Having AI assistants ruling our future lives? That's so sad. Alexa play Despacito At the annual spectacular of crap that we optimistically term the Consumer Electronics Show, I found myself locked into a room with Alexa. And yes, disembodied AI servant though she may be, she still filled every corner of a rather spacious room with all of the various products now connected to and enabled by Alexa - …
I won't have Alexa in my house.
I'd love to get one. It'd be potentially very useful to me. But, and its a Diane Abbott sized butt, I'd have to have absolute confidence in what data it was sending outside of my house, when, and why. And I'm afraid I just don't have that confidence.
How do I really know its not sending conversations with other people out to Amazon? Because they're marketing department say it isn't, isn't actually good enough.
I decided that market trust is what keeps Amazon from snooping. The risks of them being caught are pretty much 100% (traffic analysis, insider leaks) and the upside isn't that big. So I have two Echo dots, and I can control my room thermostat and my main home theatre ("Alexa, turn on Apple TV") and play music in sync over both. That's pretty much the extent of its usefulness, it's a gimmick and if the interwebs go away, nothing bad happens. If Amazon is caught making a boo-boo, my $70 go into the trashbin and I'll have to look for the Harmony remote again to switch the telly on.
One of the reasons I bought a couple was to find out how these things could become genuinely useful, but so far I've come up empty-handed. It's just a toy, pure and simple.
The risks of them being caught are pretty much 100% (traffic analysis, insider leaks) and the upside isn't that big.
But the downside is even smaller. Most people only care about privacy when actively prompted about the subject. Even then they know little, and any concerns they have won't be translated into actions or enduring behaviours. My Android handset has all the permissions that I can bolted down or refused, nothing installed and running unless I need it (ueah, I know its still Google-infested). But other than people round here, the vast majority of Android users don't give a tinker's cuss about that stuff, and blithely let Google decide what should run. They install and use Facebook apps, etc etc.
So I'm wholly unconvinced that any "smart assistant" will honour data protection out of concern for the consequences.
Wait, you mean the "downside" that Facebook is experiencing due to their repeated bad news stories about privacy? Too bad about the hundreds of millions of users fleeing them....oh wait, there is no detectable loss of users, they are just fine! People grumble a bit but they keep using Facebook and keep seeing all those ads Facebook is getting paid for.
Amazon knows that no one will care - the only ones who do are people like me who don't and never would put an Amazon or Google listening device in their home. So they have nothing to lose by snooping on the rest, and you are willingly allowing them to do it to and your family. You "market trust" argument is just another way to live in denial in exchange for a slight bit of convenience.
"Denial" goes both ways you know..
With the device being present in user homes it can be checked and they will get caught. This has not happened - over 4 years. They are not ambiguous that this is not done, the trigger must be detected and they light up when uploading.
People will understand "eavesdropping" and react far more strongly than facebook selling their "data" and "privacy". It is the reason for the allure of voice interaction - it is the predominant interface humans use.
When people write their diaries on facebook, they will care about privacy a lot more.
There are valid concerns with voice assistants, but "constantly uploading conversations" is not one with the current set of echo/home/siri devices & firmware. There isn't a shred of evidence other than paranoia and cynicism.
Bugs likely in third party implementations causing false detects would be a valid concern.
I agree with what you've said.
I find it unlikely a commercial operation would, for commercial reasons, intentionally turn these devices into bugs.
But note there are 2 caveats there, 1) commercial reasons, 2) intent.
As you pointed out, most likely any such situation would be caused by either inadvertent bugs, or some developer/engineer thinking something would be a cool feature to add, not getting it properly cleared (e.g. not explaining properly to management what a new feature is actually doing, or just not telling them at all). In the latter case, while intentionally introduced by the dev teams, it may be against company policy and/or just not thought through fully what this new fancy feature could lead to. In which case once it was realised what the consequences are or management becomes aware of and understands it, it'd be quickly removed.
However, honestly I am more concerned about the first caveat, "commercial reasons". There might be legal reasons for them to introduce such firmware to enable bugging. Such situations as the Apple v FBI case where the FBI were trying to force Apple to specifically write hacked firmware and directly upload it to the targeted phone(s) to assist the FBI in cracking those phones. This case was never resolved, as when it wasn't looking too good for the FBI they 'magically'* found an alternative way and dropped the case (without prejudice, i.e. no binding judgement was made and the FBI can re-litigate the same or similar cases).
Therefore I think it more likely that these devices would be hacked for profit/disruption/fun illegally or that Amazon or other such services might receive an NSL or All Writs Warrant to, for example, digitally sign the firmware the TLR agency has provided and to force update specified target devices (or all devices at a specified address) - no questions asked.
*) the FBI actually had the means to crack these phones already via 3rd party commercial forensic software, but conveniently 'forgot' this fact as they were actually attempting to set a precedent favourable to them, as opposed to actually needing it for that particular case. Which is why they dropped the case when it was no longer looking like a slam-dunk win, as they didn't want to risk setting a precedent against them with this case since they did have the alternative means to crack the phone.
"I find it unlikely a commercial operation would, for commercial reasons, intentionally turn these devices into bugs."
These devices literally _are_ bugs by design and definition. "intention" is irrelevant because you can sell anything they collect, therefore money. Privacy for sale.
Encrypt everything and deny any wrongdoing, no worries. For the company.
"With the device being present in user homes it can be checked and they will get caught. "
It can? How exactly and who does it?
Everything it sends is encrypted and therefore by definition non-checkable, at least not by you or anyone you know.
That's the basis for not getting caught: Encrypt everything and deny any spying; No proof, no spying.
"... but "constantly uploading conversations" is not one with the current set of echo/home/siri devices & firmware. "
Of course it is. How you think it knows someone commands it to do something? It has zero local analysis capability and it's basically a microphone with internet connection at hardware level, it can't do anything at all locally.
Also, how would you know, have you access to firmware source code and if you do, how much are you paid to lie about it?
"There isn't a shred of evidence other than paranoia and cynicism."
So you deny the existing evidence? Nice. Another lie.
Because there _is_ way too many "coincidences" to just dismiss like that. And even anecdotal evidence is enough in this case: Proving something exist needs only one example.
Proving it doesn't exist is much harder: Where's the analysis of traffic Alexa sends, unencrypted and all of it? Or firmware source code? Oh, there isn't. What was your point again?
Summary: How much Amazon pays you?
"Everything it sends is encrypted and therefore by definition non-checkable"
Not possible to check content but possible to detect frequency and volume. So it should be fairly easy to check if it's sending all the time or just when receiving a command
"How you think it knows someone commands it to do something? It has zero local analysis capability"
AFAIK the way it works is that it has enough local processing power to recognize the keyword, and only sends out to the internet once keyword is detected.
Of course keyword recognition is itself a bit dodgy - personally I would by far prefer an assistant that only activates with something a bit less ambiguous like a button press combines with a "Hello Assistant". That is double excellent because it's like intercomming your personal secretary in the next office :)
So it should be fairly easy to check if it's sending all the time or just when receiving a command
Sending may only happen when the command is issued, but that doesn't mean it isn't recording and compressing at other times or for other trigger phrases.
I don't want to sound too tinfoil hat about it, as its possible, even likely that they're only doing what they claim to be doing.... but the possible downsides if they're not are quite significant - the upside while useful, seems limited.
"Of course keyword recognition is itself a bit dodgy - personally I would by far prefer an assistant that only activates with something a bit less ambiguous like a button press combines with a "Hello Assistant". That is double excellent because it's like intercomming your personal secretary in the next office :)"
I'd prefer it if I could set my own activation word or phrase. No button required because, like most people in the world, I have no idea what it's like "intercomming your personal secretary in the next office" :-)
"Not possible to check content but possible to detect frequency and volume. So it should be fairly easy to check if it's sending all the time or just when receiving a command"
I'm a bit late with this but..
On a decent ADSL connection, how long would it take to upload 5 minutes of audio recorded at say 128k? What if it's at 96K - still decent enough for conversations to be clear.
How much background traffic does the device do? Does it 'poll for updates' every hour, and how much data is sent?
What law is there that prevents the makers from having it NOT give any lights when it is working, and what prevents them from writing a bit of code that makes the device upload random (or not) junk when idle? If I wanted to test if it was uploading everything or not, I'd have a traffic monitor on it and see how much it uploads when there are conversations (not using the trigger word) vs when there aren't conversations in the room. If there's virtually no traffic in both instances I think it's safe (until LEO orders a change, or until a certain other trigger is reached). If there's a constant low level of traffic then I'd assume that it's sending bits of the conversations during the day at quiet times, or it's sending junk to fool anyone watching for traffic spikes. If I was writing something I knew could cause trouble, I'd find a way to smooth over the bumps in the data. And of course when it is communicating with base in response to a command, I'd send every bit of extra data I could get away with at the time :)
You mean Facebook, that company that saw its userbase growth slow down markedly and saw its stock drop by over a third after all the scandals? It seems that at least some people care.
Also - Facebook's primary product is selling data about you. Amazon's primary product is selling stuff to you. There's less of a network effect and more alternatives in the latter, and as others have remarked - outright snooping in your homes is a tad worse than Facebook figuring out what you do and showing you ads in response. I use iPhones over Android for the same reason - I'm more the customer and less the product.
I guess everybody makes their own decisions in this area. It's a fine balance between trying to protect your privacy and enjoying some of the advantages of the modern age. YMMV, etc.
"The risks of them being caught are pretty much 100% (traffic analysis, insider leaks)"
Nope, it's more like 0%.
Everything it sends is encrypted and you can bet there's only handful of people in Amazon who knows exactly what is sent/received. And they won't talk because of NDAs.
So the argument about "trust" is basically null and void: There's literally nothing at all which would justify trusting Amazon.
"It's just a toy, pure and simple."
It's not pure or simple at all: Alexa records and sends everything spoken to Amazon "for analysis" and you can bet that data is sold immediately to advertises.
Too many "coincidences" happening for that not being a proven fact and Amazon claiming it doesn't happen (without opening any of the traffic of course) is just BS: They lie anything to keep the revenue got from selling that data.
>> Nope, it's more like 0%.
Encrypted traffic isn't 0 bytes - what rubbish. Check traffic when quiet, check traffic when speaking, check traffic after trigger words.
>> the argument about "trust" is basically null and void
By your logic, no internet connected device is usable. "They'll have a <sensor of choice> in there and sell it to advertisers..."
Why are you on a laptop - are you sure the mic isn't recording? Your phone? What about your modem? your mobile operator, your subway rides...
So the internet is actually a spy network designed by the us military.. and so use the internet at your peril.
"Too many "coincidences" happening"
If you believe what you say, you couldn't have posted on this forum.
Why are you on a laptop - are you sure the mic isn't recording? Your phone? What about your modem? your mobile operator, your subway rides...
Yup. No microphone, physically disabled unless I switch it on (and yes, I know what a microphone looks like and no, it's not easy to hide one elsewhere on the board). I know my phone isn't, 1) it's a dumbphone, 2) it's not got a data plan, 3) it's not in the same room as me.
The modems are a long way away in another building, and pretty sure by now someone would've found a microphone on them. Network-wise I use several other layers.
The nearest subway is a few thousand miles away, and the local public transport ops have no way of knowing who I am.
There is an odd doohickey on my early 90's car engine that I don't recognise and haven't bothered to look up. Maybe that's a hidden microphone? OH NOES! DAYZ SPEYEINK ON ME!
The most 'interesting' things I do is post the odd subversive comment here or on the local rag's forums. If I was every to have anything more interesting to discuss, it'll be in an outdoors environment some distance from the nearest electronics. If we're really worried we'll wear a scarf over our mouths.
If your in another room Alexia and Google cannot hear you. However most routers do not contain switches and rely on CSMA and broadcast data . To the best of my knowledge nobody addressed that the personal assistance are looking at all the data good, bad and ugly and passing on to third parties.
Another reason not to have one!
"I won't have Alexa in my house. My in laws do, and my offspring uses it to play their favourite music when we visit. It wouldn't be so bad if their favourite music didn't keep changing every 20 seconds..."
Next time you visit...
Alexa, pour me 5 olympic sized swimming pools of water.
Alexa, shut down all audio input for 3 hours.
Alexa, have three tons of rock salt delivered to the front lawn.
Alexa, call the pool-boy and have the swimming pool drained.
Alexa, have my cars towed to Duluth City Hall.
Alexa, order 14 dozen pizzas.
Alexa, book a three month safari in sub-Saharan Africa.
Alexa, stop all automatic payments. --or-- Alexa, pay all my bills tonight.
Etc.
It's going to happen. It's inevitable. We all know it. Now ask me if I'll ever buy into the daft concept.
“Alexa, pour me 500ml of water.” That’s why. It’s less about activation than precise control - something easy for a smart faucet and much harder for a human
I have a brilliant idea. Why not get a plain glass jug, and paint a line on the side to show precisely where 500ml is? And as a bonus you could paint more lines at each 100ml point - or even do it in fl.oz. as well.
Can I patent this idea please?
I think the other poster means that some types of flow meters have a rotating spinning element, and the sensor senses the rotation of this known object in the flow with each space between tooth of the element forming a known volume. Or it has a chamber that measures a given volume and displaces this chamber once per operation. Variable stroke if it needs to service different quantities. So in these cases it doesn't care about the colour of what is making it rotate, and in the case of the latter, it doesn't care how gloopy the liquid is, along as it passes through the valving inside.
OK, add sugar levels, viscosity, coefficient of thermal expansion, room temperature, liquid temperature, flow rate, nozzle shape, thixotropic nature, corrosion, flow velocity, materials used to make the tap' .... struggling now.
Sorry, I was an instrument lab tech for thirty years.
"OK, add sugar levels, viscosity, coefficient of thermal expansion, room temperature, liquid temperature, flow rate, nozzle shape, thixotropic nature, corrosion, flow velocity, materials used to make the tap'"
Surely for lab environment these are necessary, but do they make THAT much difference? In a pub setting is it really that much if a serving is 499ml or 501ml rather than 500?
What have density or colour to do with volume? Liter is a measure of volume. 500 ml of helium is the same volume as 500 ml of water, just the mass is different. So it doesn't matter if one beer has a higher density than the other. I suspect the staff of your craft beer bar consists of young hipsters who have no clue how to pour a beer properly from a tap.
What have density or colour to do with volume?
Nothing. See above re. colour. But density and viscosity will affect flow. In a setup you will have a known pressure and pipe size but some liquids will flow at different speeds. That said, I'm not a fluid dynamics engineer so I could be talking absolute shite! ;-)
I think the taps are there both for deskilling and efficiency: getting skilled bar staff is getting harder and harder but also efficiency. The focus is on lots of small servings so if you can reduce the time it takes to pour these then it can make quite a difference.
That said, I'm only trying to point out that Silicon Valley solutionists often find the wrong problems to solve.
What have density or colour to do with volume? Liter is a measure of volume. 500 ml of helium is the same volume as 500 ml of water, just the mass is different. So it doesn't matter if one beer has a higher density than the other. I suspect the staff of your craft beer bar consists of young hipsters who have no clue how to pour a beer properly from a tap.
The point was I believe, other properties may have an impact depending on the measurement method. I have a jug than can measure 500ml of water just fine, but I would really struggle to use it to measure 500ml of helium
FWIW the local craft beer bar has flow control pumps for serving, er, precisely 200, 300 or 500 ml.
I find it ironic that a craft beer dispenser is dispensing with the craft of dispensing beer. Admittedly, its been a while since I dispensed beer for a living, but most of my regulars had preferences about how their beer was pulled :)
I find it ironic that a craft beer dispenser is dispensing with the craft of dispensing beer.
It is. But the "craft" beer market has little to do with the "real ale" one which I prefer. The prices at the place are eye-watering, so I don't go very often, but fair play to the guys as they seem to be making a go of it in a slightly less fashionable part of the Düsseldorf Altstadt and I'm generally in favour of anything that goes against the trend of Heineken everywhere.
It's not a difficult problem to solve at all.
I used to be a pub manager in a previous carreer (early 1990's) and we were using flow-controlled pumps for beers back then. No complex electronics - just a simple activation switch for the pump, which was invariably of the rotating-chamber type. there's usually a pressure-assist in there to help with maintiaining an even volume of fluid regardless of how gloopy it is.
... and let's also face it. Different beers don't have /that/ much variance in their gloopiness.
Years ago on a cold night I filled the car with petrol and drove the 20 miles back home. When I got up the following morning it was a lot warmer and the petrol had expanded and overflowed! So I ended up with more petrol than I started with. Not totally on topic I know, but I found it interesting at the time.
Actually, no thanks, foul stuff.
I got a non-smart Breville "Hot Cop" machine. When activated by hand, it dispenses 500 ml of good boiling water from its spout, after 60 seconds.
If you don't want an exact multiple of 500 ml of boiling water at 60 second intervals then you are out of luck, and I find it more than I need - in fact I needed extra-large cups. But a current model of the machine now allows the quantity to be varied. But, as far as I know, not verbally.
It just occurred to me, maybe I could put glass marbles or something in the machine and get less water out. (Is that what they've done?) Supposing that I can buy Pyrex marbles...
Not to mention that when cooking, precise control of measurements is antithetical to cooking.
Recipes provide approximate quantities that are perfectly fine if within +-10% - at least! - of the specified value. One of the arts of cooking is adjusting the quantities used to suit tastes and/or compensate for substitutions and/or compensate for different amounts of various ingredients added.
It's unlikely with nearly any recipe that putting 450ml or 550ml of water in when it it specifies 500ml would make any difference at all. And any subtle difference it does make could be compensated by adjusting other factors if necessary, e.g. adding more flour if you added too much water (if doing pastries), or simmering shorter/longer time.
Most of the fun in cooking is varying the quantities - or substituting or entirely adding/removing ingredients - specified in a recipe to suit ones tastes or just for the hell of it.
how about releasing them blades when the owner, baffled by the sudden silence, does the obvious, by getting their mug close enough to the device to inspect the pulp, or, perhaps, putting their pinky into the blender to give the blades a gentle push? Because that's what humans do (when guns don't fire, mixers fail to mix, etc.)
Then again some may dream of telling Alex's to go screw herself when the voice response isn't particularly responsive.
Which leads inevitably (for some) to errant thoughts of soft porn responses.
Alexa go f**k yourself!
I'm sorry, I don't know how to do that.
Hand on, Cortana says she will show me how...wow...oh!
{Cue girl on girl porn sounds}
Siri - what are you......oh!
Alternatively
Would you like me to use (product placement)?
You use it like (instructions).
It sounds like (electronics).
And it feels like..........
Would you like me to oh oh oh order one?
Yeah, just me.
Not all of us! As I get older, I'm going the reverse direction. Doing it myself gets me off my ass and moving and I know it's done right. Measure 500ml of liquid? That's what measuring cups are for, read the bottom of the meniscus and you're close enough. Why spend $$$ on Alexa, when I already have a Tiki to do all that and more? Dream of automating to become indolent? No thanks!
I won't have any of these "AI Products" reasons being:
1: It's just another thing to break and make stuff harder to repair.
2: Its yet more IOT $%&t you will be lucky if it gets even gets 1 patch leave out a decade's worth that a lot of white goods will need.
3: It's a microphone in your home controlled by the creepiest companies in existence.
These are enough reasons for me to avoid this stuff like a plague.
The problem is when it becomes either mandatory, or significantly cheaper than the alternative. For example, you can't rent DVDs from local shops any more. In London, you can still buy paper tickets for the Underground, but you'll pay twice the price of the equivalent contactless fare.
Alexa may not be mandatory yet, but the "Eleven" lift is the kind of thing we can expect. Link: https://www.youtube.com/watch?v=sAz_UvnUeuU
ermm all your examples are cases of things being replaced by more efficient technology and so is cheaper.. Everything has something good and bad about it, and older tech being more expensive than newer more efficient tech - the efficiency is correctly being costed..
That's exactly as it ought to be. I don't expect snail mail to be cheaper.
Do not confuse having choices across multiple vendors for new technology with not adopting new technology at all. That is a valid problem Alexa might be creating - the only voice assistant tech.
When you come up with an "AI" that can run locally without any internet connection we can start debating whether the idea has sufficient merit. Not until then. And I'm serious about "not connected to the internet in any way shape or form". If that means it can't give me weather forecasts, place online orders anywhere at all, or give me the number of cars currently impeding my arrival to work then so be it: I'm perfectly fine with that. Whether that leaves any useful features worth bothering with remains to be seen.
Voice-control systems are easy to build using off-the-shelf components. A few electronic pieces, CMU Sphinx, TensorFlow - you could have a pure-local voice-controlled "AI assistant" up and running in a couple of days, or maybe a week if you're technical but new to the domain.
Personally, I don't see the appeal. I find voice-controlled devices immediately and deeply obnoxious.
Yeah, like it is a matter of life and death if I pour myself 497ml of water.
This is what's wrong with AI. If I'm asking you for a glass of water you will pour some water in it and hand it to me. If it's not enough, I'll ask for more and if it's too much, I'll drink only the amount I need. You as a human will not be baffled that I didn't specify 500ml of water.
To those AI peddlers, please get a life and use you talent for some other occupation. No human being has ever drank water in precise quantity so stop pestering us with your ugly child. Just because you can does not mean you have to do it.
It's not, but pour me 497ml of beer and we're going to have to have a conversation! It's 568.261ml all the way. Presumably after March 29th all the AI will have to be re-calibrated and we'll go back to proper pints and all the 'ml' nonsense will be over?
Mine's the one with the Al Murray tickets in the pocket
Er, what country are you living in, coz the UK still sells beer in "proper" pints.
And that proper pint is actually a legally defined number of ml because the UK adopted metric before it adopted the EU. That may have been before you were born, though.
Its a well trained algorthm. This and no more.
And its a microphone in my house, managed, controlled and tracked by a corporation that has *already* made mistakes.
Much like the "my ipone camera is SOooo much better than that dslr. I mean if you did all sorts of post work on the DSLR it might look so good". Your shitty little ipone camera *DOES* all sorts of post work on the image, before you see it you tit.
Its a well trained algorthm. This and no more.
Totally agree. Most of these systems touted as AI are what in the 70's, 80's and 90's we'd have called an "expert system", not an AI system. And that's not because there was no concept of AI, that it's just a new name for "expert system". The concepts of AI have existed for decades. It's just that now people are changing the meaning of AI, it's like AT&T appropriating "5G" to label 4G services for marketing as "5GE".
AI is no longer a technical term with a solid definition, it's just become a marketing buzzword like Cloud that only has a vague meaning.
You would think somethings such as the IoT, Drones & AI Assistants would be warmly welcomed by the frail aged and people with dementia, but most miss their own capacity, to be able to stand up walk across a room and turn on a tap, to reach for the tea caddy, carry the tray to the table by their chair.
Their mind, legs and hands weak & dysfunctioning prevent them from often understanding the order of tasks to make a cup of tea, turning a tap, lifting a teapot.
They depend on carers or do without. Alexa could do little to aid them, the primary focus of these devices is upon the mass consumer market of able bodied people that just could not be bothered or whose lives are so lacking they need some novelty.
I-pads & other tablet devices (primarily old ones) only later became useful with people suffering from autism, who for whatever reason could not speak and so opened up the possibility of communication - Technology for the disabled and infirm is not a rich market but a large and needy one.
Think of them when you make tech and look for markets.
have been bricked by updates, or changes to T&C ?
So even if you buy something that provides a useful service (now), the only real guarantee that it will continue to provide same useful service later, is to isolate it from external connections. This has the added benefits of preventing malicious actions, AND ensures your privacy.
The smartest thing you can do, is buy dumb.
brother has an alexa...
God its annoying.... knowing full well its listening in to your chat, and waiting for the 'alexa' keyword before responding.
And its nothing like an AI assistant.... because that pause between you giving it the command and it actually doing something is where its sending your voice data to god knows where, processing it and sending a command back.
But its going to be a boon for law enforcement
"Alexa.. play back the last 12 hrs of conversation at number 11 Acardia avenue"........ And you wont even need a warrent
Since recently, I've been thinking about what the future holds for my social life, and it appears pretty bleak. I know I wont' have any alexa, siri or other google pseudo-ai crap in my house. But then, what about my, say 16-year old (when they get to this age) want one in their room, because "all me mates have it!" and when I tell them, for the 1000th time to fuck off and get a life, he reports me to authorities for violating his basic human rights? What about when I visit somebody else and see this alexa crap in their living room? Do I storm out, do I add them to my blacklist never to visit again? And what about when alexa-google-siri crap gets embedded in people's "smart" glasses? Do I stop talking to people? It's all funny ha-ha, although many people, clearly, have this crap in their homes already, so it's only going to get worse...
What underhand seditious plots do you talk about when you go to a friends house? In fact, now I think about it, I bet you don't actually have friends and this is very hypothetical. If you do, though, and you really want to cause fun then when you are at their house say something like "Oh! You have one of those Amazon things? What does it do? Can you say things like "Alexa order me 10000 rolls of toilet paper?" Then watch the panic.
But then I have spent over 10 years living with people with serious disabilities. If you can tie a smart assistant into a tap that gives tea or coffee temperature water you would turn what for fully able people is a 5 minute task, but for a disabled person 20 minutes, back to a 5 minute task. Contrary to what others are saying - that is very humanising for them.
Not many that you can say "Pour a cup of tea" to. There are some with sensors, etc. But if you have very limited mobility the ability to say "Alexa, pause the TV and turn on the kitchen lights" then to move to the kitchen and say "Alexa, pour me a cup of hot water". I have some of these things in place and it just makes everything easier for them. Also for us being able to check from upstairs what's ben left on, etc. I could write a solution - I could buy alternatives - I could.... but why when the Alexa stuff is so easy. I couldn't give two hoots if Amazon knows when I turn the lights on.
It's often the case that those brave/well financed enough to be the first into a new market are generally not those ultimately most successful long term. They take the pain to prove a market exists, then competition comes along and drives improvements for the customer, at the cost of the pioneers.
For the voice assistants, as the technology exists today, there is a need for truly massive amounts of data to train the algorithms on and for hugely powerful cloud compute for the context inference to work. This is a huge barrier to entry, leaving Amazon, Google and to a lesser extent Apple and Samsung as the only options right now.
Until an open data set is available and until the AI can run on less massive compute, or the compute becomes more available, I don't see the situation improving. Sadly.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
