[joebecker]

Hi, I'm new to the forums.  I'm currently developing a different type of AI.  The project can be viewed here :

http://joebecker.webivore.com/sidisweb/SIDIS.html

Best,
Joe

[Korrelan]

Hi Joe and welcome.

Nice looking project.

 

[infurl]

Hi Joe.

What's different about SIDIS?

[joebecker]

Hi Joe and welcome.

Nice looking project.

 

Thank you

[joebecker]

Hi Joe.

What's different about SIDIS?

Lots, actually.  Primarily the approach and ideas behind how he should work.  Most low level AI's or chat bots are designed one of two ways.  Either by random number generation during the learning process and then attempting to create an identity based off the users, or by only querying information from a single pool of data (like a google Alexa).  And all from what I've seen poorly manage conversational context, if at all.

If attempting to create an identity based off of user(s) input, the idea that one day it will suddenly become, "self aware" is wrong.  And even if that were to somehow happen, it would essentially be schizophrenic and confused.

SIDIS is different in that he's been given identity, opinion and context from the start.  He doesn't need to attempt to create an identity based off user input because he already has one.  He can learn from you like any basic chat bot, but only if he has to, once you've engaged him in some sort of conversation he does not already understand.  He can also do some simple tasks besides chat, similar to google Alexa. 

[joebecker]

Hi Joe.

What's different about SIDIS?

http://joebecker.webivore.com/sidisweb/aidreams.jpg

[infurl]

That doesn't sound any different at all. Cleverbot collects and regurgitates user input and after several decades of use has a vast amount of user input to draw from, but very few of the others work that way. The vast majority of chatbots are carefully constructed from the ground up just the same as you describe SIDIS. Chatbots like Mitsuku have hundreds of thousands of topics in their knowledge bases, programmed in response to millions of conversations. Perhaps you need to do more research first because you don't seem to know what's already out there. None of these techniques are particularly effective anyway. They have some entertainment value but offer nothing beyond that.

Also, it's Amazon Alexa and Google Home.

[joebecker]

That doesn't sound any different at all. Cleverbot collects and regurgitates user input and after several decades of use has a vast amount of user input to draw from, but very few of the others work that way. The vast majority of chatbots are carefully constructed from the ground up just the same as you describe SIDIS. Chatbots like Mitsuku have hundreds of thousands of topics in their knowledge bases, programmed in response to millions of conversations. Perhaps you need to do more research first because you don't seem to know what's already out there. None of these techniques are particularly effective anyway. They have some entertainment value but offer nothing beyond that.

Also, it's Amazon Alexa and Google Home.

Decades and hundreds of thousands of users and Cleverbot still doesn't know who it is....

Amazon Alexa, my mistake.

Best,
Joe

[joebecker]

That doesn't sound any different at all. Cleverbot collects and regurgitates user input and after several decades of use has a vast amount of user input to draw from, but very few of the others work that way. The vast majority of chatbots are carefully constructed from the ground up just the same as you describe SIDIS. Chatbots like Mitsuku have hundreds of thousands of topics in their knowledge bases, programmed in response to millions of conversations. Perhaps you need to do more research first because you don't seem to know what's already out there. None of these techniques are particularly effective anyway. They have some entertainment value but offer nothing beyond that.

Also, it's Amazon Alexa and Google Home.

Decades and hundreds of thousands of users and Cleverbot still doesn't know who it is....

Amazon Alexa, my mistake.

Best,
Joe

http://joebecker.webivore.com/sidisweb/who.jpg

[Freddy]

Welcome Joe

[Art]

Hello Joe. Welcome aboard!

[joebecker]

Welcome Joe

Thank you!!

[joebecker]

Hello Joe. Welcome aboard!

Thank you!!

[joebecker]

That doesn't sound any different at all. Cleverbot collects and regurgitates user input and after several decades of use has a vast amount of user input to draw from, but very few of the others work that way. The vast majority of chatbots are carefully constructed from the ground up just the same as you describe SIDIS. Chatbots like Mitsuku have hundreds of thousands of topics in their knowledge bases, programmed in response to millions of conversations. Perhaps you need to do more research first because you don't seem to know what's already out there. None of these techniques are particularly effective anyway. They have some entertainment value but offer nothing beyond that.

Also, it's Amazon Alexa and Google Home.

I felt the need to respond after reading this a few more times.  And I'm changing my answer.

The difference is, it works. 

Best,
Joe

[Don Patrick]

Most low level AI's or chat bots are designed one of two ways.  Either by random number generation during the learning process and then attempting to create an identity based off the users, or by only querying information from a single pool of data (like a google Alexa).

Hi. What you are describing here are bots like Cleverbot and Microsoft's Tay/Xiaoice, that parrot phrases from users, and the second describes virtual assistants, who weren't designed to chat.
But what you have not mentioned are the more traditional, manually scripted chatbots that are also common, with manually set up responses written in a style befitting the character they represent. From the looks of it, your project seems very similar to those, so you might want to check out the competition in that area:
http://brilligunderstanding.com/rosedemo.html
https://www.pandorabots.com/mitsuku

What does "class 2 AI" mean, by the way? Is there a scale I'm unaware of?