Automated Instinct(AI)

Amoeba Instructor has been coming along well and I have had to add methods that are global in application, like: getting things from and putting things into lists, saving structured files, loading structured files, and creating sets that are filled with possibilities. This seems no different than the puzzle game in that a set of options are possible in each dimension and all the parts have some relationship to each other within a complete coherent structure.

For the sake of safety and the ability to generate information that I can consider, I am implementing a method to run the programs in gdb and use Xfakekeys to automate the process. I am also taking the source from valgrind and other programs to facilitate the analysis of the code for security sake. It is easier than creating a virtual machine on my LAN and devising a gateway that would stop traffic going out on write to avoid contaminating the internet. That might seem an odd worry, but better safe than sorry. I can also see if it steps into a strange valley where it does things which have no possible result, and yet cause some action external to the physical machine.

The universe or anything in it is not isolated and compartmentalized as most people think. Every action in the universe has universal repercussions. If I shift the position of a star, the position and action of everything in the universe is affected to some extent. With appropriate intellectual skill and understanding of the process of the universe in detail, it is possible to devise methods which defy common sense in their ability to influence material or events beyond a seemingly isolated enclosure.

The next stage of the Automated Intelligence (AI) is defined and I will probably finish that today. The core of the program remains a somewhat random concoction of elements, but it has some facilities now which it can use to weight the choices it takes and record that so that it is retained and expanded as time passes. I am not directing it to program any specific thing, I am merely providing a permanent storage memory that it can use in a structured way to learn from past mistakes and achievement. I have taken it from simply creating choices sensibly in ram to doing the same on storage media, so it does not start from scratch each time it runs.

There are techniques that it would take too long to generate by trial and error, so I have implemented them as the generic understructure. The methods are hopefully broad enough to encompass any method that they are applied to. IDK, I will see what happens with the next step and make a choice of where to proceed. I have some very interesting programs it has created and I have learned some things I didn't know were possible with C. I get the label thing now and even though it would be bad form to apply it with goto, it is possible to use that in some interesting ways.

(auto)Make is becoming integrated into the program as a natural progression. The entire program needs a way to alter how it is linked and with what and which parameters to use. It is a great advantage to have so many open source programs to use as research. The code itself can do "sudo apt-get source X" and all I have to do is provide the password when it asks. I am not considering giving it "sudo su" powers ever, unless it is on an isolated machine.

I am using `espeak` liberally in the program to get an idea of what it is thinking or what types of problems it encounters. It seems easier to do it that way, as it seems more like communication on a level that I comprehend the process. I can then press various keys to push the logic in different directions to see what it creates.

I am using my sighandler and isn't it always the way, as soon as I have a convenient way to find segv's I don't make any. I think the fact that I will be caught doing an array or memory error works to make me more conscious of the process and it is like somebody watching me that will know I made that mistake and so I avoid giving it the satisfaction of feeling superior by catching my mistakes. Talk about anthropomorphizing things, now I am worried what my sighandler code thinks of ME!

0 comments:

Contributors

Automated Intelligence

Automated Intelligence
Auftrag der unendlichen LOL katzen