It's distressing that more and more chatbots are spawned every day to take the human element out of customer support or when engaging with a business. Chatbots work when they're a complement to human customer support, but when they're up front and center they tend to be distracting, frustrating, and their attempts to act human often amuses me. I do like Kik's and Messenger's approach to a bot and I think they are on the right path.
Online shops have usually a link where my orders are displayed and updated in real-time. Worked well.
I am only half joking. The reason nerds like CLIs is that they can just tell the computer what to do without having to first coax it into offering the appropriate menu.
The same is true for non-nerds, except they no one really wants to learn the exact commands. If for some reason, clicky-window GUIs had not been invented in the '80s -- then the hot thing in UI research might have been fuzzy, english-like text interfaces, including online help.
That is basically what is being re-discovered via chatbot technology.
So while I am skeptical, I am starting to understand the value of a chatbot and possibly a single way to do things. Maybe the "easy" path via the chatbot and if someone is a real poweruser they can use the window and type all information themselves.
For example: we had a bot that could take care of HR, expense, and IT issues. So instead of having to look in one of the systems or the documents on SharePoint, you could just ask: "how many days of PTO do I have left?" Or "when will my last expense report be paid?" or "can I get email on my corporate phone?", and the bot would be able to handle it. Especially in larger companies with lots of systems finding that can be difficult.
HN is probably not the intended audience for chatbots :)
edit: typos
I've always felt like bots for use in customer service were trying to solve a problem by adding more problems. So, now, not only is someone talking at your company because they have a problem that couldn't easily be solved by your web presence, now you're making them navigate an uncanny valley of dumb questions and answers...following the same old support script only now with even less intelligence and autonomy on the part of the company rep.
I am doing something like that myself as a side project, not exactly as "AI" more as voice interface responding to voice commands that uses humor and character voices to reply. In way, the problem of bots is actually "AI".
You can create a more flexible and lively experience if you put "AI" aside and really think of how people interact and create responses that are not based on "mean" values, but on real conversations. You don't even need the whole AI backend, you can just use javascript, audio files with or without accompanying text. Because so much of the conversation is bland, having a "voice" will make a huge difference.
AI developers should consider collaborating with writers and comedians:)
You're right that the typical chat is quite bland and feels scripted, but I think companies are picking up on this and are trying to improve on this.
Chatbots in China over WeChat act as a lightweight Interactive Voice Response (IVR) system, which is highly successful due to a set of market factors which are (at least from my perspective) unique to China.
But, as I've been saying for about a year now, chatbots, in their current incarnation, don't make a lot of sense outside that niche.
Looking at it from either the American or Japanese perspective, there are a set of expected standards for customer support, which are either already met via web-based mostly-automated systems (Amazon is really good here), or by being able to directly chat with a human being (perhaps supplemented by AI) to solve a very specific and usually uncommon need.
Replacing that setup with nothing but a text-based IVR system is a non-starter for any company that wants to remain in business.
Hence, it was a natural progression to allow payments to businesses which is where chatbots came in.
Perhaps i'm a strange case, but the only times I or anyone I know have ever had to use support are for these cases.
If an issue is common, its not an issue for support, its a failure of ux or design
I think there's definitely two distinct types of people in this regard. Just like automated responses from phone lines, I much, much, much, much prefer "speaking" to a robot that:
a) is often less stressful than speaking to a human
b) is often able of hearing and understanding me better than a human
c) is often more easy to understand than a human
d) is often more straightforward to get help from
e) is often "happier" to help
f) is often much faster to interact with, especially in repeated tasks like paying bills, authorizing payments, etc
And if you do need to speak to a human after all, they typically treat you better knowing you've already gone through the "easy stuff" and are seeking something extra.
I'll regularly put off or completely avoid calling any companies (support, bills, etc) that don't have automated processes, and will just take my business elsewhere at the first opportunity.
I wholeheartedly wish every line I had to call were automated, and extending the real-time support capabilities of online communication only makes quality support more ubiquitous and easier to get, IMO.
--
I spent an hour on the phone with UPS support this morning (first their 1800 number, then their Reykjavik office, then one driver, then another driver, then the Reykjavik office again) that could have avoided a ton of back and forth and miscommunication if I'd just spoken to a bot in the first place that could take my request ("Hi, your online tracking says a package could not be delivered to my address because 'I moved', when in fact I did not. Please deliver it anyway.") and put it directly into a database available to the office/drivers.
People will work to support a community, because they gain self actualisation from helping others and getting status from others. Some more "karma" can be generated from early access programs and expert feedback status on new products. That's all though - there is not an endless well of good will out there to be exploited by social media and bot infrastructures. People like real gold!
For example in our case, clients often have problems logging in. With our chatbot they can easily get a one time password sent to their phone. With a regular search/faq they would have never known about that, because they are often too frustrated with the UI experience for finding that information. With a natural language conversation, it is easier :-)
Getting a regular non techinal user to search and read the documentation for this is just insane.
An obvious solution is a search box, but most sites also have terrible search functionality, so users have learned to ignore it. For the rare gem that has useful search, dressing it up as a chatbot could get people to actually try it. Too bad it looks like a lot of devs are just dressing up their shitty search as chatbots, so users are being trained to ignore them as well.
I agree, but that doesn't mean they're as bad as them.
It's _much_ faster to read a text and reply with a word/number/tap than it is to wait for the option you want to be read out to you: "for x press 1, for y press 2... for z press 9". Or worse, have to speak what you want, wait for it to parse what you said, and then ask you to confirm.
I work at a company that's helping businesses automate parts of their customer support with chatbots and you'd be surprised to learn that CSAT improves when they employ some automation. Usually this means getting a quicker first response to the customer's question (and some of the time successfully answering it!) as well as automating tedious data collection (what device are you on? what's your email? etc.).
It's early days for chatbots and I think we'll see a lot of improvement where automation through conversation should be and I think we'll also see the rest shake out.
More to your point: I agree with you, IVRs are usually really bad, but some of them are impressive. Have you used Apple's recently? It's very good at routing and making cases where it's uncertain invisible.
I certainly prefer chatbots over the menu selection when I call. But that's about it. Mose of the chatbots are similar rule based systems anyways.
Have a look at Speakeasy for a bot that is trained on reddit using deep learning, it produces very plausible responses to queries but has a few weaknesses.
Firstly it's not task directed, so if you ask it about changing the oil in a car it doesn't know any useful facts and it doesn't present them - it just fobs you off.
Secondly, type "you're" into it - it will be very rude to you... because... reddit.
So what's needed is something that can read technical documents and use them at the right time in a conversation. Deepdive (http://deepdive.stanford.edu/) gives you a bit of this capability, but when I used it (version 0.4) I struggled, maybe things have improved! Non the less; I do believe that this is state of the art for machine reading.
IBM did a tonne of work on question answering in Watson but I am now so confused about Watson that I literally don't know what to make of where their work has got to.
Hope that helps, would be interested if there are different opinions out there!
As with the brief flurry of the Semantic Web era, I think the lesson with chat bot VI/AI will again be as much in striking the balance between things like Deep Learning and things like Expert Systems, assuming Deep Learning will never be perfect and Expert Systems will never be easy/cheap (building a good one still requires a domain expert in that field to train it in whatever rules engine you are using to express the domain).
However I think that the cost of knowledge engineering is still massive and the challenge of this kind of project so sharp that it makes me quail to imagine execution! I think three or four years of full on tool development might make this more practical...