Voice Assistants Six+ Years In
We have many voice assistants in our lives.
I have Google Assistant on my phone. It is great.
The Gotham Gal has Siri on her phone. It is OK.
We have Amazon Alexa and Google Home in our apartment.
We have Siri on our AppleTV/Siri Remote.
We can talk to our car.
But the honest truth is we rarely use any of them.
The one we use most is the Siri Remote when using AppleTV because it is by far the best way to control that device.
It is not an issue of the quality of the voice recognition on these services. It is great.
It is a question of the relatively weak utility of the experience relative to alternatives combined with not building the muscle memory to use voice assistants more.
I was walking home from the gym this morning and was wondering if we are typical.
So I just ran this Twitter poll:
Do you regularly use a voice assistant in your life?:
— Fred Wilson (@fredwilson) April 8, 2018
Twitter polls only allow for four options so this is far from scientific and also suffers from the bias inherent in my Twitter followership.
But almost 500 replies in, over 60% replied no.
About 20% say they use a voice assistant regularly in their home.
And about 10% say they use a voice assistant on their phone.
I’m happy to learn that we are not in the luddite category when it comes to voice assistants.
More like the mainstream.
But certainly not in the early adopter cohort which is not small.
It does make me wonder how many of those Alexas and Google Homes are sitting around collecting dust.
Comments (Archived):
I think the challenge here is in saying, “they’re great.”They’re not great because they *only* work 95% of the time.And we’re not used to a digital experience that fails or frustrates us that often. Particularly when there’s an alternative that feels like it works 100% of the time.My hunch is that as the training moves forward and the tech develops exponentially, we’ll be surprised to discover that voice works 105% of the time. Better than direct interaction. That AI will combine with the voice recognition to make things even more convenient.Tim Wu’s article on convenience is a must read. We’re all in a hurry to do less, decide less and give up agency, it seems.https://www.nytimes.com/201…
maybe we are in a hurry to do less so we can do more
Although I find most of his writing to be superficial shtick, Seth is almost certainly correct. The Apple Newton was a failure when viewed at within a few years of it being launched. But if the Apple Newton were seen as an alpha version of the iPhone then it was wildly successful. Google Glass should be looked at in the same light.Most new technologies have promise that is only fully realized when the last 5 to 10% of the “kinks” get worked out. Automobile driver’s stuck in the mud were initially derided with comments like, “Get a horse!” Paved roads and steel belted tires were innovations that helped make the automobile what it is today.On-demand couchsurfing? How could that ever turn into a real business? I guess that is why Fred passed on Airbnb. I am baffled that Fred fails to see that voice is really still in alpha or beta, despite the marketing spin. Voice control will likely be so ubiquitous within the next 5 to 10 years that we will forget that there was resistance to it.Fred need to be concerned about being a Luddite, but he does seem headed down the path to becoming a curmudgeon. “People these days don’t embrace alpha/beta technology immediately! What is wrong with them?!?” Relax Fred. People are people. Users don’t care about features but rather benefits. Stop focusing on “bright shiny objects” and instead focus on what delivers perceived and actual value to customers/clients.When voice assistants work “105%” like Seth indicated, people will very likely embrace it as long as the privacy concerns are addressed, most likely via legislation of national governments. But at 95% voice assistants are not worth the hassle for most people. Sheesh. This is part of “Marketing 101.” C’mon Fred. Focus man! Get your head in the game! This is neither a complex nor sophisticated concept.Next time you might run your marketing questions/concerns by Seth before posting them on this blog. I am not a fan of Seth’s. Not at all. But he knows “Marketing 101” down cold.
I guess that is why Fred passed on Airbnb.I think from what I remember Fred passed on airbnb because he couldn’t wrap his head around the millennial notion of letting strangers use your hard earned place of residence and peek into your drawers. I think he said the older members of his team agreed with him mostly but the younger members disagreed and were in favor of the investment (or something like that). I don’t think (my point) that it had to do with potential and I am pretty sure by the time he was approached by PG it was clear it was more than blow up mattresses.I agree with him on this 100%. The idea of letting someone use my house and my hard earned things is distonic to say the least. Never going to happen.By the way more ideas fail and you never hear about them than succeed. I can’t prove this because you never hear about them. Segway isn’t exactly a resounding success in the way it was presented. Flying cars never happened. Self driving cars? Let’s see when we get there with that. UBI? Ditto. Blockchain? Ditto.Seth is very good at what he does and makes money. That’s the beginning and end of it for me. The fact that many of the things he writes are obvious to people who have been around isn’t important. As I said in my ‘idiot paradox’ if everyone was as smart as me I wouldn’t be able to make any money. I mean it’s fine to think that owning a small fine restaurant is more important than owning a chain of rib joints but it really isn’t if you aren’t making as much money at the small fine restaurant.
+1If it were more rewarding and practical to interact with other human beings who are in our physical presence using digital screens, cue cards or sign language, we would. But why would we, when just talking is such an incredibly useful and efficient interface between sentient beings. (For those who doubt it, read the transcript of a conversation sometime. Talking is not even remotely written text said aloud. It’s human jazz.) As Seth points out, the kinks are still very noticeable but that’s short term. Alexa, Siri and Google do not need to be sentient, per se, but as they come to be able to seamlessly mimic sentience, then voice interactions will become as common and second nature as talking to another person. In fact, I predict we here today will be the last generation to not routinely do things like confide in, rage at, and apologize to machines.
When we built our voice assistant into Boomerang for iPhone, one of the things we thought a lot about was fallback behavior when we only got 95% of the way there.If you ask Boomerang to reschedule your meeting with Fred, but the voice recognition mishears “Fred” we’ll still draft a reasonable email for you and let you type in Fred in the To field. If you have a similar hiccup with Siri or Google Assistant, you won’t get reasonable behavior at all.From working with the technology over the past few months, it doesn’t feel like we’re very close to getting to 105%. But we can get a lot of the way there by making fallback behavior still useful enough to save a lot of button clicks.Edit: forgot to link the app! https://itunes.apple.com/us…
Following @sethgodin:disqus ‘s comment, they are not great because we can’t really TRUST them. They are fairly limited in supporting real world tasks and today they are just a “shortcut” to buttons on your phone (i.e. “play music”, “Directions to YYY”, “Remind me to ZZZ”)
I think it’s still very early days and voice assistantance/automation is going to follow a much longer adoption cycle than other recent tech. The “problems” it solves are similar to a remote control in that it only automates simple tasks at the beginning of it’s development but as complexity increases it will be unimaginable to live without it. This is especially true with home automation.
Weirdly owning an Apple Watch Series 0 has lead me to use Siri more, mostly with that. If I use the Siri feature to give it a command the watch will buffer the command in its world of siri and then execute it on (ever so slowly, the bain of having the 1st gen product) without problem. Tapping around the watch is too slow. On the phone/IPad I’ll use it to set timers but I find with the Siri setup saying “Hey, Siri” does not manage to be super reliable. Tried a HomePod in an Apple Store and it clearly fixes this but now sure I’m excited enough to pay that price point.I would say the main reason I do not use voice interaction more is I spend most of my day in an open office space or on public transit. So there is no privacy in these interactions and they are likely to annoy people around me.
We have 6 echos and one Google home. The killer app was integration into Phillips hue lights So . I use it all day longLights News Music Cooking timers Weather Traffic Shopping lists Breaking news notifications Making phone calls Sending text messages Playing gamesShoppingEtc!
Was un this camp until we got my 5 and 7 year olds have Echos in their rooms. They are amazing for them:Playing musicSetting an alarm10 minute reading timer after we tuck them upTurn down their light (the Hue portable light is fantastic).They call their Grandparents
not enough killer use cases, i play jeopardy on my amazon tap while riding bike in the am, pretty cool and get the juices flowing, siri for apple music in car (think its safer), if i had a faster watch id probably use siri on watch more as the tap ui is poorProbably can do a lot more multitasking in my home office, schedule appointments, check weather and scores.
Somewhere there is a correlation between serious noise pollution everywhere, open working spaces with no privacy and acoustical dysfunction, and voice command usage.
We have four ‘Alexa’ units in the house controlling lights and more. I was not a fan prior to winning a unit an industry event (Home Automation Seminar), and found it fun to play with.Once we added more for intercom, messaging, alarm, ordering (from Amazon), distributed music/radio – they became almost indispensable.I listen to my 30 – 45 min of News flash sources every morning, call folks when I need (without having to fumble for a phone or wash my hands in the middle of cooking, etc.it does take some getting used to, Alexa wants to hear clearly enunciated words. Just like when we had to practice the palm pilot graffiti, so too one has to practice a slower more deliberate speech.In short it beats the heck outta ‘Gesture’ based systems.
The hands-free use case I find myself using most often is to listen to a radio station, or music from my library, while doing something around the house, mostly in the kitchen. The hands-free aspect makes a lot of sense when your hands are doing something else; and the Home is a better “internet radio” than a traditional radio because, at least where we live (surrounded by tall buildings) the AM and FM reception are poor.There are some real privacy concerns, though, that aren’t addressed. The Home (and its competitors) do in fact listen to everything going on; they couldn’t work any other way. We’ve learned the hard way that Facebook’s appetite for innovation has outrun its commitment to protecting user privacy. Are we going to learn the same about Google, Apple or Amazon as they build out the capabilities of these devices?
We have two Alexas in our home and Siri on a couple of phones (mine and my wife’s). The latter never get used. The former are used often — by our 12- and 9-year-old children. My totally unscientific conclusion is, if we’re representative of anything close to the norm, the true value of voice assistants will be realized not by my generation, but by my kids’. For them it will be second nature as they build the muscle memory from a much younger age. My muscle memory still compels me to call the TV remote the “clicker,” even though TV remotes have not clicked for decades!
i think that is right
This is my experience, as well. We bought a Mycroft assistant, and my kids (8 and 10) immediately wanted to use it to do everything. The only reason they’ve stopped is that the functionality is still a little rough around the edges.I think, once the experience meets the expectation, voice commanding will be a big part of our kids’ lives. Some things you just know are going to happen.
voice commanding will be a big part of our kids’ livesTo my point in another comment kids are definitely less worried (in that use case at least) what others think of what they are saying. It’s a millennial thing.I was shocked recently that my step daughter, as I was driving her somewhere, told me about some girl in gym class ‘that doesn’t wear a bra’. Wow that took me by surprise. Later when I told the story to my wife she relayed that she had told her about some girl ‘that ______’. Won’t say the word but figure it out out. (Hint: First letter ‘m’). [1]I attributed this to not being judgemental parents. (Like my parents were). The idea of saying things like this to parents is off the bell curve the way I was raised. But my stepdaughter felt comfortable saying this (and more from what I heard from my wife). That’s actually a good thing. Means they will be less likely to keep secrets. And this is not because we are ‘cool’ parents or anything like that or their ‘friends’. We aren’t.[1] I once quizzed my stepkids on what ‘bad words’ they knew. It was a contest. One said one word then the other had to say another word. They offered it up as if I was asking for examples of their favorite foods. Can’t imagine my parents doing that or feeling comfortable if they had asked me. Don’t think I ever even heard them use 4 letter words actually. (At least not in english). They covered just about everything.
I do not use any of them, and feel like maybe there’s an “early adopter” camp for privacy.It is becoming abundantly clear that we can’t trust that these services won’t scoop up other information about our home and work lives, no matter what we’re being told. Or how we use the privacy settings.
Our Christmas Alexa is still sitting in its box for this reason. I’m probably going to go ahead and set it up, but I feel *much* more comfortable with the Mycroft.
My wife bought me an Alexa but I unplug it every chance I can. Not only does it not work (it doesn’t understand why I’m saying per my other comment) I hate the fact that I’ve given large tech company the ability to listen in and record all of our phone conversations. I have no faith at all that any of them can and will to the right thing and protect our privacy. Rather I assume they are recording everything they can and selling it to the highest bidder.
Yes. Or that the information will be released in some kind of hack/leak…
Or, perhaps most likely, some gray area between an explicit sale and a hack.
+1000
My 5 year old spends at least 30 minutes every day talking to Alexa. It is his source of education and entertainment. He asks her for NBA scores, checks spellings, plays music, asks her to laugh, etc. When his mom was out of town a few days back, he actually asked Alexa when she would come back.Maybe he will grow out of it in a couple of years, but she is such a big part of his life that i wonder if voice is the default computing platform for the Alexa generation (born after 2012). The voice assistants will undoubtedly get smarter and better over time and these kids will grow with her.I am long on voice assistants. As adults, it is too much change for us to use it, but for kids, it is a natural way to engage.Now, do they impinge on privacy, etc.? Sure and I hate them for that, but my sons really do not seem to care.
my 8 year old is using the dictate function on an old iphone we gave her to use around the house for emailing her cousins and extended family, to do research and for simple asks. Definitely a generational thing in our house.
A friend told me he found that his child was ‘abusive’ to their voice assistant, when they didn’t understand them or got the context of the question wrong. He felt the need to intervene and teach the child to be more respectful. An unintended consequence I hadn’t expected to hear about, but which makes sense. It would be interesting to see how much abusive language, Siri, Alexa and Google get.
I have a friend that said the same thing. You can see it in yourself. At least I can. Think about how pissed you get when you are trying to do something on the phone and you don’t have signal or it doesn’t work. I’ve caught myself cursing at it when I can’t pull up a plane ticket, etc. I’ll admit to having cracked a screen slamming it when I couldn’t pull up a conference call info on the road.I mean think about that……internet, email, cell phone, Skype conference call. It doesn’t work halfway across the world and I slam the phone. That is serious behavior change since I started business travel in the 1980’s
I definitely swear at my alexa from time to time
It’s also third party apps.
For most people its nice-to-have a voice assistrant, but for elderly and people with disabilities its a life changer, and hopefully many apps and use cases will be developped around that. Similar to @stevebrewer:disqus comments re the use with kids.
What if voice was the “only” option? Then we would have no choice but to use it, assuming it would work. I’m hearing there’s a new generation of voice processing technology that is much better than the current state we’re in.What scares a bit is when a service operator tells you “you’ve been voice print authenticated” – how can they be so sure about that, when voice processing is not so sure yet. Is Voiceprint as accurate as fingerprint?
I can’t recall ever using the one on my phone.I use the Echo a lot though, though mostly for a few narrow categories – controlling the lights, setting kitchen timers, and playing music, and asking the weather forecast.There’s a niche for is, but I think it’ll always be just a niche. It’s the most useful when your hands aren’t free (like when cooking or driving). It’s the least useful when your hands are holding a device anyway, like when you’re looking at your phone – you may as well input what you want with your hands, which will often be faster and more accurate (and quieter).
Per earlier comment and you’ve seen it in action, Android speech-to-text is incredible for email. I used to dictate a lot in my young lawyer days (yes, my laziness goes back a long way back) so I’m familiar with all the syntax of how to insert punctuation, etc. Siri is getting better but from a productivity standpoint, voice on Gmail is incredible
These are early days, and indeed many of these are in use as “gadgets”, meaning they are hardly being used. However, there is a divide growing, and we see a subset of owners who now own three or more of the units and are using them beyond playing music and asking the weather. As well, not all interactions are ideal simply by voice, so we can expect development merging voice with visual results both video and text to get to the kinds of results people want to use. In all of this, the fact is that AI while exploding on the scene is in very early stages so much of the “intelligence” people expect when talking is in fact illusion. This, like all other technology will continue to improve and evolve rapidly. People like the idea of talking to these things right now, and I expect this will be a strong, but not the only interface. Looking forward to see where this goes.
Hearing the weather by voice is an anachronism. It’s like watching the 11pm news to find out the weather or reading the WSJ for stock pricing (they are still there amazingly). . You can typically get that info much higher quality by parsing visual info on your cell phone or laptop (which the weather apps have not actually made easier they’ve made harder). Besides it’s never accurate to begin with most of the time unless there is an outlier event that has greater certainty.
Someone needs to invent a good text-to-voice translator that can read text aloud in my headphones
I don’t want to read anything anymore. I just want to see pictures and hear stuff on my headphones
They don’t work (for me at least).I don’t use any voice assistants because they understand what I’m saying >50% of the time. I moved around when I was very young between the NE and midwest when I young and learning to speak. I think anyone who knows me would agree I have a pretty neutral American accent. But these things don’t understand me even when I’m in a quiet room speaking slowly. Like at all. And the difference between “Fifth” and “Sixth” which pops up a lot of NYC? Fuhgeddaboudit.
Probably most of them. Its like self driving cars or AI, they’re just not quite good enough. Plus, as I’ve said before, I have enough people yapping at me. The last thing I want is my machines to talk to me, especially unsolicited.
I personally do not use voice control for the same reason I use texting rather than voice calls. It’s more discreet and less invasive to others around me to use my hands.
Amazon’s Alexa music has been a game changer for me as far as listening to music for short bursts. A bit of a challenge for pulllibg up different versions of an opera, e.g. mozart’s the abduction.There are soooo many ways to improve Alexa. Remind me so much of desktop to mobile. They are related but are cousins not twins
I don’t like giving mouse clicks to a computer. So, I certainly don’t want to give voice commands to a computer. No way.The user interface to computers that I use and that I develop for my own tools is based on command lines. Why? Because command lines are easy to use and to program and easily make the automation cumulative where one such program can call another one. E.g., if some program X can do something, then program Y can call program X and have X do its work 100 times. In strong contrast, graphical user interfaces (GUIs) are a pain to drive with software; so, to do something 100 times, have to do all that clicking 100 times — bummer. So, for the tools I develop, I don’t use GUIs. For a voice interface, to drive that with software would be worse: I’d have to program the sounds!!! Ah, I’d drag out the Nyquist sampling theorem again!Well, I don’t like to use the Xerox PARC, Apple, Microsoft, etc. GUIs. So I certainly don’t want to use a voice interface.At one point circumstances pushed me to rush out and just buy a computer, and I got an HP laptop with Windows 10.Well, it appears that Windows 10 has voice to text functionality. Okay: Next time I move, of course I will need to pack lots of books, computer cables, tools, kitchen equipment, sheet music, CDs, DVDs, VCR tapes, office supplies, etc. For that packing, I want for each packing box a list of what it contains. That is, I want an inventory list. Actually, I do that now: For my computer work now, I end up with various items, e.g., spare cables, thermal paste, spare fans, various adapter cards, etc., and I pack this stuff in file folder boxes and, then, keep in a file on my computer a nice index of what is in each such box.So, maybe the next time I move and have to pack everything, the voice to text software of the HP laptop would be good for that entering the inventory list: As I pack a box, I would speak into the voice to text software as a way to enter the list of contents of the box. Then, soon, maybe right away, I’d use my favorite tool, a programmable text editor, to clean up, correct, etc. the results of the voice to text. So, the voice to text on the HP laptop with Windows 10 might be an easier way to get inventory lists. Maybe.With my startup, it’s not at all important or even worth any time to keep up on the apps, gadgets, etc. of mobile devices, devices to talk to, etc.I can see why some people can make good use of a portable phone. Otherwise, I don’t see much utility or future in mobile computing. I know; I know; Apple has made a lot of money with the iPhone, etc., but I don’t have a smartphone and don’t want one.My work doesn’t need much traveling, so I don’t do much traveling. I used to travel — used up lots of gasoline, rubber tires, time, money, energy, and effort, all for next to nothing of value. Since I don’t travel much, I don’t much need mobile devices. For me, a smartphone would be a really big step down for me, my work, etc.I know; I know: When I’m out in public I’m supposed constantly to be using the latest apps on the latest iPhone, with the latest ear pods, etc., look really busy, look like I’m in the middle of really, can’t delay, super important big deals, make a big show. No thanks. Sorry Apple: I don’t want an iPhone. And I don’t WANT a voice user interface; if I had one, I’d see that going for command lines I could easily program would be a really big step forward!!!My guess is that lots of other people will see that several hours a week traveling apps, voice interfaces, etc. are big wastes or just silly and, thus, work to quit doing that stuff.A big reason to avoid the latest fad stuff is the acronym “AI” commonly stuffed in there: To me, AI is so far 99% hype and a fad with nearly no utility at all.
>The user interface to computers that I use and that I develop for my own tools is based on command lines. Why? Because command lines are easy to use and to program and easily make the automation cumulative where one such program can call another one. E.g., if some program X can do something, then program Y can call program X and have X do its work 100 times. In strong contrast, graphical user interfaces (GUIs) are a pain to drive with software; so, to do something 100 times, have to do all that clicking 100 times — bummer.Yes. GUI’s definitely have their place, but automating batch tasks via GUIs is not a very good method, except maybe for some specialized tasks via certain made-for-the-purpose GUIs (or tools like AutoHotKey or AutoIt), whereas command-line tools both permit and facilitate automation, composition via pipelines, and scripting.For anyone interested in designing and writing their own command-line tools, here are a couple of related posts from my blog that may be useful:An article I wrote for IBM developerWorks – a tutorial on Developing a Linux command-line utility in C:https://jugad2.blogspot.in/…It has both a case study about a specific command-line utility that I wrote for a client, as well as some general principles and techniques for writing command line tools.The Bentley-Knuth problem and solutions:https://jugad2.blogspot.in/…Excerpt from the latter post above:[ I recently saw this post about an interesting programming problem on the Web (apparently initially posed by Jon Bentley to Donald Knuth.For lack of a better term (and also because the name is somewhat memorable), I’m calling it the Bentley-Knuth problem. ]In my post I showed solutions to the problem, which was initially posted about here:http://www.leancrew.com/all…in both shell script and Python.
You explained the interface issue better than I did.For a revision, and more succinctly, (i) my computer, (ii) the software I find easy to write, and (iii) I all agree — it’s easy both to send and receive characters. So, that’s the convenient, productive interface paradigm: command lines of characters.Yes, I do use a mouse: (i) For a lot of the little windows put up by Windows, a mouse is necessary. And (ii) I use a mouse when Web browsing. Still, for both (i) and (ii), lots of people and the industry more generally have noticed that (A) often it would be good to automate that work and (B) writing software to do the work of a mouse is not easy. E.g., quickly server farms want to be able to boot a computer over a LAN and install and configure software, on a few thousand computers, all with just one command. And there are such means, including IIRC from Microsoft. And, for getting data from Web pages, called “screen scraping”, that, too, is not always easy, really should involve running the JavaScript of the page, but no doubt there are now relatively good means developed. E.g., some rumors are that there are some good “packages” for Python.But recently as I have, yet again, been installing software, I want to scream that I want text files and command lines. I used to scream, but I got no progress and only a sore throat, so I no longer scream. For installing and setting options on Windows, just for installing and setting options of Microsoft’s Office, I want text files and command lines. E.g., for a Web browser, now there are a lot of options to set and, in particular, some lists of URLs that can/cannot use cookies or run JavaScript. Well, I want those options and lists in text files. I’ve typed in that data too many times.More generally, to heck with AI and, instead, here is a much more obvious and productive future: Have humans managing computers managing computers, …, several levels deep managing computers doing the work. So, we need a universal means of doing that. And I’d guess the means would be in text. Actually, I’m partly correct: At one time I got fairly deep into the OSI/ISO CMIS/P ASN.1 (a lot of gibberish — international standards organization, common management information system, abstract syntax notation .1 gibberish) object oriented data definition standards. Some such things should be part of the needed computer to computer communications for all the system managing in the hierarchy I suggested. No doubt XML can have a role, etc. When I was at FedEx and wrote the software for scheduling the fleet, I designed the input data to be a file of text with key-type-value triples with nested block structure with scope of names. And I wrote the corresponding code.A voice interface? A mouse is bad enough. Voice would be much worse!Thanks for the references.But you are deeper into this stuff than I am.Gee, I just read your Bentley-Knuth problem!Gads, as some months ago I said to a technical guy at venture firm A16Z, “I’ve done that a gazillion times.”So, my favorite text editor and my favorite tool is KEdit from Mansfield Software in Connecticut. It has an elegant macro language, Kexx, KEdit’s version of Mike Cowlishaw’s elegant Rexx.So, I have about 200 macros for KEdit and write new ones right a long. One I have in draft, waiting until I complete rebuilding from my computer that quit, creates a standard file system tree name from any of the several formats of the same data used by various pieces of software I run.But for the Bentley-Knuth (B-K) problem, one tool is my macro TOKENIZE: It defines a “lexical scan” token and then, for a text file, e.g., input to the B-K problem, parses the file to have one token per line. With that, sort the lines of the file — KEdit is good at that. Then run another little macro I have that counts duplicates. Then run another little macro I have that builds a simple, character mode histogram of the frequencies. And I have another macro, delete the duplicates.Then the guy at A16Z asked, how would handle a large file. Well, for a solid answer, sure, one that will work well over the horizon, is just to sort the file, that is, for a file of n words, run in time O(n ln(n)) then read the file in time O(n). So, the whole solution is just O(n ln(n)).But in some cases, maybe usual cases in some contexts, have a heap, as in heap sort. Give the words in the file to the heap one word at a time, and have the heap add the words to the heap and keep track of the count of duplicates. Gee, for English, space for 80,000 or so words should be enough, twice that for words from text from hackers who didn’t know how to spell or even use a spell checker! Uh, right, my favorite spell checker is a little command line thing with a console window character interface! But I have the spell checker code, darned clever, and could call it and automate a lot of spell checking. So, we have a heap with space for ballpark 80,000 words. So, write some code to turn a heap into a priority queue, that is, the words are the keys in the heap. So, O(n) for reading the input file and something O( n k), for usually some small k, for the heap insertion, that is, again just O(n), and done! Sure, a heap is a bit poor on memory locality of reference which would make its use of the usual virtual memory inefficient — although now 80,000 words would fit in a tiny fraction of the real main memories. But, just in case, there’s a paper for how to tweak the heap data structure and algorithm to get good locality of reference and good performance in virtual memory. There is also the extendable hashing, hashing but with some extensions to have no collisions, and that could also be used for the B-K problem.My approach to software is mostly just pragmatic — I’m trying to get my startup work done. I believe my startup has some advantages, but the strongest is the core applied math. The computing for my startup is relatively simple. So, now in computing, I just want enough for my startup. So, my computing is routine and not one of the advantages or a barrier to entry.So, right, with just a pragmatic approach to software, how’d I get into using the heap data structure and algorithm for a priority queue? Well, your mission, and you have to accept it, or at least I did, is to look at a list of numbers, maybe 100 million, quickly now, right away, all the numbers in main memory, maybe several times a second, a different list each time, and report the 20 or so largest. So, yup, I used a heap for the solution. So, I wrote some quite careful heap code.Ah, my usual B-K problem is stopping at Burger King for two Whoppers with french fries, onion rings, and the largest diet soda for lunch!! Haven’t done that for a long time!
You’re welcome. It’s because I have been working on this stuff for some time, as well as have an interest in it.I had played around with speech recognition on Windows using Python some Python libraries a while ago. It did work, partially, There were some issues, it did not always recognize words correctly. But it is a hard problem of course. Not tried it out on the Windows 10 version you mention.
Not to be overly cynical, but “dust” probably isn’t the only thing those Alexas and Google Homes are collecting while they sit within earshot of most of your intimate conversations.Otherwise, echoing the comments of @stevebrewer:disqus and others — the younger children in our house are the most engaged with our digital assistants. We had a similar experience a generation ago with the first home computers.
We have an Echo in the kitchen and a Sonos One in each of the rooms that our children share. We also have an Echo Dot in our master bathroom. The whole family loves using Alexa to play music, ask questions, and control things. It always surprises me that our children prefer to ask a question via voice (“What’s the score of the Bulls game?”), vs grabbing one of our phones and using Google. I think this will continue to play out as voice assistants grow in capability and availability on the go and at work. My kids feel like a great litmus here.
We have Alexa/Siri/google, but the only functions we regularly use are asking the weather and asking Alexa to play a song for us.I think the biggest problem is not the voice recognition but the inherent bandwidth limitations via voice + audio. Other than the above simple requests, it’s easier/faster to look at and click links/buttons on your phone screen than to listen to a voice response and ask follow up questions.
It will go into the history books as the most successful (state-sponsored?) surveillance tool of all times.
Aren’t you reading this the wrong way – i.e. 40% of people *are* saying they’re using voice assistants regularly, which would be an amazing success. I suspect however that the phrasing of the questions (options 2, 3 & 4 aren’t orthogonal to 1) is leading to a higher “am using” than actual.
This was my thought as well. It would be great to know if the growth rate is accelerating — I suspect it is.
40% of people *are* saying they using voice assistants regularlyAs Fred said the results are people on a) twitter and b) follow Fred. This isn’t even close to representative of ‘people’. My guess is that the number is drastically lower than ‘40% use regularly’. And there is almost certainly a bias where people even anonymously will answer with a reply that makes them think they are advanced in some way over the other population (further exacerbated by ‘a’ and ‘b’).Typical ‘stick to the knitting’ with Twitter staying with only 4 choices btw.
40% of the people don’t know what Twitter is. That is the echo chamber.
In the US, under 40 years old–I seriously doubt it.
50% of the population is over 40.
With Trump I simply don’t think there are many who haven’t heard of Twitter is my point.
Yes, I will have to concede that point.
Maybe part of the reason that it hasn’t taken off and isn’t ubiquitous is because people are not comfortable with others hearing their business. [1] I don’t need people around me (even people in my own family) knowing what I am doing or thinking or wanting to know. Not that it’s any big secret if they found out. But it does not feel comfortable. Maybe this stems from having judgemental parents weighing in on all of your actions.Some other reasons:- Social proof not firmly established. [2] Does anyone remember how strange it felt the first few times that you used hands free and had a long conversation in a car when you were driving by yourself? After time though enough people were doing it (social proof) and of course you just got enough value that you overcame that anxiety.- Per Matt Zagaja’s point there are places that it simply isn’t practical to use voice. And as such it isn’t going to ever be used 100% of the time. So it’s not reinforced in the same way that things that are 99.9% acceptable are.[1] And unlike talking on a cell phone in public the use case isn’t as strong and there is an alternative right at your fingertips.[2] How many of us are old enough to remember that you’d never go out to dinner at a restaurant dressed like a schlep? In a T-shirt? In dungarees?
I think you are right here.In an age where people say “there is no privacy” people want some.Now this might be me projecting but I literally hate people looking over my shoulder when I am on the computer. This includes even my wife.Now she could easily (well not really because she is horrible with them) just jump on when I go to the bathroom or go get something to eat. I don’t lock.But I really hate people looking over my shoulder.And people project to kids, but do my kids want me seeing their Instagram page?
I am pretty much the same way. But did you ever notice that when a workman does work for you they typically like when you watch what they do? Or a repairman. That is when someone takes interest in their work or asks questions. That has been my experience. Ditto if I do something to help my wife on the computer. I hate when she dumps it into my lap and then walks away. For that I like her to watch and to teach her.
I know this is off topic, but please tell me you’d never invest in something like LegalFling?
It’s brilliant marketing. They can branch into other areas and the story can be that the original idea (did/did not) work so now they are doing something else (as well). What a way to get attention. This is mainstream bait.Concept of course is kind of a buzz kill. Ditto for the name.http://www.dailymail.co.uk/…Remember the movie with Woody Harelson, Demi Moore and Robert Redford, Indecent Proposal?http://www.imdb.com/title/t…”The night will come and go but the money will last a lifetime”.
Starting to use Siri for random trivia. Otherwise I feel like a nitwit talking overly loud into my own machine. My daughters will no doubt laugh at me when they’re older
Occasionally Google Assistant. when s/he becomes a conversationalist i will talk more.Peter Sagan. Oh YEEEEES!!!Best wishes to Michael Goolaerts for a full recovery. Awful.Update;He died. Shocking. The only consolation for his family and friends is that he was doing what he wanted to be doing with his life when it happened.
MY Borg friends and I do not have any problems
We have a few Alexa’s and use them often. But not for a lot of tasks. Timers in the kitchen. Music. Weather. And lights.If your house works for controlling lights via voice (eg simple lighting set up without the need to specify things such as ‘living room lamp that’s behind the couch’) it’s delightful.We also use Amazon fresh for groceries. Adding things to a grocery list via voice, that’s actually our shopping cart, is delightful.We don’t use it at all beyond these 3-4 things. But for these things it’s been embraced by myself, my skeptical wife, and our 3 y/o.
I believe the relatively low usage is both cultural and generational. People under 25 use voice more frequently (according to a study but sorry, no link). Also, in certain countries, e.g. in China, people tend to use voice dictation (or sending voice directly) rather than typing on their phones when using messaging apps. I presume they use voice assistant more frequently too.
Before you allow artificial intelligence into your life you must know it’s source of truth. I was all set on buying Alexa a year ago. But then I learned that it’s answers reflected the progressive views of those on the West coast. Want an example? Mohammed is a revered religious leader. Jesus is a fictional leader of a made up religion.Now maybe those views reflect your own and you have no problem. So I’d ask you what if Alexa’s views reflected those of Donald Trump? Your kids would ask it questions and get the Trump world view. Imagine your surprise when your kids quote Trump back to you at the dinner table? What if you purchased an inexpensive Alexa clone and found it reflected the views of ISIS? There’s a huge threat if we get this wrong and we’re already headed on the wrong path.
That’s why should an Alexa show up on my front porch, and the candidates are:Use it for drop kicking practice.Use it for rifle target practice.Toss it up and use it for shotgun practice.See how long it would still work left out in the rain, snow, and heat.And the winner is, may I have the envelope, please [drum roll]. And the winner is none of the above because of NSFW!!!Instead of the above, I could try to teach it to do something useful, but that would be like a human trying to teach a pig to sing: It’s hopeless, frustrates the human, and irritates the pig.
For the record that is not what is says. When I heard that I had a friend check that day. He probably used Simon says feature.That does not mean your point is wrong.
A year ago there were a number of videos making the same points I did. Today this is the only one that I found in a quick search. These guys go pretty deep into the weeds overall but they provide some key examples. https://www.youtube.com/wat…If Amazon has updated Alexa removing political opinions in favor of encyclopedia level facts then I haven’t heard it. I’ve got a friend that I can check with who has an Alexa on some of these questions.
Seriously???? I know I am in the top 5% of “conservatives” on this board. When people say “fake news” this is what they mean. Crowder? Really??? Come on. Fox fired the guy. He is the only one that “got it on tape” Yeah, right.I am not going to argue if Alexa is “liberal” or “conservative” What I am going to argue is people distorting facts to get ratings, and yes you will correctly say but so does CNN for instance. But that has nothing to do with facts. Both sides have gotten so distorted they feel that they can just repeat made shit up and not check it.Just repeating stuff without checking. I mean how hard is this to just check? How damn hard. Did you? You must know somebody that owns one. But no, everybody just repeats. When my wife told me I immediately called a friend. It repeats the wikipedia entry. FYI.Yes, it is both sides. But damn this gets old. I’m with Fred. Mainstream news on both sides is just a sham.I had somebody cite an article here saying “you just walk into a Walmart and walk out with a hand gun”. Here are 20 things that are harder to buy. Even CBS news repeated and had to retract that. Ok….how hard is this to check? First Walmart doesn’t sell handguns. Hasn’t for over a quarter of a century. Second if you find a place at minimum you will have to get a Federal Background check. Depending on state is how long you have to wait, among other requirements. In Florida it is 3 to 5 days. But no. People report you go in and buy one in 15 minutes. I’ll argue about gun control, but not with somebody that spews bullshit. Again how damn hard to check?? How can CBS write that?? Because they were too stupid and lazy to check. They have an agenda. Hey, so and so said that, there is my source.Seriously, how long can that take to just check? You don’t even have to get off your fat ass, you can just call a Walmart.
I live in a small 550 sq foot studio on my own, and a google home is the perfect companion:1. I ask it for the weather when I get ready in the morning2. I ask it to play soothing music or set an alarm when I’m lying in bed3. I tell it to add “TODO” tasks whenever something pops into my mind4. I use it to call my friends & family while I’m doing things around the house5. I play podcasts and audibooks while I’m cleaning, cooking or stretching at home. I can stop/pause/rewind and modulate the sound hands free.From personal experience I can say that voice assistants are great for those living on their own, in a small space, with a never ending queue of content.
One thing I will note about voice assistants is that they are being ‘pushed’ on us. In the past things have taken hold have been more ‘pull’ and sell themselves. My cherry picked examples in no particular order and off the top:a) Cable tvb) Tvc) Telephonesd) Fax machines e) Car phones (cell phones)f) Ezpassg) Sexh) Drugsi) Food addiction/restaurants and diningj) Fedexk) Gamblingl) Pornm) Radion) Sportso) Computer gamesp) Air travel/Airplanesq) The iphoner) Gunss) Rock and Rollt) MoviesForgetting the actual adoption curve none of the above were a hard sell.While everything has to be sold and sure marketing money was spent on some of the above overwhelmingly the market and people immediately liked in a way that it wasn’t a hard sell. Meaning that ‘s’ took hold without a hard sell but Opera simply isn’t appealing in a way that any amount of marketing or effort will ever make it popular.
Google assistant’s voice to text is incredible. I should exclusively SMS using it but still very rarely remember to.
My five kids regularly use Alexa to get weather information when getting dressed. Absolutely saves us a lot of hassle on busy mornings as we think our oldest at 13 is still too young for a smart phone.Wife and I not so much except for Google maps directions which is simple enough for me to remember the command for.
I’m not sure who you surveyed but I’d encourage you to ask the question of the under 25 generation. My observation of 13-20 year olds (albeit a small sample) is that they are constantly talking to their phones, are avid users of their alexa devices and view typing as “old school”. Personally, I use Siri more and more for dictation of emails on the go and am a regular Alexa user – though haven’t really figured out what to use my Echo for beyond a substitute for the radio and some basic searches. My wife uses it much more extensively.
Collecting dust & data 🙂
I think that’s the reason that the Google Home ads on billboards tout ‘setting a timer’ as the lead use case, or at least the one they think will resonate with most people.
I literally read this minutes after installing a third Amazon Echo device (along with a Google Home), which I am using to listen to radio (also use for podcasts and music) while working. (I got it, and an Echo Dot from my bank free of charge in exchange for using a credit card.) We use it for our shopping list (easier while cooking or doing dishes, and hands covered in whatever), sometimes to send a message to each other, occasionally to tell it to play a program from our FireTV stick which, plugged into our Samsung smart TV, will turn the TV on and flip to that show on Netflix, Amazon Prime, etc. Mostly, though, to lay audio — music, podcasts, NPR radio stations, etc. I’ve experimented with a morning news brief, and for my dose of Japanese news, but find it uneven for that. I’d love to be able to have it read me articles or books — which is something for which i use my iPhone “accessibility” function quite a lot.
Not really an assistant but I’ve been shocked by how useful voice control is on my Xfinity remote (Just switched to the service). Voice recognition is better than Siri or Alexa and the search results on the TV are fantastic.
For me I’d use them more but usually I’m at work in a very open working area so it means I’d have to get up, go to a free room (if I can find one) and speak my request … it’s less painful to just type it in usuallyWe do use Alex a bit for home but so far its limited- Can’t control the TV unless I spend a bunch of time figuring it out- Too painful to get it working well with my smart home controller — partly due to young kids sapping most of my time- Not obvious how it hooks into my google or exchange calendar yet- None of our appliances are fancy enough to hook up- Can’t auto-start my car yet (although this could be a mixed blessing). I’m sure there is a way but I don’t have time to look into it …There are a ton of applications here but I think its still early days
4 silos; if you count erstwhile Cortana. None work well separately. Balkanization. Enough said.
Surprised you don’t use Alexa to control your Sonos