Hi, Im Richard.

  • 12 Replies
  • 3605 Views
*

Richard

  • Trusty Member
  • *
  • Roomba
  • *
  • 12
  • Lua, JS, C++, Java, and LOOP
Hi, Im Richard.
« on: April 24, 2015, 07:44:30 am »
Well where do I begin. Right now I am working on a very clean and crisp Natural language Processing algorithm; although the main focus is much more than that.

I am working on a Artificial Intelligence called Cere, short for Cerebris, which is Latin for Cerebrum/brain. The main goal of this Artificial Intelligence is to reach a CQS level of 9, the equivalent intelligence and consciousness as a small child or toddler.

I am 18, and I am still in highschool; Although that is not a reflection on my resolve to complete this. As of now I have nearly 30 pages of notes on the English language completed, and about 10 pages of alternative algorithms i can build on to deliver a "all-in-one" Natural language processing program for Cere to communicate with. (Speech synthesis will eventually also be apart of the program)

If I would have to sum up my presence here In this forum, I would say I am here to get feedback on occasional updates, provide many theories on consciousness and its relation to Cere and computers, and raise intriguing debates from  world news on AI to theories of AI.

I hope this will turn out to be the right community-

Thanks for accepting me.

*

Art

  • At the end of the game, the King and Pawn go into the same box.
  • Global Moderator
  • ******************
  • Hal 4000
  • *
  • 4410
Re: Hi, Im Richard.
« Reply #1 on: April 24, 2015, 10:41:23 am »
Greeting Richard!

Nice to have you with us.

Your project sounds engaging and challenging...hopefully rewarding.

We are a community of fairly like minded people from well, all walks of life and with a variety of likes, talents, disciplines and viewpoints.
While you may find someone's view might differ from yours, don't take it personally. Respectfully agree to disagree and things will usually sort themselves out.

You should get quite a bit of feedback during your project's development and hopefully some will be usable.

Most of us here are friends or acquaintances from one time or another and we all try to be a bit self-policing but most of all, Search through the previous postings or use the Search box at the top. Post or ask questions and feel free to join in on current discussions. New perspectives are always helpful.

Lastly, enjoy yourself!

Good luck and again, Welcome!
In the world of AI, it's the thought that counts!

*

Freddy

  • Administrator
  • **********************
  • Colossus
  • *
  • 6092
  • Mostly Harmless
Re: Hi, Im Richard.
« Reply #2 on: April 24, 2015, 12:36:27 pm »
Hi Richard, welcome to Ai Dreams, I hope you enjoy yourself here and find some interesting things to talk about. It seems to me that you came to the right place. Have a hunt around as we have certainly been talking about those things you mention.

Here's one that is current : http://aidreams.co.uk/forum/index.php?topic=8023

Don't worry about digging up old topics either. We like to go back to things here if needed or there's something useful to add - maybe due to new events.

And when it comes time to testing let us know if we can help.

See you on the boards  :)

*

Richard

  • Trusty Member
  • *
  • Roomba
  • *
  • 12
  • Lua, JS, C++, Java, and LOOP
Re: Hi, Im Richard.
« Reply #3 on: April 24, 2015, 10:12:34 pm »
While you may find someone's view might differ from yours, don't take it personally. Respectfully agree to disagree and things will usually sort themselves out.

I love it when others have different view points, I mean that is one of the main purposes of a public Forum, to discuss and perhaps debate certain topics with individuals of separate views

Thank you both for the warm welcome, I have been reading some of the previous topics that have been posted and I think i will be in this forum for a long time to come.

*

8pla.net

  • Trusty Member
  • *********
  • Terminator
  • *
  • 792
    • 8pla.net
Re: Hi, Im Richard.
« Reply #4 on: May 11, 2015, 11:47:00 pm »
Hey Richard,

One thing I would suggest... And this is only a suggestion... If at all possible, make your A.I. web-based.

Out of curiosity, for the sake of conversation, were you influenced by the dwarf planet Ceres?
My Very Enormous Monster Just Stopped Using Nine

*

ivan.moony

  • Trusty Member
  • ********
  • Replicant
  • *
  • 713
  • look, a star is falling
Re: Hi, Im Richard.
« Reply #5 on: May 12, 2015, 11:18:15 am »
@ 8pla.net
I like the idea of web based applications, even if it is an AI. However it can turn up that in-browser scripting may be too slow, but it is worth of try, you get a lot more users and viewers.
Wherever you see a nice spot, plant another knowledge tree :favicon:

*

ranch vermin

  • Not much time left.
  • Starship Trooper
  • *******
  • 369
  • Not much time.
Re: Hi, Im Richard.
« Reply #6 on: May 12, 2015, 09:32:19 pm »
Web based is consumer friendly, programmer unfriendly, and goes pretty slow.

A really easy NLP idea, brain dead nailed in its boring overdone coffin, is exchangable phrases, and its easy to know, and impossible to do fully automatically, and in a scoped taxonomy. I imagine you can half automate it by handing in the description/meaning, and then automatically assigning it to the words that are different as letters, but the raw data matched in their meaning! and getting a lot of mistakes in the process.

For example,   I dont like icecreams,  automatically becomes I not dont not like icreams!  Seems silly, but the logic is the same...  unless, you want to make it not detect double negatives as a positive, then the result is slightly different.

The phrase exchange is a trippy thing,  and its not quite computable, and it depends on many things, and slang logic.  Who knows how it pops in our heads.

And saying that, a machine will never be like a toddler, and especially conscious, its just different,  not saying it isnt clever tho, or it wont become amazingly powerful in the end.  But it just will never be conscious, or feel or be alive, or even horribly... sane at all.

Another way to try and get phrase exchange, is by decyphering to another language, then exchanging between them, gathering more information in the process. :)     If you want to exchange with your eye, its crossing an undetectable boundary, but now the similarity is simpler (more raw measurement matchable) if you get there.
A bit from here, a bit from there, and bring it together and see the whole picture.

*

8pla.net

  • Trusty Member
  • *********
  • Terminator
  • *
  • 792
    • 8pla.net
Re: Hi, Im Richard.
« Reply #7 on: May 12, 2015, 10:44:58 pm »
@ 8pla.net
I like the idea of web based applications, even if it is an AI. However it can turn up that in-browser scripting may be too slow, but it is worth of try, you get a lot more users and viewers.

You are right about that, Ivan.  In my experience, stand alone A.I. is usually more powerful than web based A.I.

My Loebner Prize contest entry is stand alone A.I.
So, I do both, stand alone A.I. and web based A.I.

My Very Enormous Monster Just Stopped Using Nine

*

8pla.net

  • Trusty Member
  • *********
  • Terminator
  • *
  • 792
    • 8pla.net
Re: Hi, Im Richard.
« Reply #8 on: May 12, 2015, 11:06:26 pm »
Web based is consumer friendly, programmer unfriendly, and goes pretty slow.

PHP, Perl and ASP.NET are reasonably fast for web based A.I.

JavaScript is pretty awesome for web based A.I.
For example: Text To Speech can be done in JavaScript, and much more with HTML5.

You are correct, Ranch... Compiled C++, for example, is faster and more powerful.
However, it is not easy to reach the limits of web based computer programming.
My Very Enormous Monster Just Stopped Using Nine

*

ranch vermin

  • Not much time left.
  • Starship Trooper
  • *******
  • 369
  • Not much time.
Re: Hi, Im Richard.
« Reply #9 on: May 13, 2015, 12:06:22 pm »
I better be more descriptive than just saying that.

web based code->

a) consumer friendly ->  nothing beats not having to download anything except a web page, if you picked the right web based programming thing then theres no hassles for the user to just go there and start using straight away.

b) programmer unfriendly -> since running code with no permission is a complete nono, because hackers and destructive people will take advantage of the system, so once the code is finished, then there is always some horrible process to actually getting it on the internet where there is very little help to get it on, because the information required to do so is on a need to know basis,  and they dont think you need to know.

(deal with this page, and then thats what its like doing it web based.)
https://www.khronos.org/webgl/security/

You eventually just go stuffit... its just crapper, i might as well just make an exe, download at your peril.  I dont care if people dont use it! :)

c) slow -> i actually am going to contradict what i just said,  and web based gpu code might actually kick ass,  but since your setup is $800 worth to run the program even terribly, im not sure if putting it on the internet web based isnt any good because noone forks out that much for a video card except for you.  webgl probably is really good.  i just thought of that!!! But your average java or html5 job is cpu only, and I wasnt thinking...     i spose there is web based gpu stuff for sure!  if i ever wanted to learn something new, and get over the security issue.
A bit from here, a bit from there, and bring it together and see the whole picture.

*

korrelan

  • Trusty Member
  • ********
  • Replicant
  • *
  • 678
  • Look into my eyes! WOAH!
    • Google +
Re: Hi, Im Richard.
« Reply #10 on: May 14, 2015, 08:11:47 am »
Surely the easiest/ securest method is to use a either a small web app or .exe to create a vpn pipe between the user and your main servers.  Use the incoming IP or assigned ID to tell the users apart and encrypt data in both directions.  Very easy to write (chat app basically) and easy to keep tabs on your AGI.

I have an audio input module/ vision input module/ output module that can run on any machine anywhere; also helps with developing because you can easily off load the input processing to another machine on your network.

EDIT: The modules convert the users input back into sound/ voice etc so my AGI can listen through its own audio cortex/ microphones.  When a user types a message or question the input is reproduced on a monitor, my AGI reads the input through its own visual cortex/ video cameras. 




« Last Edit: May 14, 2015, 10:08:24 am by korrelan »
It thunk... therefore it is!

*

Don Patrick

  • Trusty Member
  • *******
  • Starship Trooper
  • *
  • 386
    • Artificial Detective
Re: Hi, Im Richard.
« Reply #11 on: May 14, 2015, 09:10:09 am »
Sounds cool, Korrelan.

My first AI version was in Javascript. I got it to parse basic English sentences before I learned that Javascript could not read or write knowledge to files, and had a low limit on the amount of variables that the program could "keep in mind" during a session. Since I needed huge arrays just to keep the program's vocabulary in, I was going to run out of resources fast.
Of course nowadays you have php and SQL databases. I don't know if they are better at accommodating language processing AI. Cleverbot for instance takes several seconds to answer.
Personal project: NLP -> learning -> knowledge -> logical inference -> A.I.

*

ranch vermin

  • Not much time left.
  • Starship Trooper
  • *******
  • 369
  • Not much time.
Re: Hi, Im Richard.
« Reply #12 on: May 14, 2015, 01:17:43 pm »
Your classer looks good Korrellan!  I dont know what you know about the real brain, I dont bother doing anything biologically correct,  winds up exactly the same anyway, IMO.

Imagine taking the class map - and using the classes to morph the document to another document. :)   Not very useful as far as the meaning of the text,  but if you were looking at the letter shapes, could be a good way to make an automatic captcha solver.

java runs out of memory real quick from my experience.  you can write little things, but i dont recommend going for a big fish in it, c in visual studio is a lot better that way, you have to worry about resources a lot less.
« Last Edit: May 14, 2015, 02:27:02 pm by ranch vermin »
A bit from here, a bit from there, and bring it together and see the whole picture.

 


Users Online

21 Guests, 1 User
Users active in past 15 minutes:
Freddy
[Administrator]

Most Online Today: 31. Most Online Ever: 208 (August 27, 2008, 09:36:30 am)

Articles