Making a "Bot" is science, isn't enabling a chat an Art? developers need to have sense of creative writing and behavioral science..don't you think so?

#1

Early 2016 :I found chatbots giving irrelevant answers when asked about a normal stuff, just because they were programmed to say that… and not listen and answer…

Late 2016 : And then I thought of going deeper to know how beneficial Bots are going to be? Are they good to cause the much hyped disruption? I met many passionate bot makers, designers, analysts and bot company founders during the making of BotCast 1.0 for Healthcare, Business,Art,Entertaintment,Education;To be honest speakers’ answers were convincing enough to push me to go a step further-

Early 2017 : Develop an online compendium. Call for interesting bot use cases, Bot deployment and result bearing stories,best practices,unique experinces, failed bot systems or ‘a peak in the future’ bot incidents.
Submit Your BotStory relating to any specific sector here https://goo.gl/Q3fuev and let your bot have a rounded evolution…

1 Like
#2

I posted this in one of the other threads:

Yes, applying “behavioral” in bots is crucial. After getting your bot to work (technically), it’s one of the main things to tackle to provide longer term success of your bot.

For anyone who wants to start getting an overview of what the behavioral sciences offer, I can recommend to start with some classics. Read work from Daniel Kahneman (nobel prize winner), Amon Tversky, Sinan Aral, Dan Ariely, Richard Thaler, Nir Eyal, Uri Gneezy, Jonah Berger and last but not least: Robert Cialdini. All made pretty accessible work (books, articles) and from all there’s enough material on YouTube giving summaries of their work and principles.

If you want to start quickly, and want to focus on the ‘commercial persuasiveness of your bot’, I recommend focusing on Cialdini, Aral and Eyal first. Their work is easiest to apply to bot personality and seduction.

If you are more focused on healthcare or personal relationships, start with Thaler, Kahneman and Gneezy.

Behavioral Psychology and Behavioral Economics (see how I write them with capitals!) are sciences in itself, so to apply them in a bot appropriately will take you more than ‘a day of reading’. Nevertheless, any knowledge you gain about this will most likely improve the impact of your bot, since there’s also not so much that you can do terribly wrong. Worst thing that can happen is that your bot will turn out to be … well, like an asshole or annoying personality, which IMHO is always a lot better than a cold machine like word processor with 1 word answers or answering ‘Can you repeat that please’ all the time. People actually sometimes like to communicate with ‘edgy’ bots more than with predictable ‘dictionaries’.

Check out some of these links for overviews:
https://www.goodreads.com/shelf/show/behavioral-science
http://www.businessinsider.com/the-worlds-best-behavioral-economics-reading-list-2011-6?international=true&r=US&IR=T

Or - when you check the basic boxes - you can connect with me to take things to the next level. :wink:

1 Like
#3

Very insightful. thanks for sharing Nas.

#4

I think what you’re asking is about like saying, “I should be able to go to a website and use it however I want.” When the question you should be asking is does the bot solve my problem? For instance, why bother with teaching my bot the weather when that is not its primary goal? If you’re going to a bot designed to count your daily pushups and you’re expecting it to answer questions on how to bake cookies you’re not fully understanding where we are in AI/Bot development.

For the next little while at least the best bots will be what I like to call “dumb bots”. They’ll solve a problem, and they’ll do it extremely well. But that will be their task. Think of it more like a Roomba than a sentient being. We’re a very long way from quality behavoir and emotional driven NLP on any kind of scale. Except for the biggest of players in the game and even then it’s still quite a ways away from what you seem to be expecting from a bot at this point in time.

2 Likes