me then, most of what Alexander hears is what's on the TV or music I listen to. If they want to train alexa on that, their fucked
It’ll be really good at telling people to shut the fuck up if it’s using my data for training.
These types of projects are driven by metrics, and teams have some kind of quota/goal that they need to reach by a certain date to keep the project on schedule. Bonuses or job security may be on the line here, and so you may see some desperate employees "going the extra mile" to reach their goals.
Relatedly, Alexa's voice activation sensitivity is essentially a tunable number. It can be changed to be more sensitive, so that it will activate more easily (e.g. maybe you say "Alex" instead of "Alexa"). The people who control this are likely on the team with that deadline, so the incentives are there to lower this value in order to collect more data by recording personal conversations "accidentally". Maybe a bad update goes out that causes Alexa to activate randomly, and they quickly fix it after a few days when they collected all the non-Alexa personal conversations they need for their AI.
That's maybe a bit too deep into the paranoia/tinfoil hat spectrum for some, but history has shown that you can't give big tech the benefit of the doubt. Especially when you see some of the documents from the Google trial, where executives discuss rolling back new features to improve arbitrary metrics in the short term so that they can get their bonuses for the quarter, even if it hurts consumers.
it is only hard if you want it to be hard
WTF is this bullshit. Not everyone is equally capable and this reductionist shit is hurting those that do need the help.
well the title says "as long as you can", not sure what is the issue here, just do whatever you can, don't fold if you don't need to, that's it
Is this a surprise to anyone?
This was already my understanding when I got the first pre-release one in 2014.
In that time, it has mainly learned how to"dim the living area lights to 50%" and "set the AC to 22 degrees". That is about 99% of it's use.
Wonder if that's helped it's AI much...