I am going to write a quick AI-Simulator. If it works I should be able to make it capable of demonstrating some emergence for itself even though I am starting with a system that will allow me to encode input. It will also try to interpret input. I am going to be developing my theories about the use of highly integrable conceptual cross-categorization, cross-generalization and cross-abstractions to create more intelligent AI. This is not going to be the same as old AI although I will be using a text IO. In order to do this quickly I will be abandoning my long standing data and database management code that I planned to use for an AGI project so that I can cut the program design down to make it as simple as possible. I will use a few of the classes I have already developed for the data structures but I will only start with 3 fundamental data structures. (I will probably need a few more for data management but simplicity is a key design element). I will not be using the active file data access methods that I developed and instead I will just open the file, run the code using virtual memory and save the file by menu selection when I am done. (If the program works I can add updates-only file saving as needed since I have already developed those kinds of functions.) And I intend to use the cross-categorization, generalization and abstraction systems to develop indexing for the system as well.
So the program will allow me to program the AI (which makes it a simulation) using a labelling (or annotation) of the input. I will not be using a context-free compiler or interpreter or anything like that. I expect that I will need to develop the encoding language (the AI-Simulation language) as I go but it is not going to be a highly developed formal programming language. I want it to run into some ambiguities. It will also need to be able to interpret statements using the AI as it is developed. And when it does run into interpretation problems I intend on using text to clarify any confusing interpretations or to disambiguate as needed. (One of my AI principles includes something I call generalization-levels so clarification, explanation and disambiguation are really relative not only to general context levels but also to a variety of kinds of specific contexts.) So, if it works at all it should be unlike any other AI programs that I know about. Jim Bromer ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
