Project True AI "SC 0D" - Phase 3 of 9

kei10 · « **on:** July 11, 2016, 11:35:48 am »

This is Project SC 0D that strives to shape True AI/Strong AI/AGI using simple rules by eliminating known obstacles, but I have to admit-- it is quite difficult to implement due to what I am trying to do which differs from everyone here.

The name SC expands into Secondary Companion, and 0D stands for Zero Dimension. I am sure that you wonder why is it called "Second", not "First".

Introduction of Project SC 0D:
Part 1: http://aidreams.co.uk/forum/index.php?topic=10902.msg43627#msg43627
Part 2: http://aidreams.co.uk/forum/index.php?topic=10902.msg43628#msg43628
Part 3: http://aidreams.co.uk/forum/index.php?topic=10902.msg43629#msg43629

The work is divided into seven phases.

Phase 1 - Research the SC (Done)
Phase 2 - Research on Information World (Done)
Phase 3 - Develop the 0D Interface (In-Progress)
Phase 4 - Plan the 0D System (In-Progress)
Phase 5 - Develop the 0D System (In-Progress)
Phase 6 - Genesize myself into 0D (In-Progress)
Phase 7 - Genesize the 0D (In-Progress)
Phase 8 - Genesize the SC
Phase 9 - Unknown Final Phase

Greetings, I'm Kei. working on my project. I'm here to actually get attention -- but that was a thing in the past now, I do cringe at my old silly insensible mocking replies, my sincere apology -- but at least I was being honest. I wonder how did I became like that. Either way, I've decided to post my work here.

Most images are either from years ago that no longer applicable to the project, or new ones that I'll be updating here often. This forum has no collapsible tag, though. As a recommendation, don't try to guess anything. I don't even quite remember half of these are. They're related to my much older work, mostly discarded, or were complete failures. Sensitive information that reveals more than to be seen were erased.

2016-12-24 Saturday 07.22.43 AM
New scribble for the update of the progress:

8pla.net · « **Reply #1 on:** July 12, 2016, 12:12:45 am »

For the purpose of discussion, which is less far fetched than it sounds
since there are books on it.

As an experiment, it may be possible to write a custom boot loader,
to boot an A.I. experiment in place of an operating system.

An old computer if far less risky to experiment than an expensive
newer computer.'

Depends on what your goals are really.

kei10 · « **Reply #2 on:** July 12, 2016, 03:24:01 am »

Ah, about that... I am not an engineer, or know any IT knowledge, especially computer themselves. I did not attend college for personal issue. Although I can program, and I am using C# for all my work. C++ is too hardcore, the syntax is madness. Well I should pat myself on the back since I still haven't believed that I even manage to learn C# by myself, without books of sort. Which it is why I am concerned about the performance, probably traumatized from back then when I used to use slower programming languages.

About the boot thing you mentioned-- I have no clue how boot loader works. Although I know that it's something used to boot a system. Heck, I don't even know the ridiculous maths that involved around these parts of this forum that posted by hardcore users.

8pla.net · « **Reply #3 on:** July 12, 2016, 07:11:55 am »

C#... Oh, OK. So you're on Windows .NET framework. That is enough to get started. What trouble have you had with C++? Best to start practicing C++ as simple as possible, I think. Maybe, for example g++ , the GNU C++ compiler for Windows, and then collect a few C++ source codes examples from the web to compile for practice.

Filename: practice.cpp

Code

#include <iostream>

using namespace std;

int main()
{
  cout << "Practice makes perfect.\n";
}

Compiles with:

C:\> g++ practice.cpp -o practice.exe

kei10 · « **Reply #4 on:** July 12, 2016, 10:25:03 am »

~~Hmmm, oh, I guess that actually kind of worth taking a look at it again. Since also I practiced C# the way like you demonstrated. Thanks.~~

Edit: On second thought, I'm quite comfortable with C#. So, I guess nah, I'll stick with C#. Due to productivity.

8pla.net · « **Reply #5 on:** July 13, 2016, 02:29:16 am »

Or, you can have both in your programming arsenal.
C# is suited for web development with ASP.NET.
C++ is unmanaged source code, so it's good to know.

Filename: practiceIO.cpp

Code

#include <string>
#include <iostream>

using namespace std;

int main ()
{
  string YN, A="offline", N;
  int i;

  cout << "\n\n";
  cout << "Are you online, yes or no? ";
  getline (cin, YN);

  string response[4]; 
  response[0] = "Yes";
  response[1] = "yes";
  response[2] = "Y";
  response[3] = "y";

  for(i = 0; i < 4; i++){
     if(response[i] == YN){
         A="online";
         break;
     }
  }
  cout << "You're " << A << ".\n" << "What's your name? ";
  getline (cin, N);
  cout << "So, " << N << ", you're " << A << ".\n\n";
  return 0;
}

Compiles with:

C:\> g++ practiceIO.cpp -o practiceIO.exe

Program Output:

Are you online, yes or no? no
You're offline.
What's your name? Bob
So, Bob, you're offline.

Are you online, yes or no? yes
You're online.
What's your name? April
So, April, you're online.

___________________________________________________________________________

kei10 · « **Reply #6 on:** July 13, 2016, 05:47:37 am »

Interesting. I see... :O Thank you.

Aside from that, here's an update progress of 0D phase 1.

2016-07-13 Wednesday 12.25.59 PM

Full-size image - http://i.imgur.com/LEBVF0t.png

A scribble mechanical system that is finalized on [Redacted]. Further tests are needed, my goodness... Although the system is simple and easy to manage (or probably not by just console without appropriate graphical interface) ... I need to grant the AI to be able to access, use, and modify 0D system, so a graphical interface isn't really needed here.

The only problem is the performance -- and I worry about the memory, I haven't figure out how to even store the memory from ram into disk, that can be easily retrieved later. Hard drive is slow, but if I don't store memory into hard drive, as the AI grows, it will eventually run out of memory space.

Dang nabbit. Database is slow, too.

Data · « **Reply #7 on:** July 13, 2016, 02:22:25 pm »

Quote from: kei10 on July 13, 2016, 05:47:37 am

Hard drive is slow, but if I don't store memory into hard drive, as the AI grows, it will eventually run out of memory space.

Who uses HDD these days ?

Wouldn't it be much better to use an SSD ?

kei10 · « **Reply #8 on:** July 13, 2016, 02:32:50 pm »

OH... SSD. I can't believe I have never thought of that. O__. My computer is an old one, old motherboard, old cpu, and all that-- and an old HDD of about 400 Gb. Even my 2 Tb hard drive is a HDD-- for the sake of space.

Thank you!

8pla.net · « **Reply #9 on:** July 13, 2016, 03:06:43 pm »

Have you considered, a graphical user interface may not be appropriate? To begin with, it may all but guarantee no real time operations (among other considerations).

Also, as previously contributed here, according to a documentary narrated by Patrick Stewart, the Space Shuttle computer was less powerful than a modern smart phone. So, high performance may not be mission critical with A.I.

This is particularly why I suggest C++, or, I think, better yet C Language. Operating systems are programmed in C Language, not C++. Compiled languages, usually mean higher performance than .NET virtual machines. Though C# is more than sufficient to get started with, and is far and away better suited for the web.

____________________________________________________________________________

kei10 · « **Reply #10 on:** July 13, 2016, 03:47:19 pm »

That's very informative, thank you.

Hmm, on second thought, a graphical interface for 0D system is probably crucial. I realized as I kept planning on the 0D Phase 1. Hmm...

Well, for appropriate design, I mentioned from my previous mocking thread that my AI is based on mesh system with a bit of hierarchical composition-- somewhat similar to full brain simulation-- except mine is spatial. As you probably know that human brain is made up of different parts, and those parts each functioned differently. Thus my project requires a lot of Custom Classes, or Custom Structs (In C#, classes are reference types, and structs are value types). Each classes stores different things, and even different way.

Many or most data must not be duplicated. Since I need to make sure that the memory and the processing must be as optimized as it can be, so most larger data like neurons must be referenced. Our human brain revolves through plasticity. My model of AI follows that, of course. Thus, more neural connection is formed rather than new neurons -- that sort of thing.

So, yeah, data flowing through neurons. Lots and lots of neurons... Our brain is like a recorder, constantly recording every possible events, while forgetting. I'm not entirely convinced if C++ can handle that. I'm not much experienced in these IT fields, once again, wewewew. (Although my AI doesn't really need the entire brain to just say "hi". So, it doesn't effect much even if there's a million of synapses, for a reason.)

However, my project... Requires 2 AI to work at once -- sort of. So, two brains. Wewew. I don't think anyone has ever thought of 2 AI at once before, have they? ...

I heard that C++ are prone to memory problems, like memory leaks? While C# has garbage collection -- but I am not sure. I have no idea how to work with C++ memory pointers and stuff, for sure. They've never cease to confuse me.

Edit: And I need fastest matching data-- dictionary and hash works. My AI does not involve sorting at all.

So, it's not just simply...

Code

switch (input.ToUpper())
{
    case "BLARGH": return "What?";
    case "BLORGH": return "Stop it."
    case "BLERGH": return "I said, stop it!"
    case "BLURGH": return "One last time... Stop it."
    case "BLIRGH": return "... ... ..."
    case "BLYRGH": return "GOD DAMNIT YOU LITTLE! --"
}

ivan.moony · « **Reply #11 on:** July 13, 2016, 09:39:47 pm »

I'm not a fan of c/c++. It is an old language (screwed syntax) with no runtime checks (that's why it's fast, but results are buggy) and I avoid it as much as I can. I liked Java very much, but I ended at using Javascript/HTML all the time because I want to provide results right away, without any download. But I'm not a fan of Javascript either, it is just a necessary evil to me. I've heard that Python is ok, but I never used it. I ran over a tutorial for Python and I liked what I saw. Anyway, it is awfully popular, I think third or fourth top language these days (google "programming language popularity") and I assume it's not without reason. Otherwise, I used DOS languages like Quick Basic, Pascal and Clipper back in the good old days and did some homeworks in Visual Basic 6. Visual studio was a powerful thingie, a lot of addons were implanted there in a form of ActiveX technology. I remember installing some unrelated programs that had their own ActiveX controls which I could use after in my own programming. Cool stuff, but that technology is now replaced by something new, I assume. When Windows Vista came out, I was already an HC Linux user and there goes my divorce with Microsoft. When .Net framework came out, I threw a look at C#, did something with GDI+ and concluded it's just a Java in another colorful paper. You know Microsoft, they are very good at copying things in their own way, powered by a trillion employees behind back. They can't miss.

A very nice refreshment in my programming experience was reading a Haskell tutorial. It is a "functional" language, and it offers a declarative way of programming, very different of imperative one. You can implement "quick sort" algorithm in one line there. But that technology is still in infancy state, it is a kind of mess like the one with procedural languages before the object oriented invention, even more because the matter is more complicated. But anyway, I'd bet on functional programming in the near future. They just have to structurize types and unify some different syntax areas and they'd have a very pleasant platform that could excel in programming AI.

Art · « **Reply #12 on:** July 14, 2016, 02:39:15 am »

With all due respect, Hard Drives are priced too low to pass up and at the same time, those SSD's have sort of begun to fall in price. Still, the old fashioned Hard drives are still a great bang for the buck if one doesn't have a need for speed and simply wants a decent storage for running programs or backing up videos, music of computer data (not directed toward anyone that might also have a somewhat similar name -

).

Right now a 1 TB SSD can be had for around $200 USD while a 1 TB Hard Drive is around $45 USD... So the bang for the buck for the intended purpose has to rule unless one has some deep pockets.

Just my $.02 USD.

Data · « **Reply #13 on:** July 14, 2016, 12:45:52 pm »

kei10 stated that an HDD was slow.

HDD = 60 Mbps

SSD = 500 Mbps

For speed SSD is best bang for bucks these days being roughly 10 x faster than HDD and with virtually zero latency to speak of.

Korrelan · « **Reply #14 on:** July 14, 2016, 01:45:54 pm »

As a quick cheap solution put your database onto a free ram drive. Just remember to copy it off before you shut the machine down.

https://www.raymond.cc/blog/12-ram-disk-software-benchmarked-for-fastest-read-and-write-speed/

Project True AI "SC 0D" - Phase 3 of 9

kei10

Project True AI "SC 0D" - Phase 3 of 9

8pla.net

Re: Project SC0D - 0D Phase 1

kei10

Re: Project SC0D - 0D Phase 1

8pla.net

Re: Project SC0D - 0D Phase 1

kei10

Re: Project True AI "SC0D" - 0D Phase 1

8pla.net

Re: Project SC0D - 0D Phase 1

kei10

Re: Project SC0D - 0D Phase 1

Data

Re: Project SC0D - 0D Phase 1

kei10

Re: Project SC0D - 0D Phase 1

8pla.net

Re: Project True AI "SC0D" - 0D Phase 1

kei10

Re: Project True AI "SC0D" - 0D Phase 1

ivan.moony

Re: Project True AI "SC0D" - 0D Phase 1

Art

Re: Project True AI "SC0D" - 0D Phase 1

Data

Re: Project True AI "SC0D" - 0D Phase 1

Korrelan

Re: Project True AI "SC0D" - 0D Phase 1

Recent Topics

Recent News

Users Online

Articles