Ask HN: Is anyone actually excited about using chatbots?
6 comments
I'm not sure "excited" is the right word. Some technology is going to break through the keyboard/mouse/touch barrier and bring us into more natural contact with all the data and processing power around us. It's almost certainly going to be based on language. What else is there? Gestures have been tried, as have custom "swiping" languages, but the momentum is shifting strongly toward natural language. So maybe not "excited" but definitely "fascinated."
That's a really interesting point. And it certainly would be phenomenal if we could interact with computers using natural language.
At the same time, fascination doesn't necessarily make a person an active user of any particular product.
At the same time, fascination doesn't necessarily make a person an active user of any particular product.
A person doesn't have to be an active user of a chatbot for it to be useful. Chat could be the ultimate discoverable API - with bots talking to one another.
Or the ultimate undiscoverable API. It is quite similar to speech input in cars. If you don't know what is valid to say and what not you are quite lost and will probably only use a subset of the total available API. And writing/yelling random things against a device just to check if it will work is something that most people won't do.
This applies to human users, but I think it would be similar for non-humans as well.
This applies to human users, but I think it would be similar for non-humans as well.
It wouldn't be hard to build in tools to aid discoverability. A 'help' function that returns a list of features for example. That's pretty much what WSDL[1] tries to be, but that's really to hard to use well. If there were a much more flexible "natural language" interface to it then it'd make finding and consuming APIs both easier and more robust.
I'm just musing on the possibility really. I'm absolutely sure there'd be difficulties in the technical implementation, and obviously it'd be quite inefficient compared to a known and well-defined API, but if it could be made to work the advantages may outweigh those problems. Plus, implementing it for humans goes 90% of the way to implementing it for computers too, so it's almost zero-cost if you're considering a chat interface to your service. It's something to think about anyway.
[1] https://en.wikipedia.org/wiki/Web_Services_Description_Langu...
I'm just musing on the possibility really. I'm absolutely sure there'd be difficulties in the technical implementation, and obviously it'd be quite inefficient compared to a known and well-defined API, but if it could be made to work the advantages may outweigh those problems. Plus, implementing it for humans goes 90% of the way to implementing it for computers too, so it's almost zero-cost if you're considering a chat interface to your service. It's something to think about anyway.
[1] https://en.wikipedia.org/wiki/Web_Services_Description_Langu...
True indeed.
> Chat could be the ultimate discoverable API - with bots talking to one another.
This is something I've thought about a lot recently. It would be amazing if natural language becomes a viable protocol for machine-to-machine communication and machine-to-human communication. The real potential for bots may be in forming a massive network of specialized agents which talk to each other and collaborate to fulfill any single request. People could easily be part of this network too - a bot sends a person a text message asking a question and waits for the answer.
Of course, just from the perspective of a lazy developer it would be great to be able to interact with APIs by writing requests in plain English instead of having to lookup endpoint docs.
> Chat could be the ultimate discoverable API - with bots talking to one another.
This is something I've thought about a lot recently. It would be amazing if natural language becomes a viable protocol for machine-to-machine communication and machine-to-human communication. The real potential for bots may be in forming a massive network of specialized agents which talk to each other and collaborate to fulfill any single request. People could easily be part of this network too - a bot sends a person a text message asking a question and waits for the answer.
Of course, just from the perspective of a lazy developer it would be great to be able to interact with APIs by writing requests in plain English instead of having to lookup endpoint docs.
Chatbots on irc and things like reddit (e.g. mirroring images or quoting a bible verse from the name, or getting info about a user),or twitter (eg the wolfram tweet a program bot, or the deep forger bot) can be convenient for some things, but I'm not particularly motivated to make any purchases with one, or do anything with one that requires registering, or anything that makes me think it profits by collecting information about me.
I can think of many cases where I prefer a text-based interface to a phone call. I can't think of any case where I specifically prefer a bot-driven text-based interface to a human-driven text-based interface, though. If chatbots let companies build more and more effective text-based interfaces to replace call centers, then I'm at least a little excited, but it's the result and not the process that has my interest.
Most Slack integrations are essentially first-generation chatbots (parsed and tokenised text input linked to a finite state machine to actually do things), and they're hugely popular among users. I see Microsoft and Facebook's moves as bringing those things to a wider audience - instead of telling Hubot to deploy some code using Slack we'll tell a brand we've bought something from that an order has been delayed using Facebook.
People already seem to prefer to interact with a brand using Twitter than the phone. There's no reason why many Twitter interactions need a human being behind them.
Chatbots are essentially Blade Runner for customer relations. Sort of.
People already seem to prefer to interact with a brand using Twitter than the phone. There's no reason why many Twitter interactions need a human being behind them.
Chatbots are essentially Blade Runner for customer relations. Sort of.
> People already seem to prefer to interact with a brand using Twitter than the phone.
Do you think they prefer to use Twitter because of the conversational UI? I've always thought it was because Twitter is public, so brands respond to complaints quickly out of fear that they will look bad otherwise.
Do you think they prefer to use Twitter because of the conversational UI? I've always thought it was because Twitter is public, so brands respond to complaints quickly out of fear that they will look bad otherwise.
I don't know why, but if it is speed and visibility as you suggest, a chat bot hooked up to Twitter would be a really valuable asset to a business.
AS others said, they are not new but the way are being used is something which matters, specially AI bot. Like we used to hear There's an app for that, we soon will hear There's a bot for that
I tried to develop an FB bot and record details here:
http://blog.adnansiddiqi.me/develop-your-first-facebook-mess...
I tried to develop an FB bot and record details here:
http://blog.adnansiddiqi.me/develop-your-first-facebook-mess...
The question hinges on the definition of "chatbot". Since I never get excited by the prospect of chatting with a scripted human, I'm not excited about chatting with a bot. On the other hand, natural language API's available via text appeal to me as an alternative to the-hamburger-of-intuition thinking that dominates interactive design.
The thing is: all the excitement I've seen is coming from people who are hoping to make money from other people using chatbots. I haven't seen much excitement from people actually using chatbots. Am I wrong?
Do you ever use chatbots? If so, what do you use them for? Why do you use them? What's great about the experience?