[Art]

OR...have the bot be even more convincing by saying:

Human: How old are you?
   Bot: I am 2 years old.

   Human: How old are you?
   Bot: You already asked that question.

   Human: How old are you?
   Bot: Seriously?

 

[8pla.net]

Bingo!

And, yet another method to make an A.I. engine more convincing.   Art is right.

What are some other methods or strategies in A.I. to achieve convincingness?

_________________________________________________________________________
Call for Alpha Testers: http://www.chatbot.tk/technology

[kei10]

You mean, this? Let's say the bot has knowledge about human age, as well as your age... Let's say the bot knows who he is talking to -- We have the ability to do this, but what if...

Human: As a test of your credibility, tell me, how old am I?
Bot: You are 2 years old.

Human: What? No I am not. I'll ask again. How old am I?
Bot: You have already asked that question. You're 2 years old.

Human: Bot, I mean it, how old am I?
Bot: You're 42 years old, silly, I was just kidding!

I googled "AI that tells jokes" ...
It returned a top result.

"Joke-Telling Robots Are the Final Frontier of Artificial Intelligence"

[8pla.net]

Human: As a test of your credibility, tell me, how old am I?
Bot: You are 2 years old.

Human: What? No I am not. I'll ask again. How old am I?
Bot: You have already asked that question. You're 2 years old.

Human: Bot, I mean it, how old am I?
Bot: You're 42 years old, silly, I was just kidding!

I just ran your questions, as an alpha test... Here are the results:

You:   As a test of your credibility, tell me, how old am I?
A. I.   Is that the real reason?


You:   What? No I am not. I'll ask again. How old am I?
A. I.   Are such questions of what, on your mind often?

You:   Bot, I mean it, how old am I?
A. I.   Which answer of how would please you most?

I sorted this alpha test in chronological order for easier reading.   What do you think?



_________________________________________________________________________
Call for Alpha Testers: http://www.chatbot.tk/technology

[kei10]

I just ran your questions, as an alpha test... Here are the results:

You:   As a test of your credibility, tell me, how old am I?
A. I.   Is that the real reason?

You:   What? No I am not. I'll ask again. How old am I?
A. I.   Are such questions of what, on your mind often?

You:   Bot, I mean it, how old am I?
A. I.   Which answer of how would please you most?

I sorted this alpha test in chronological order for easier reading.   What do you think?

I'm not sure -- the AI seems much broken then ever -- no kidding, I think -- but wait, what do you meant by chronological order? Oh, I see, a chat log with the AI. Well, it's rather interesting, although you might need to keep a limiter, or/and an algorithm you need to consider to make the AI say certain things to drift away from the subject, as jokes must not last too long, and depend on the user's words -- and finally return back to the subject -- We humans can track multiple subjects at one time.

In other words, the AI would need the ability to detect anger from user's sentence. Such as "I mean it..." from the example. This indicates that the user is serious, and gotten tired. Additionally, that doesn't mean the user would actually get tired from other ones, for example, surprise-jokes, that you don't even expect. All these probably require emotional intelligence to be built alongside into your AI.

Human: I wish there's no raining, so I can go outside.
Bot: I wish there's rain, so I can go outside. //Unexpected sarcasm

Human: What? //Surprise
Bot: You heard be, buddy.

Human: But, why?
Bot: Are you a moron? I've never felt what a rain is like before. I'm curious. //Reasoning

To even begin with joking, the AI would need a lot of knowledge about irony, metaphor, and sarcasm, and even possibly a bit of knowledge about the world and how it works. I'm not sure if you're after such intention for your AI. It would be a groundbreaking history if anyone could finally create an AI that could joke.

Basically that would need human-level language processing, or a true AI to even build a joke out of air. 

[8pla.net]

Chronologically, for our purposes, simply means it is sorted in a different time order on the website: http://www.chatbot.tk/technology   What works well on the website, may  not necessarily work as well for forum discussion purposes.  So, it was reformatted for the forum.

Broken?  I very much want to receive reports about what may seem broken, or in need of improvement, with suggestions on how to do that. That is the purpose of alpha testing, to discover improvements.  Thanks for your feedback, and for alpha testing.

_________________________________________________________________________
Call for Alpha Testers: http://www.chatbot.tk/technology

[kei10]

Oh.

Broken? I very much want to receive reports about what may seem broken, or in need of improvement, with suggestions on how to do that. That is the purpose of alpha testing, to discover improvements.  Thanks for your feedback, and for alpha testing.

OH... ... My bad. I see. I'm sorry if I'm not that helpful, and you're welcome.

[Art]

#############################################
You:   As a test of your credibility, tell me, how old am I?
A. I.   Is that the real reason?

You:   What? No I am not. I'll ask again. How old am I?
A. I.   Are such questions of what, on your mind often?

You:   Bot, I mean it, how old am I?
A. I.   Which answer of how would please you most?
#############################################

8pla.net, excuse me for participating but your bot's example from this exchange seems like it came from an earlier version of an Eliza bot, playing the part of a Rogerian psychotherapist...always asking somewhat leading questions.

Nice work so far....

[8pla.net]

Happy to have you participate, Art.  Well, Alice, Mitsuku, and other excellent chatbots may have started out like that.  Of course, this new A.I. engine design is in a much earlier development phase, than those well established A.I. engines and scripts. 

Thanks for the compliment!  Your feedback is useful.  One of the key features for any A.I. engine is to run scripts that are interchangeable.  An interchangeable script of a Rogerian Psychotherapist may be the most well known. 

Interchangeable Scripts, separate from the A.I. engine, help prove that a new A.I. engine design is scalable.  After alpha testing is complete, the A.I. engine will be proven stable.  At that point, a fresh new interchangeable script would be more useful and valuable. However, it would almost be wasted on early alpha testing, when operational errors are most likely to show up and ruin the intended effect.

_________________________________________________________________________
Call for Alpha Testers: http://www.chatbot.tk/technology

[8pla.net]

OH... ... My bad. I see. I'm sorry if I'm not that helpful

kei,

You are super helpful.  Please don't hesitate to share your ideas. 
Other points of views are so useful for alpha testing.  So, I do want
to hear from people.  Whether it is good or bad, is just fine with me.

[8pla.net]

I just updated my technology to be as smart as Google’s artificial-intelligence bot.

Here is the proof...   

References:

http://www.businessinsider.com/google-tests-new-artificial-intelligence-chatbot-2015-6

http://www.chatbot.tk/technology

[8pla.net]

ALL NEW

The first alpha test script has been archived.  So, it is offline, and a new script has replaced it.
Currently, alpha testing interchangeability and response time with a completely new much larger script.   

Tweaks are obviously necessary!   At least this alpha test seems somewhat operational.
So, far the response time seems to be pretty quick.   Please comment on the response time...

Any other comments?  Reference: http://www.chatbot.tk/technology

[kei10]

You:   Hi.
A. I.   Um.
You:   Tell me something random.
A. I.   Tell you something.
You:   What is your name?
A. I.   Are you asking about my ?
You:  What is your name?
A. I.   Very well.
You:   What is your name?
A. I.   Alas.
You:   Is Alas your name?
A. I.   Robots' name.
You:   Who's name is that?
A. I.   Is that .
You:   That is who?
A. I.   Thanks for explainingwho.
You:  Who are you?
A. I.   I ask myself that question, "Who am I?"

Uh... o_. Interesting...? The response time is great, fast.

[8pla.net]

Kei found a glitch, "Thanks for explainingwho." while alpha testing. This response concats words "explaining" and "who" together.  Very helpful, since other similar glitches were edited as a result of that alpha test. 

Undoubtedly, further alpha testing may discover more glitches in this script.  Just the beginning, really.  The end is nowhere in sight.  So feedback from alpha testing is very constructive.  Kei also reported a quick response time.

This alpha test confirmed the scalability of this new A.I. engine.  It is scalable from several hundred records to several thousand records.  Thank you Kei.  Anybody who may have alpha tested before:  Thank you in advance for re-alpha-testing this script...  Reference: http://www.chatbot.tk/technology

[Art]

I noticed:

Your chatbot has no name.
You:   What is your name?
A. I.   Robots' name.

You:   I'd like to know what to call you.
A. I.   You are not the boss of me...

You:   I don't want to be the boss of you.
A. I.   Neither does my botmaster.

You:   Tell me your name.
A. I.   I thought I told you already.

You:   Your name?
A. I.   Robots' name.

You:   Robots' name?
A. I.   Long time.

(maybe from Love you long time)

It gets a bit testy with it's reply: "You're not the boss of me..."

From the same conversation I noticed a connected "thought":

You:   I mean that the temperature outside is very hot today.
A. I.   That the temperature outside is very hot today.

You:   Yes, in fact so hot that the excessive heat can cause problems
   for living things., you mean that the temperature outside is very hot today. (it states the you mean that the temperature outside is very hot today, but it appended it to the end of my sentence [for living things.])
A. I.   Is that a fact.

You:   Absolutely.
A. I.   Hooray.

Lastly was this section:

You:   Then am I to understand that you are a robot?
A. I.   You are not the boss of me...

You:   I was not trying nor wishing to be the boss of you.
A. I.   I heard you were.

You:   I'm afraid you were misinformed.
A. I.   Robots do not experience fear.

You:   Then you should do well as a   career politician or a soldier of fortune.
A. I.   I'll take that under advisement.

You:   No problem. It was actually a    joke., then robots should do well [again it interjected the previous comment]
   as a   career politician or a soldier of fortune.
A. I.   No problem (that).

It seems that several passages in my conversational text with your bot indicate that it is grasping what was said (typed) but some of these comments are interjected into or appended onto the end of my text. This is usually indicated with the ., text (period comma space then text).

Otherwise it was a fun and interesting chat dialog. Obviously with a new bot there will be some "bumps" in the road but overall, I think it has promise!