Hacker News Reader

Comment by chatmasta | Hacker News Reader

Chatbots were never the “next big thing” by any measurable metric other than artificial hype created by the companies trying to create a new platform for extracting revenue and data from users.

For this argument to make any sense, there would need to be data that showed growing usage of chatbots, followed by a plateau and then a drop off. I’m not sure what that data looks like, but I’m pretty sure there was never a high growth phase where users were actually interacting with the chatbots. It was all hype created by Facebook and a few idiotic VCs who wasted money on what they thought would be the “next App Store.”

I, and many others, were saying around the time of this hype that chatbots would never become a category defining product, simply due to inherent usability flaws in their design that have been discussed ad nauseum.

The only people surprised that chatbots failed to become the “next big thing” are the people who mistakenly thought they ever would be. This assumption was never grounded in any real data of user desires or real problems. Chatbots were then, and are now, a solution looking for a problem. I’m not surprised at all.

View on HN

I fully agree with you. I also dismissed Google Glass and other hypes. I feel smart about that.

On the other hand, if someone would have pitched Facebook to me, I would have also dismissed it. I keep that mind to stay humble.

For any trend/hype there will always be people dismissing and praising it. Thus, if it is successful or not, there will be people who say "I told you so" afterwards.

It does not matter what a comment on HN says. You personally have to decide, if you want to speculate/invest in some hype or not. You have to decide which new technology you learn and which ones you ignore. In hindsight, it would have been great to learn machine learning five years ago.

Eh, Facebook was a natural evolution of social networks going back to MySpace, Friendstr, SixDegrees. Each one fixing mistakes of the past and becoming the next big thing until their own mistakes caught up to them.

The one that I dismissed was Twitter. In fact I thought it was a joke when I first heard about it, something making fun of how long winded boring blog posts on services like Livejournal tended to be. I also thought Vine would be doomed to fail.

I also thought twitter was a joke until I realized most Americans are too lazy to write a blog post, or even read one.

“No one ever went broke underestimating the intelligence of the American public.” - Mencken (paraphrased)

http://www.thisdayinquotes.com/2011/09/no-one-ever-went-brok...

How many blog posts are being read and written by the French? The “Americans are too lazy” trope is just ridiculous.

I find that many comment like gp online are equally valid with the word “Americans” replaced with “people”. It irks me as well.

Yes, but it does preserve some uncertainty about the rest of the world. The author may only feel comfortable enough to make a comment about Americans.

This comic kind of captures what you two are getting at, though I think you are applying the idea a little more generally.

https://xkcd.com/1138/

"Yo" did die (or at least appears zombified, with 32 Patreon supporters). One bit appears to be too few.

Twitter is the weird thing it is because it began with an SMS interface (hence 140 characters). It took off originally because you didn't need a smart phone or a data plan or a PC to use it.

I thought it took off because they managed to get celebrities on it.

This. Ashton Kutcher and Demi Moore were amongst the first celebrities on Twitter to actually post the content themselves rather than by a manager/PR agent. This got a lot of media attention and many other celebs followed shortly after.

I still remember finding it strange to see how Ashton and Demi personally wrote short messages, since every form of communication from famous people at the time were formal statements released via the press.

That happened a couple of years after it initially took off. Celebrities only took to it after it had gained solid traction among Silicon Valley types and social media early adopters.

It’s not about being “lazy”. Most of the people I follow are both bloggers and podcasters. But not everything is not worthy of a blog post and I would hate for my RSS reader to be polluted with everything they tweet about (even if it is interesting.)

I never thought Uber nor Airbnb would be successful. Who invites strangers in their home or car. I was told as a kid to never get in a stranger's car ... now, as an adult, I am riding with strangers all the time.

You never take taxis? Airbnb is even less of a stretch since B&Bs were the original hotel.

Uber was not a surprise at all for me because I'd been complaining about the bloated and horrible taxicab system for many years. It was a market that was crying out for disruption. I remember in 2008 my local taxicab company had exactly two ways to hail a cab: phone or fax. They might or might not show up at the appointed time, and they'd certainly never send you any kind of updates. It cost upwards of $5 to step into the cab, and after that there were fees out the wazoo (per 1/6th mile fee, per 45 seconds fee, fee for having a bag, a fuel surcharge, a license surcharge, a regulatory recovery surcharge, and a couple others I forget). The cabs might have credit card readers but they were always broken so it was cash only. Drivers took weirdly roundabout backroads paths and always claimed it was fastest if you asked them about it, even though I knew the highway was not busy at 4am. Every single thing about the experience sucked. When they started crying later that Uber was killing their business my response was "good".

Except Uber didn't start as a taxi replacement company. If you only remember it as a taxi company it was already after it had started to become successful.

It was initially a carpooling app. The original pitch was that regular drivers would just drive to their own intended location and have an ability to see people on route who wanted to go the same way. Much harder to pick that idea out as a future billion dollar company.

What are you talking about? The original pitch was a black car service to fix the hailing problems of SF. It was even originally called UberCab. https://medium.com/@gc/the-beginning-of-uber-7fb17e544851

The original pitch was a black car service, but the legend that TK talks about, in Paris in the snowstorm of December 2008, was that he could ask “anyone there” to drive him, which is closer to carpooling.

(I was at that conference: everything went wrong, but it was mostly super cold. Cars were stuck in slippery snow, so someone suggesting the idea of carpooling sounds odd, but I suspect anyone there would have loved to wait out the traffic inside a heated car.)

I think you are talking about Lyft not Uber.

Yup. Looks like it was me that misremembered. My mistake.

>Uber was not a surprise at all for me...

When I moved to the US in 2003, the lack of something like Uber was a surprise.

Not in the smartphone-app sense, of course (those didn't exist yet), but in the sense of regular-Joes-making-a-buck-giving-a-ride sense.

Back in Ukraine, there was no taxicab monopoly. You could hail a ride on the street simply by raising your hand - and someone would pick you up pretty fast. Some people did it for a living, some would use the extra income on the way to/from work if you were going in the same direction (not unlikely if you are standing on a major street going to downtown, for example).

The fares were unregulated and negotiated on the spot, but there was an equilibrium point for every route. (Unless you were a tourist, in which case you'll be almost surely ripped off - nothing any local would care about though).

And, of course, there was no rating system, which took away the incentives for drivers to go out of their way to maintain a nice appearance. Don't like it, don't take the ride; the next car is will stop by a minute later for you, the next passenger will be around the corner for the driver.

Other than the lack of ratings and fixed pricing, this worked like Uber works now in other aspects.

What might be surprising is that central pricing and ratings really don't make ridesharing much different from what I described. It made the system more accepted in the US, but the end result is pretty much the same: I can hail a ride anywhere from people I don't know. The system stabilizes at a price point and expected level of service.

And there's a remarkable similarity in the downsides of this system. My father made some money giving rides back in Ukraine between jobs, and while the short-term revenue looked great (compared to average monthly salaries), so much of it went into gas and maintenance that the whole gig was hardly worth it - and it took more time than a job.

This is what Uber drivers are discovering themselves right now. The ones who took out loans for their shiny cars are especially screwed.

>Not in the smartphone-app sense, of course (those didn't exist yet), but in the sense of regular-Joes-making-a-buck-giving-a-ride sense.

These existed nationwide in the United States in the early days of the 20th century, until about WWII. In many cities they were called "jitneys." Some were organized enough to operate as simple, private, anonymous bus companies with fixed (but flexible) routes.

They started disappearing after WWII when most people outside the urban cores got a car, and municipal governments started taking over public transit.

These still exist in some places, but are mostly in ethnic minority communities and spread by word-of-mouth. (I know first-hand because I got hit head-on by one full of schoolchildren in the late 90's.)

I can’t speak to all of the US, but circa 2010 there was still a ton of people driving “private cars” in Brooklyn. You would negotiate up front with the driver and pay cash. My guess is 100% of those guys use Uber now.

True, but you wouldn't hail them from a random location in the city. You'd have to be in the right place to find them.

AirBnB is also less of a stretch because there were already pretty successful attempts paving the way for it.

I was using VRBO a decade ago for vacations. AirBnB is more distinguished by a slick UI than business model innovation.

I've never taken an Uber in the original sense of car pooling. It's always been as a cab.

Uber is still not a successful business model. It’s losing millions of dollars a quarter. Once they start charging enough not to have to subsidize rides and they can turn a profit and people still use them, then it can be called a success.

We still don’t know how price sensitive the average Uber customer is.

If I’m not mistaken in mature markets, like US, they are actually profitable. They lose money (a lot of it) to expand in new markets.

Before Twitter, people on AIM and similar services would use status messages to broadcast short little notes to their buddy list. I had the idea of turning that into a service in the early 2000s.

What surprised me was seeing a serious journalist like Wolf Blitzer regularly reading people's tweets on his prime time show.

The degree of Facebook (and Whatsapp and Messenger) amount of users is insane though. How many people possibly thought they would be at numbers or around 1.5B and 2.2B users?

This is an excellent point!

We should also keep in mind that chatbots are clearly a thing that people want; we're just using capitalism to test whether society, technology, etc. are "there" yet, so it's not entirely irrational for investment to flow into this area. (We might debate the scale of the investments, or we can just let the market determine that for us.)

The famous HN Dropbox comment!

Glass wasn't hype by any means, it just didn't take hold in the consumer market (which, who's to say Google didn't predicate that). It's been in use by massive enterprises for years now, from Boeing to GE to large healthcare organizations.

I wouldn't dismiss Facebook for the simple fact that it plays to something deeply human: to connect with other people.

But while it does that, it also plays to something even more deeply human: to watch other people from afar, and to replace complex, anxiety-ridden face-to-face interactions with easier, gentler text-based communication.

Google Glass and all the other hyped techs never struck at anything deeply human. VR is nice and all as a tech, but it's not something we've always done.

I would say it is still early days for VR. I was a skeptic at first till I got my own VR rig, now I'm convinced it's the future even if I don't know what time-frame that is going to be. As for Google glass it was just a little ahead of it's time and it's even gaining traction on the enterprise side.

It seems you and GP are confusing VR and AR.

VR, virtual reality, is a complete immersion in a virtual world. A 3D game, a 3D movie, a 3D Minecraft, you name it.

AR, augmented reality, is additional "augmented" information in the real world.

VR is nowhere NEAR ready for mass consumer market. There's 2 screens on the glasses with very low resolution. Because the glasses are very much near the eyes, pixels are very apparent. That's ugly. Furthermore, you need a big fat gaming rig with a 1000 EUR/USD graphics card for the current low resolution. And then there is the issue of the data transmission. Those huge wires (its not wireless yet, and wireless suffers from more lag). Finally, it is pricey.

AR from Google Glass flopped due to privacy concerns. It would aid users in their day to day life like a smartphone or smartwatch does. Its being used for this purpose in business settings (also due to its price) just like some other expensive tools by Microsoft such as Hololens and Surface Hub (84"). The user interface isn't quite there. For example I'd say eye tracking is useful. Some cars also get HUD in windscreen.

(YMMV but as a glass user who uses glasses which also work like sunglasses in the direct sunlight, I'd love to have a pimped glasses with AR.)

Anyway, in short, my conclusion, is:

VR is mostly fun, though some useful sims (e.g. flight trainers, but also things like 3D meetings would make Hyperloop and air travel redundant which saves time & environment) do use it as well. It is more of a niche than AR. AR is going to be part of our every day life in useful/productive ways. AR will be vastly more rampant, and it will be in use in a massive scale earlier than VR will be.

As a side thought imagine going on vacation in VR and it being just like the real thing cause you're on a robot. Your 5 senses get stimulated while you walk through Cairo. You smell the spice whilst you walk through the market on your robot which you control. Why actually still go on vacation then? I already have enough pleasure with Google Photos or the new Windows 10 login screen as it is. Going to see the Eifel Tower which you already saw hundreds of time on pictures is boring as is receiving that piece of metal on a postcard. Oh well.

Of course, there are wireless VR headsets. Check out the Oculus Go. I think this might be the start of VR going mainstream.

Interesting, didn't know that, but you'll need a hell of a rig to render two screen of like say 4k. And the quality isn't like having a 4k TV at 2 or 3 meters distance.

The requirement of a high-end GPU, and the current price of high-end GPUs thanks to the cryptocurrency mania, meddle with the adoption.

We're far from VR being mainstream; AR costs far less resources.

Yeah I think it will take another couple of years for VR to become mainstream (i.e. smartphone mainstream). But I think the Oculus Go is the beginning of the mainstream adoption (maybe like a Palm, and we need 1-2 years more for the "iPhone").

From my experience, VR is a huge gamechanger (a new paradigm that will enable a lot of new innovation) whereas AR is also cool but doesn't have that new paradigm potential (like e.g. the internet, smartphones etc. did).

"It seems you and GP are confusing VR and AR" I'm doing no such thing. I don't know where in my comment you got such an impression. If you bothered to read it you'll see I mention that I own a VR rig. I certainly know the difference. Also I would have agreed with your comment about VR being nowhere ready 2 years ago. While it's not quite ready for mass market primetime it's definitely at a stage where the average Joe can get a lot of mileage out of it (unlike AR which is mostly hype at the moment). I would say it's currently at the stage that PCs where at in the mid 80's. BTW you don't need a 'big fat gaming rig' and a dedicated graphics card - I don't know where you are but here in the US you can get the Oculus Go which is a standalone unit for less than $200. It's not the full VR experience but its good enough that I would consider buying it over a gaming console or even a tablet. I think you should have more experience with VR before you make grandiose statements knocking it publicly. As for AR I haven't really had a lot of opportunity to try it out (except for Glass). It may well overtake VR at some point but at the moment most AR projects are at the concept stage or just hype.

> If you bothered to read it

Oh I read your post, its against HN guidelines to assume and express someone else has not read your post or the content.

If you read the post of your GP (and that GP as well) you'll see they're casually mixing up AR and VR. You do the same, and I called you all out on the difference between the two. There's no need to feel offended about that.

You say you haven't had an opportunity to try AR, and mention Google Glass. The thing is, Google Glass is just one example of AR (a very known one) and also a very multi-functional one. The usability I had in mind is more subtle, more specific.

Examples: Layar has been out for ages (it adds AR on your smartphone screen from camera input e.g. adding complimentary digital info to a paper magazine or food you bought), if you have multiple smartphone cameras you use it, LIDAR uses it, Google Maps ecosystem as well (Street View for example). AI in general can use it. There's all kind of usages of AR such as this traffic light [1]. Not even mentioning about the experiment on how biking lanes are being lit up due to solar energy in the evening/night. Google Glass could've potentially done that all, but there's no reason that cannot exist in the future. It was just the wrong company to come first with the product. Google was also too early with it, just like Apple Newton was too early and GM was too early with the EV.

My first time using VR was on vacation in Spain as a child, using those red/green glasses we were watching a dinosaur movie. It was rather primitive, but immersive. I'm not anti; I just see it rather limited. I've read enough reviews about VR glasses and read reviews of VR games -including realistic screenshots- to know that the quality isn't there yet.

Long-term, VR will make more impact, but that's because it is so different whereas AR is rather complimentary. On the short term, AR is going to be more widespread and useful. I mean, I'd love a HUD on my windscreen telling me I am speeding. Like, of course.

[1] https://dutchreview.com/news/dutch/smart-crossing-dutch-inve...

>If you read the post of your GP (and that GP as well)

>Its against HN guidelines to assume and express someone else has not read your post or the content

Hmmm. Uh-huh well, if you're talking about limited AR such as in HUDs or Pokemon Go or glass, that's already here and it's cool and useful, but not very exciting (at least to me). No doubt AR will be a thing in everyday life. I don't see why people (myself included) talk about both technologies (AR & VR) as if its a zero-sum game and we'll only have one but not the other. No reason they both can't exist in the brave new future.

I wasn't implying you didn't read the posts (like you implied with "If you bothered to read it"). I asked the reader (e.g. you) to read the posts, as in reread them. That's a subtle difference. I didn't say reread but that's what I meant though. When I wrote that, I actually went to reread them myself to double check if I really misunderstood the context. Frankly, after rereading it, I came to the conclusion nope I did not though I might've better replied to your parent. Either way, the two were clearly being mixed up throughout this discussion.

I'd say, that when AR is "boring" that actually means it succeeds. It blends in without being annoying.

If you remember the time when desktops and mobile phones and smartphones become boring; that is when they started to gain traction among the common man; arguably, when they started to become good.

Take the traffic light example I mentioned earlier. If that's rolled out and seen as "boring", I'm happy with it. Because that means it works. Right now, it isn't yet boring, btw. It is seen as an innovation.

Same with innovations like electric vehicle and autonomous cars. Once they're "boring", their functionality is accepted. And we are talking about the "West"; it hasn't been adopted in poorer countries at all, not yet.

VR def. has the wow factor, I agree on that. Look at all the VR movies the past 30 years. AR is much less apparent, though its there as well.

Ultimately, for me games are just spielerei they might be fun (not so much for me anymore as I grow older); they have little to nothing to do with productivity. AR is going to be more useful for productivity than VR. I just don't see a lot of use for VR, especially not on the short term. You can notice this how the VR glasses are largely marketed for gaming, while AR has professional applications which -ultimately- aim to make our lives easier.

>I wasn't implying you didn't read the posts (like you implied with "If you bothered to read it")

My apologies, my tone was probably a little aggressive.

Gaming is the original VR killer app intended to gin up interest and spur development but current VR offerings go beyond that. One of the more interesting applications so far is 'live events' where your digital avatar can attend live sporting events, concerts etc with other digital avatars (you have to experience this to understand what a potential game changer this is). Even watching regular (i.e 2D) movies in an immersive environment on my headset is more fun than watching it on my tv, with the drawback that as of now you can't share the experience with other people not wearing a headset. I'm just scratching the surface of content that is already available NOW (there's even porn if that's your thing, though I personally don't find this very compelling mostly because of the low resolution). Ultimately, if things go the way I think they might, your VR headset would potentially replace not just your gaming console but also your TV set and your laptop/home office (currently I can play chess in a virtual room with only primitive controllers so I don't think working while completely immersed in a virtual environment is a long way off). If you do have the chance I encourage you to take some time to catch up with the current state of the art.

I agree with you that 'boring' is good but in the world we live in unfortunately new tech has to be 'sexy' and 'exciting' to attract venture money and developer interest it needs to get to the boring stage.

Google Glass and chatbots solve both something though. Onboarding new users in production and reducing customer queries

I think there's a lot of hate in this thread for the bot concept. Yeah, most of these text-based bots are bad and useless. Expecting to chat with a computer over text and get meaningful results is a big request, and without the best of the best NLP researchers, you're not going to get anywhere.

However, I do think there's some truth to this article - the rise of digital assistants in Google Home, Amazon's Alexa, Apple's Siri, is, in a sense, the rise of the chat bot. You can text them and they'll do things. You can talk to them and they'll do things. You can even get them to schedule a haircut for you[0], as Google recently demonstrated. I think these bots are semantically identical to what we think of chat bots as doing. If you would've told me 3 or 4 years ago, that people would willingly let a "bot" into their home that listens to everything they say and talks back, I would've been skeptical, to say the least. Now, even my hippie roommate has one of these things. I still don't get it, but to claim that "bots are dead, long live humans" misses the mark about how fast speech to computer/computer to speech tech is evolving. Watch the video and be... amazed:

[0]https://www.theverge.com/2018/5/8/17332070/google-assistant-...

So yeah, the idiot VCs missed the mark - this isn't going to be a consumer revolution, led by a few scrappy start ups, one-man teams, and dreamers. It's a revolution in data collection, human-computer interaction, and AI that's already been taking place behind the scenes at Apple, Google, Facebook, and Amazon for years, and will continue to as long as they hold the tech world's best AI talent.

Indeed. The bot interface is being produced by the giants because they are the only ones with the budget for it. Just like google was the only firm with the budget to make self driving cars actually work, and will likely dominate that field for years to come as a result.

There will be room for startups to put the glue between the bots be everything else, but the core tech for the conversational interface is not going to come from startups.

It reminds me in the early days of "smart phones" people would say where is the Linux phone OS we keep hearing about. They just didn't understand that Android was Linux based.

The Voice Assistance is a vocal chatbot. People were thinking it was going to be text.

No one missed anything except the actual product everyone was talking about. I feel like this happens all the time. We get caught into one strict interpenetration and miss the HUGE thing staring at our face, or ears in this instance.

> It reminds me in the early days of "smart phones" people would say where is the Linux phone OS we keep hearing about. They just didn't understand that Android was Linux based.

It's actually the people who believe that Android is Linux-based that are mistaken. If you write for Android, you write for Android and nothing more. You can't just run the application on a regular Linux (i.e. the one with libc and X11), and if Google ever rips Linux kernel out of it and replaces with, say, DOS, most of the applications won't even need to be recompiled, because they were using Android's API/ABI.

That changes nothing about Android being Linux based, though. The word "based" has a definition, and you're trying to change it.

Then what does it mean "to be based on"? What's its definition, and a one that's well-established and commonly agreed upon, since you're trying to invoke it?

"a fundamental principle or groundwork; foundation; basis"[1]

[1] http://www.dictionary.com/browse/based

And in what way is the Linux kernel the "fundamental principle" or "foundation" of Android? How is it so important that you can rip it out and replace with something totally different without much change to the applications running on top?

You can replace NT with Linux (or Linux with NT) as well. Userspace has nothing to do with this, kernel has its API and the userspace is built on top of it

> You can replace NT with Linux (or Linux with NT) as well.

No, not quite, not without enormous effort. Look at Cygwin or WSL or Wine: tons of work poured into and still not there yet.

On the other hand, typical Android application only ever touches Android's API and doesn't even leave its JVM. Replacing whole kernel (with obvious single point of work in Java interpreter) would be practically invisible.

I think you’re confused about Linux being anything more than a kernel

How so? Android runs on the Linux Kernel.

Ehhhh... I'd say they're less effective than a simple switch so far. Like the chatbots that are good in constrained environments because they're glorified menu systems.

When we can reasonably "chat" with our voice assistants then we have actually gotten there. More specifically, when I can tell my voice chatbot I just want to say "alexa" in the morning when the alarm goes off to snooze. Or when they are adaptable and pattern observant enough that I can say "alexa the usual" in the morning and get the higher volume as I walk through the house and news and my favorite podcast as I ask for every morning. When they come close to accomplishing those things, then we will have vocal chatbots. When they can "remember" just as much as the original Eliza chatbot then we will be there. Currently they are more like vocal light switches than vocal chatbots.

Articles on the features that get real use seem to suggest that people tend not to use these home assistants much, and if they do, it tends to be for music or turning on the lights: http://www.businessinsider.com/amazon-echo-most-used-feature...

Which is to say, they're not getting a whole lot of usage of features you might consider to be chatbot functionality.

I think there is a huge discoverability problem created by the limitations of these devices. Because we don't believe it can understand anything we say, we instead limit our vocabulary and interactions to what we are confident they will understand. It's the opposite of AI in that this is the machine training us how to communicate with it. So we sit there barking out commands because we are unsure of what it can handle, creating a giant hunt-and-peck problem.

This, to me, is likely why Amazon sends me emails saying "What can Alexa do now?"

Perhaps the assistants could have more initiative then.

Ex: "Hi (...), since you always turn off the heater when you leave, would you like me to schedule to always do it for you?"

But there's a big problem in finding the right moment. At many times I'd probably be annoyed.

This. I had to disable google's "assistant" notifications because it kept pestering me about things I had no interest in. They always seem to have an "I see you're trying to write a letter. Do want some help?" clippyesque feel to them. When I activate a system is probably the only time it's ok to add a guiding interaction (like do you want me to make this default?). I really wish I could tell the voice activated things, "Ok, from now on when I say to play 'rockin beats' I'm talking about my Pandora channel, don't make me say 'play rockin beats on pandora' every time." I do almost the same things every day with Alexa, but it still messes up simple context. For instance: I will never ask for alexa for news while THE ALARM IS GOING OFF. If it "hears" news while the alarm is going off then it heard wrong. I said SNOOZE.

That’s exactly the direction Apple is taking with their Siri Shortcuts, it can tell when you’re performing actions and then later suggest them in similar circumstances.

Anecdote: I've discovered features with Alexa on accident before. I was watching Jeopardy at my parents house and someone yelled that Jeopardy was on, and Alexa was like "let's play Jeopardy!" and booted up a game for us. I would have had no idea that Alexa could do that if it weren't for the fact that it's.. uh.. constantly listening to everything that goes on in your house.

> the rise of digital assistants in Google Home, Amazon's Alexa, Apple's Siri, is, in a sense, the rise of the chat bot

Actually, I'm still quite bullish on this. Advocates always talk about how 'VUI' (Voice User Interfaces) will be the next big thing but from what I've observed most people - actual users, not financially invested in their success - still remain quite bullish on them, or they only use them to set timers.

I don't believe Amazon, Google, or Apple have ever actually released sales figures for their assistants. I believe its because the numbers would look so insignificant compared to everything else.

Do you mean bearish? If not, your comment confuses me.

Huh, apparently I do. It looks like I've been using that word incorrectly for quite a while.

I mean that I'm, put lightly, skeptical of the whole digital voice assistant thing.

When I first heard the terms bearish and bullish I thought they were confusing and un-intuitive... Until I heard the following explanation: Bulls thrust their horns up. Bears swipe their claws down. Bulls horn up, Bears claw down. After that it was easy to remember and now it's second nature.

You're one of today's lucky 10,000! :D https://xkcd.com/1053/

I can never remember which is which. Thanks for that mnemonic!

Huh. I always just thought of it as bears hibernating / being lethargic.

And the other ones the other one.

ahhh can see that. Bears are lumbering and Bulls are wildly exhuberant. Although, usually that exhuberance smashes everything in it's path (China shop) or tries to fling a human and gore them (rodeo). Then again maybe that is still apt -- everyone is happy about a bull market until it gets out of control and we are all flung off if it!

The expectation for chatbots to completely solve and answer user questions and take over support is as flawed as the expectation that self-driving technologies in cars allow people to watch Youtube videos and not pay attention to the road or have a hand on the wheel.

The successful bots are ones that don't attempt to understand and respond to every single question perfectly, but instead act as a supplemental tool / guide and offering concrete decision points and actions for the user to take. For example, instead of asking "How can we reach you?" and letting the user enter free text and figuring out if the user entered a phone number or email or some random text that just short circuits the bot, show two buttons "Via email", "Via phone" and clicking each one would then ask for an email, or phone number.

The successful bots also know when to failover to a human being and failing over fast. I've hardly ever had a good experience dealing with sites that employ chat bots and it's not great to be frustrated by a bot when I'm already an angry customer who hasn't received my order.

The one case where chatbots were better than people was at a company I used to work for. We kept getting customers contacting us via facebook to ask the same questions over and over again. "How much does `x` cost?" "When are you open?" "Where is your location?" "Do you have `y`?"

I had to have an internet-connected phone on me at all times because people would send messages at any hour and response time was a factor in our discoverability on facebook.

Of course all of these questions were answered on our website (AND ON OUR FACEBOOK PAGE) but simply responding with a URL resulted in attrition. So I found myself running through the same script several times, all day, all night, every day, every night.

The best solution would have been to set up some kind of text parser which would allow people to navigate the website through facebook messenger (oversimplifying) and then alert a human if it couldn't parse the input. We could even have hooked it into the comment feed because facebook was really bad at notifying us of comments. Taking it a step further, the Mero Mero's dream of offering facebook marketing to other companies as a service would actually be reasonable, because it wouldn't just be me sending out hundreds of copypastas 24-hours a day, forever. But I was never able to put a system together because of other time-consuming duties the company needed me to perform.

This, a million times this.

This is how chatbot functionality should be implemented. Like those "chat with us" links you see on a website that drops you into a window and you get put in someone's queue. It doesn't actually wake them up until the chatbot frontend can't handle the users' request using NLP/decision trees/regex (depending on your level of sophistication of user), and then the whole log is sent to the tech support guy when they get free of the last chat and can read through it and take control.

Maybe the optimistic timelines are flawed but I think its at least reasonable to believe we'll get there someday, in regards to both intelligent chatbot + self-driving tech

I think the problem is framing, or essentially what is a “chat bot”.

20 years ago “artificial intelligence” where taboo words for anyone hoping any funding. It would be called ‘expert system’ or ‘automated agent’ or whatever else that didn’t make people show you the door.

I think what you put as “chat bots” is having the same issue. The concept has been deployed at super large scale and people interact with bots everyday, it’s just not marketed as such.

Currently my phone company, my ISP, my health insurance company, the last airline customer support I had to deal with, they all process a crazy amount of interactions, and all the basic steps were clearly handled by a bot until I got escalated to a human. They are all real world huge scale applications, I don’t know by what other metrics they would be deemed as “failures”, and I don’t think they plateaued, I expect it’s still growing.

> I don’t know by what other metrics they would be deemed as “failures”

Usability? Sure, someone keeps paying for them, but they're not the people who get stuck interacting with them...

It’s not too bad though.

Human operators on big operations where already bound to a script for years now, and the “human contact” part was mostly spelling letters and numbers over the phone.

I’m more than happy it switched to chat, and both sides can deal with the exchange asynchronously, in particual with the bot handling the requirements to open the user info.

I feel it’s actually a win-win situation, as incredible as it seems regarding customer support technology.

> The concept has been deployed at super large scale and people interact with bots everyday

Has it? Are these experiences people enjoy and would like to see more of?

Well, about 10 years ago before iPhone appeared, was the mobile the next big thing? In hindsight of course. How did the data look like before iPhone? Check out the chart in this article https://www.recode.net/2017/6/26/15821652/iphone-apple-10-ye...

There's no data before a breakthrough. We should be wary of hypes but at the same time we shouldn't diminish pioneers.

If you looked at the trend lines, Mobile was definitely growing in 1997 (https://www.statista.com/statistics/203688/handset-penetrati...)

I was working at Radio Shack in 1995-1996 and mobile phones were definitely starting to grow in popularity. By then, you already had subsidized “free” phones and mobile plans were around $35 a month.

Okay, maybe I wasn't clear enough.

When we say that "Mobile is eating the world" or something, we are not talking about the mobile phone we can make a call with but the phone we can use the internet and applications with.

Your data is about the mobile device that allowed us to make a phone call.

If you say that "this is also data" then we have that kind of data for chatbot as well, e.g., the number of sales of Alexa and Google Home devices, or Messenger installations, etc.

There is a mythical internal document at Amazon that describes the Alexa vision. They want basically the Star Trek computer, and they are making good progress toward that vision. And the investment they are making to achieve that goal is hard to overstate.

Yes, it will take a few years. But consider that there are now tens of millions of Alexa pucks out there. The bot may suck today, but it’s getting better quickly, and consumers are receptive and ready to reward the tech giants when they finally succeed.

Amazon are the best people at this as they have the technical ability, vision, and willingness to invest in long term projects.

AWS was not great in its first 5 years. Its capabilities 5 years ago were pretty good but nothing compared to today.

Google is bad at long term development, especially with Ruth Porat at CFO. Microsoft is mixed - for every Xbox and Hololens there's a Courier.

One thing missed in chatbots is that X dot AI is a chat bot that works pretty well. There are others that are similarly domain limited where the current interaction model is an email thread or a text conversation. These tend to be doing well.

>Microsoft is mixed - for every Xbox and Hololens there's a Courier.

My perception of Microsoft is that it's a Hydra.

The upside is that it's not all bad! The head responsible for the product you care for might have been there for a long time, and is doing a really good job in keeping it up.

The downside is that the other heads might be doing something entirely different and turn the whole beast around .

And while all this is going on, some heads just get chopped off here and there.

(sniff Lumia sniff)

The courier concept is far from dead

I looked at X dot AI, but cannot justify the price. I would do 20 meetings a week and place a high value on my time, but to me their value proposition is simply not aligned with the price they are asking. I just don't believe it will help me that much.

As an aside, I tend to think about what I'd be prepared to pay for something before I click on the pricing page. In the case of X dot AI, it was about $5-10 per month. I'm sure it is good, but docusign (similar price) is MUCH more valuable to me - time kills deals and docusign helps bring them in. An AI bot that books in meetings just doesn't deliver the same ROI.

I’d rather talk to a chatbot than send an email or shudder make a phone call. I’ve found my interactions with them to be good and they always hand me off to a real person if it gets too complicated. It seems to save me a lot of time when I use them usually.

I agree. I fully support chatbots for getting you to the right person or handling simple issues. There seem to be a lot fewer accidental hangups when using a chatbot.

But wait, 66% of tweets are reportedly generated by bots.[0] Maybe they aren't chatbots, in a strict sense, but it does seem like most Twitter "conversations" involve bots. There are lots of bots on Facebook too.[1] Automating typical social media behavior is laughably trivial.

0) https://www.vox.com/technology/2018/4/9/17214720/pew-study-b...

1) https://mashable.com/2017/11/02/facebook-phony-accounts-admi...

Yep. I remember tons of marketing/advertising people going around enthusiastically talking about how chatbots are "the next big thing." No professional UX people or engineers were really pushing the tech that I knew.

What I have understood from my little experience is usually the "next big things" - Always seem way to big in the start - There is good exponential acceptance initially - But then there is always slow acceptance rate

I think its more about chat then chatbots. Chat has already become the next big thing between customer to customer interaction. What chatbots are trying to do is enabling brands to start interacting with their customer on chat. Human-based support was always there but to able to interact with customers using chat during his/her complete lifecycle was not possible using human agents. But chatbot can solve that. We are moving back to chat because it is one of the two natural media of communication (voice/chat & gesture) to humans. The hype about chatbots was bad, resulting in everyone trying their own version of chatbots and fail. But for those who used it in its actual capabilities, it has done wonders. For some top financial brands of India, it has increased no. of people interacting with a brand to about 400% resulting in ~200% increase in marketing leads and lead quality by ~150%. This is actually the NEXT BIG THING for the marketing department. Also, chat apps are more evolved then SMS/phone apps which give user/customer more control over communication, resulting in happy customers.

Chatbots are not solving any problem as they are not a solution. Instead, they should be used as a tool to solve any other problem, just like AI/Crypto.

I began my career building IVR systems in the early 2000s (Nuance, etc). In college I worked at my university's speech recognition lab. I remember an uneasy feeling while building and testing my first speech application. I thought: "This is a statistical algorithm with nothing else behind it. It's just a good guesser over a very limited domain of 'knowledge'."

At the time the tech industry was hailing speech recognition as the next big thing. There was a lot of investment in speech: BeVocal, TellMe, Nuance, SpeechWorks, AT&T, L&H, etc. Replace your call center workers with automated systems, use voiceprints to secure sensitive transactions, etc. Amazing AI would effect cost savings and make your business efficient! Sound familiar?

Very few users actually liked these systems over the previous versions that just used the phone keypad. Poor recognition accuracy really pisses users off. And most of the time the recognition errors aren't the user's fault. Ever try to recognize speech on a cellular channel using an acoustic model trained on a landline channel? Crap! This was certainly a problem during the early 2000s because of the emerging mobile phone market.

The UIs (or VUIs as they were called) were awful since they mostly replicated the touchtone versions but added "you can say 'accounts' to...". They continue to be awful because, while speech recognition had gotten a lot of funding (DARPA, etc) in the past, the UI aspects were pretty much ignored and underfunded. People interact very differently when speech is used as the medium. The interaction automatically becomes social. Social interactions with machines are decently well studied (Cliff Nass for one), but the very nuanced aspects are difficult to bake into an IVR system, especially under pressure to deliver on a deadline. The experience feels unnatural very quickly and users either run for the door or starts mashing the keypad.

Ultimately the web killed off industry interest in IVR systems. Want to know your account balance? Log in to a web site and it's right there. No need to dial in your account number, verification and run a gauntlet through a phone menu.

Fast-forward to today and chatbots are the new IVR systems, but without the speech-to-text portion. The speech-to-text portion is the easiest part by far in those types of systems. There's still the need for a parser (shallow or otherwise), a way to pair up queries with responses/actions, the ability to track a dialogue and its context, and most importantly, a "natural" feel to the interaction.

Chatting/texting and speaking are serial in nature; very slow and inefficient. A well-indexed FAQ or an intuitive, well-designed GUI is more useful than a chatbot. The former helps in shortening search time and the later helps at streamlining transactions. Information search and transacting comprise the majority of actions that users perform when they visit a site/system/application. When you talk to a human agent for these two types of actions, guess what they're using to assist you? The same interface you'd use if you served yourself. They just know it better.

As an aside, the only noticeable/innovative use of chatbots I've seen in the past 10 years is on porn sites. I really took notice of the messages in the chat window when it started nagging me for ignore it. I wouldn't be surprised this round of chatbots were inspired by that.

I think that did happen. Banks and utility companies started to offer chat-like interfaces (without a human responder) for services and help. The number of end-users could probably have been graphed as a steeply rising curve for a while. It just didn't justify the breathless expectations of the companies trying to sell chat interfaces at the time.

Yeah, this is the "trough of disillusionment" phase on the Gartner hype cycle. If anyone is looking at the exponential improvements in speech and text comprehension over the past years and unable to see that it will only get better from here, then it's not hard to imagine a future, where in 10 years chat bots are just as good as humans for any rote request.

And yet there are companies that are starting to make money from these things. Maybe this is a classic "can't take new tech and apply old business model" kind of thing.

https://www.theinformation.com/articles/for-developers-faceb...

Behind firewall, so here's a key quote:

Facebook’s early efforts to turn Messenger into a service that people can use for shopping or getting news or weather sputtered quickly. But two years on, the Messenger effort is showing signs of life.

Several developers say they have been selling more bots and applications through Messenger in recent months. Trip-booking service SnapTravel, for instance, has seen its cumulative sales through Messenger grow from around $1 million as of May 2017 to around $10 million now, according to CEO Hussein Fazal. Msg.ai, which develops tools for customer service and marketing for messaging apps, has seen enough of a pickup in business to hire a new executive and marketing team, said Msg.ai CEO Puneet Mehta.

Alexa and Google Home are simple chat bots. Say something and get a single, reasonably accurate response. They seem to be doing pretty well.

The little chat bot widgets never took off because they didn't solve NLP anywhere near as well as Amazon or Google did.

If a next gen chat bot can make a dinner reservation over the phone (like Google's demo), it definitely will be the next big thing.

So much of our day is communicating and if we can automate our communication effectively, it opens the doors for even greater productivity.

> I’m pretty sure there was never a high growth phase where users were actually interacting with the chatbots.

There was somewhere around 2003-2005, when a couple good ones started showing up on AIM. Since that was the IM of choice at the time, there was zero friction to actually interacting with one.

Anyone predicting it as a "next big thing" after around 2007 totally missed the actual hype and was trying to build on something users had already dismissed as not actually the future.

I remember being so unimpressed after Facebook's hype because their "bots" were no smarter than SmarterChild was back on AOL in the early 2000's. (Which was at least innovative for its time.)

I at least thought that there would be some Alexa/Siri-level AI going on, but most chat bots are about as intelligent as texting commands to an SMS short-code number.

I love this comment and I kind of want to replace "chatbots" with "cryptocurrency" in a few years.

Reply on news.ycombinator.com