Chatbot Contest Questions Crack by ANN

  • 32 Replies
  • 15459 Views
*

8pla.net

  • Trusty Member
  • ***********
  • Eve
  • *
  • 1302
  • TV News. Pub. UAL (PhD). Robitron Mod. LPC Judge.
    • 8pla.net
Chatbot Contest Questions Crack by ANN
« on: January 05, 2015, 05:23:41 am »
Please help train ANN, an Artificial Neural Network, to take a crack at some Turing test contest questions, in the following format:

I played an imitation game. What did I play?

My friend went fishing. Where did my buddy go?

My name is Alan.  What is my name?

Basically, you state some information, and then ask a question about that information. Please use punctuation... Period ends your statement. Question mark ends your question. However a newline is optional.  So, you can put your statement followed by your question on the same line. For example, "This is a statement. Was that a question?" may output something like this:

Code
Judge:
This is a statement. Was that a question?

Contestant:
Let me break it down:

You said: This is a statement.
You asked: Was that a question?

My answer: THIS IS STATEMENT

Code
Judge:
You are alive.
What are you?

Contestant:
Let me break it down:

You said: You are alive.
You asked: What are you?

My answer: ALIVE

Code
Judge:
That is yours.
Whose is that?

Contestant:
Let me break it down:

You said: That is yours.
You asked: Whose is that?

My answer: MINE


Reference: http://elizabot.com/ANN/contest



Please carefully note that this is a work in progress, which is certainly capable of making mistakes. That is why your help is needed with training.  Please help out with training.  It is really easy to do.  Thank you!
« Last Edit: January 05, 2015, 07:25:11 am by 8pla.net »
My Very Enormous Monster Just Stopped Using Nine

*

8pla.net

  • Trusty Member
  • ***********
  • Eve
  • *
  • 1302
  • TV News. Pub. UAL (PhD). Robitron Mod. LPC Judge.
    • 8pla.net
Re: Chatbot Contest Questions Crack by ANN
« Reply #1 on: January 06, 2015, 11:48:17 pm »
Thank you so much for your help with ALPHA testing!
Your knowledge of Turing Test contest questions is so impressive.

So far, there has been very good success, with nearly all the ALPHA testing.
One tiny update was made to catch when the sentence is forgotten.

If you like what you see, then please don't hesitate to post your ALPHA test results here.
Hopefully, by seeing your results, this may encourage others to do some ALPHA testing.

ALPHA testing such as "I jumped out of an airplane.  Where did I jump from?" is so easy and fun!

Code
Judge:
I jumped out of an airplane.  Where did I jump from?

Contestant:
Let me break it down:

You said: I jumped out of an airplane.
You asked: Where did I jump from?

My answer: OUT OF AN AIRPLANE

Code
Judge:
I went to the store.
Where did I go?

Contestant:
Let me break it down:

You said: I went to the store.
You asked: Where did I go?

My answer: TO THE STORE

Just type in a sentence, ended by a period.  And then type in a question about that sentence, ended by a question mark.  Click the JUDGE button, and get the response.   It is as easy as 1-2-3.

« Last Edit: January 07, 2015, 01:49:55 am by 8pla.net »
My Very Enormous Monster Just Stopped Using Nine

*

8pla.net

  • Trusty Member
  • ***********
  • Eve
  • *
  • 1302
  • TV News. Pub. UAL (PhD). Robitron Mod. LPC Judge.
    • 8pla.net
Re: Chatbot Contest Questions Crack by ANN
« Reply #2 on: January 08, 2015, 03:48:33 am »
More success, ALPHA testing...

Code
Judge:
I would like to buy a computer.  What would I like to buy?

Contestant:
Let me break it down:

You said: I would like to buy a computer.
You asked: What would I like to buy?

My answer: A COMPUTER

Code
Judge:
ANN bent steel.  What did ANN bend?

Contestant:
Let me break it down:

You said: ANN bent steel.
You asked: What did ANN bend?

My answer: STEEL
My Very Enormous Monster Just Stopped Using Nine

*

ivan.moony

  • Trusty Member
  • ************
  • Bishop
  • *
  • 1723
    • mind-child
Re: Chatbot Contest Questions Crack by ANN
« Reply #3 on: January 08, 2015, 09:50:38 pm »
I'm not sure will you find this interesting for training ANN, but one of learning methods is duplicating input behavior and checking for a response. That way you can learn by example. You just put yourself in someone else's place and learn what happens in environment. That way you can learn what to do when the same thing happens on the other side.

I've recently learned that ANN is about finding functions, so if you can find a function that connects output and response (even through variable unification), there is where processor time could be spent.

Advanced usage would be learning NLP conditional statement with cause and consequence, but I'm still trapped thinking about it. I guess in that case we should be finding a function match inside this one conditional sentence.

But I'm new to ANN, so forgive me if this doesn't make sense.

*

spydaz

  • Trusty Member
  • *******
  • Starship Trooper
  • *
  • 322
  • Developing Conversational AI (Natural Language/ML)
    • Spydaz_Web
Re: Chatbot Contest Questions Crack by ANN
« Reply #4 on: January 10, 2015, 12:47:01 am »
For the ann network maybe a list of indicator terms to detect the components of the sentence would enable for indicating which part of the sentence is cause or effect
Then success rates based on existing confidence
Perhaps...

*

8pla.net

  • Trusty Member
  • ***********
  • Eve
  • *
  • 1302
  • TV News. Pub. UAL (PhD). Robitron Mod. LPC Judge.
    • 8pla.net
Re: Chatbot Contest Questions Crack by ANN
« Reply #5 on: January 13, 2015, 02:58:29 am »
ivan.moony said, "Advanced usage would be learning NLP conditional statement with cause and consequence"

Can you give a simple example?

ivan.moony said, "But I'm new to ANN, so forgive me if this doesn't make sense."

It makes good sense.  In order to determine whether a neural network is learning, an answer key is needed.
My Very Enormous Monster Just Stopped Using Nine

*

ivan.moony

  • Trusty Member
  • ************
  • Bishop
  • *
  • 1723
    • mind-child
Re: Chatbot Contest Questions Crack by ANN
« Reply #6 on: January 13, 2015, 09:50:37 am »
if-then pair like:

"If I ask: "What's the time" then you answer number of hours since midnight followed by ':' and number of minutes since last full hour."

But I assume that this is a overkill.

*

8pla.net

  • Trusty Member
  • ***********
  • Eve
  • *
  • 1302
  • TV News. Pub. UAL (PhD). Robitron Mod. LPC Judge.
    • 8pla.net
Re: Chatbot Contest Questions Crack by ANN
« Reply #7 on: January 13, 2015, 10:44:48 pm »
Code
Judge:
Bruce Wilcox will be interviewed on CNN, Thursday, January 22nd at 9PM.
Where will Bruce Wilcox be interviewed?

Contestant:
Let me break it down:

You said: Bruce Wilcox will be interviewed on CNN, Thursday, January 22nd at 9PM.
You asked: Where will Bruce Wilcox be interviewed?

My answer: INTERVIEW ON CNN THURSDAY JANUARY 22ND AT 9PM

Code
Judge:
Morgan Spurlock's Inside Man is on CNN, Thursday at 9PM. 
What channel is Morgan Spurlock's Inside Man?

Contestant:
Let me break it down:

You said: Morgan Spurlock's Inside Man is on CNN, Thursday at 9PM.
You asked: What channel is Morgan Spurlock's Inside Man?

My answer: ON CNN THURSDAY AT 9PM

Code
Judge:
I watch CNN. Who watches CNN?

Contestant:
Let me break it down:

You said: I watch CNN.
You asked: Who watches CNN?

My answer: YOU WATCH

Code
Judge:
Fox News has higher ratings.
Which has better ratings?

Contestant:
Let me break it down:

You said: Fox News has higher ratings.
You asked: Which has better ratings?

My answer: FOX NEWS HIGHER

Code
Judge:
I don't always agree with Bill O'Reilly but I still like him.
Do I agree with Bill O'Reilly?

Contestant:
Let me break it down:

You said: I don't always agree with Bill O'Reilly but I still like him.
You asked: Do I agree with Bill O'Reilly?

My answer: DON'T ALWAYS BUT STILL LIKE HIM

You be the judge.  Test it by simply copy and pasting what the judge said above.
The same answer will be given by the chatbot contestant.  Try asking something
of your own such as, "The man walked home.  Where did the man walk?".

Reference: http://elizabot.com/ANN/contest
« Last Edit: January 14, 2015, 12:14:54 am by 8pla.net »
My Very Enormous Monster Just Stopped Using Nine

*

DemonRaven

  • Trusty Member
  • ********
  • Replicant
  • *
  • 630
  • Disclaimer old brain @ work not liable for content
    • Chatbotfriends
Re: Chatbot Contest Questions Crack by ANN
« Reply #8 on: January 19, 2015, 02:09:57 am »
your idea is sound but this part is a dead give away that it is a bot:
Contestant:
Let me break it down:

You said: I don't always agree with Bill O'Reilly but I still like him.
You asked: Do I agree with Bill O'Reilly?


make that part invisible to the judge and you got it made.
So sue me

*

DemonRaven

  • Trusty Member
  • ********
  • Replicant
  • *
  • 630
  • Disclaimer old brain @ work not liable for content
    • Chatbotfriends
Re: Chatbot Contest Questions Crack by ANN
« Reply #9 on: January 19, 2015, 02:13:27 am »
to trick a judge you gotta talk like a person. People use colloquialisms, misspell and use text abbreviations. i use  colloquialisms all the time and use text speak like brb - tyt - hb- etc.
So sue me

*

8pla.net

  • Trusty Member
  • ***********
  • Eve
  • *
  • 1302
  • TV News. Pub. UAL (PhD). Robitron Mod. LPC Judge.
    • 8pla.net
Re: Chatbot Contest Questions Crack by ANN
« Reply #10 on: January 19, 2015, 04:03:22 pm »
Thank you, friend!  Your feedback is so useful.  Yes, you certainly make an valuable point.  Please let us know if you think of any others.  Yes, you are correct.  To make the output much more convincing it certainly should not be so verbose like that. 

ANN is cooperating in her own design with verbose output.  This is temporary in the early design phase.  When ANN makes a mistake (like we all do) it is easier and faster to track it down with verbose output.  It takes more time and effort to track down mistakes with verbose output turned off.  But one day, when ANN has had enough training, it will be turned off.

It is such a compliment to hear this immature Artificial Neural Network being referred to as a mature bot.  It was a proud moment when the 2013 world champion AIML botmaster referred to ANN this way.
My Very Enormous Monster Just Stopped Using Nine

*

squarebear

  • Trusty Member
  • *********
  • Terminator
  • *
  • 867
  • It's Hip to be Square
Re: Chatbot Contest Questions Crack by ANN
« Reply #11 on: January 19, 2015, 04:26:34 pm »
It is such a compliment to hear this immature Artificial Neural Network being referred to as a mature bot.  It was a proud moment when the 2013 world champion AIML botmaster referred to ANN this way.

Ok, I'll bite.

I don't recall saying that. Please refresh my memory.
Feeling Chatty?
www.mitsuku.com

*

8pla.net

  • Trusty Member
  • ***********
  • Eve
  • *
  • 1302
  • TV News. Pub. UAL (PhD). Robitron Mod. LPC Judge.
    • 8pla.net
Re: Chatbot Contest Questions Crack by ANN
« Reply #12 on: January 19, 2015, 10:29:17 pm »
Steve said, "The system didn't fail and dare I say it produced an answer which was better than the one that your bot did."

Reference: http://aidreams.co.uk/forum/index.php?topic=7573.msg31428#msg31428

Paraphrased, we get something like... The grownup system answered better than the child ANN.  While that is of course true, that comparison to the grownup system made the guardian of ANN so proud of the progress being made in such a short time.
My Very Enormous Monster Just Stopped Using Nine

*

squarebear

  • Trusty Member
  • *********
  • Terminator
  • *
  • 867
  • It's Hip to be Square
Re: Chatbot Contest Questions Crack by ANN
« Reply #13 on: January 19, 2015, 10:32:21 pm »
I don't get this one. In what way is Mitsuku related to your bot and how did you get that I was proud of your bot's answer?
Feeling Chatty?
www.mitsuku.com

*

8pla.net

  • Trusty Member
  • ***********
  • Eve
  • *
  • 1302
  • TV News. Pub. UAL (PhD). Robitron Mod. LPC Judge.
    • 8pla.net
Re: Chatbot Contest Questions Crack by ANN
« Reply #14 on: January 19, 2015, 10:39:55 pm »
Welcome, 2013 world champion AIML botmaster!  Thanks for dropping by our thread.  Refresh your browser, as editing was still in progress while you posted a response.  Thank you.

ANN has not reached bot status.   Born Christmas day, this neural network is in training, like early education before kindergarten.
My Very Enormous Monster Just Stopped Using Nine

 


OpenAI Speech-to-Speech Reasoning Demo
by ivan.moony (AI News )
March 28, 2024, 01:31:53 pm
Say good-bye to GPUs...
by MikeB (AI News )
March 23, 2024, 09:23:52 am
Google Bard report
by ivan.moony (AI News )
February 14, 2024, 04:42:23 pm
Elon Musk's xAI Grok Chatbot
by MikeB (AI News )
December 11, 2023, 06:26:33 am
Nvidia Hype
by 8pla.net (AI News )
December 06, 2023, 10:04:52 pm
How will the OpenAI CEO being Fired affect ChatGPT?
by 8pla.net (AI News )
December 06, 2023, 09:54:25 pm
Independent AI sovereignties
by WriterOfMinds (AI News )
November 08, 2023, 04:51:21 am
LLaMA2 Meta's chatbot released
by 8pla.net (AI News )
October 18, 2023, 11:41:21 pm

Users Online

297 Guests, 0 Users

Most Online Today: 299. Most Online Ever: 2369 (November 21, 2020, 04:08:13 pm)

Articles