通用知能システム

Mathematical tools work well on mathematical objects (such as sets and rings of numbers), and the equation model F(X)=Y, which describes a triplet (F,X,Y), is sufficiently abstractly for defining, and expressing the pursuit of any goals (Y), with respect to any world (F), by any process that varies (X). In a sense, the artificial intelligence has been created thousands of years ago, when the first automatic/mechanistic way (a set of rules as a process) to solve equation was invented.

However, it generally proves to be hard to convert everyday problems into mathematical ones, and even today, the majority of people are unable to reduce their everyday problems to mathematical ones, and apply the formula they've learned at the elementary schools: F(X)=Y.

The idea is on how to automate the conversion of sets of everyday objects into well-defined sets, and empower the user to apply the mathematical knowledge to solve their problems.

NOTE: If anything, it's the wide accessibility to personal application of this model, is what may determine, if AI is in the hands of the many or the few, if problems we care about are solved, or not.

Method

Making the equation model F(X)=Y accessible to people.

1. Let Y - represent as all possible world's topics, or lines of code, or — governmental or business programs based on goals: whenever we define a goal, and start decomposing it into ideas how to achieve it, or further, into projects based on those ideas, and so on, into tasks to be executed by people, and lines of code, to be executed by machines... we are building a parse tree for our goals, or a program. When the program results in an atomic item (such as task, or computer operation), it becomes the (X), - the actions applied to the world. Follow this model, to organize all world's programs into a single collectively managed distributed tree. (there are details of implementation, to unite all different programming languages and scopes from legislation, to project management, to computer programming, etc.).
2. Let F - represent as all world's objects, or data (data objects can be said to be instances of schemas or classes). Construct the F by creating an address space for world's objects (including data that has methods and represents real world's objects, with their addresses and possible control operations like methods) on the world's objects as instances of schemas. Implement schemas of arbitrary nesting by recursive algorithms, allowing for storing the ontological understanding of all data thus stored. To enable that, connect schemas with concepts through a 1-to-many relation, allowing to map the any existing data record with its schema, and with possible conecpts that all instances of the schema can represent.
3. Lef X - represent all processes, as actions generators , that manifest themselves through their actions decomposing their goals Y. For example, when we humans come up with actions, we are these processes, and when a solver comes up with actions, it is this process (AI). Let the distributed system implement a plug-in system, to plug in solver processes as modules that seek optimal X parsing Goals into Ideas, Plans, their Steps, etc., into parse trees, and learning to plan based on these parse trees, providing the users with shortcuts to re-institute these plans or sub-programs.

Moreover, implement the use of time as currency to fund thus defined programs collectively through exchange of credit based on evaluation of the amount of work done during execution time (integral of time with respect to works), and its utility with respect to collective goals, making time as the general currency of the world, the value of which is determined by the measurements of how much work at any level (human doing work, processor doing work, etc.) was done during that time, using created assets as proofs of work.

Summary

Idea is to combine universal (Turing-complete) way to define and decompose goals into programs, with a universal way to collect and arrange and align arbitrarily nested data (with schemas and ontologies to understand them based on arbitrary dictionaries of concepts), and a way to automate the generation and funding of programs to achieve arbitrary personal or collective goals.

Progress

Together with a couple of friends, have created a prototype this system, that implements the above model with some of the latest web technologies, creating the ability for people to deploy, and use it both individually, collectively in a distributed or centralized fashion (universality).

We expect that this would define a pattern of defining, collecting, exchanging and merging of arbitrary objects**, and can serve as a universal intelligence, and operating system, allowing people and systems to operate together, simultaneously pursuing personal and collective goals, preserving everyone's privacy, and empowering extremely useful sharing.

_{** By objects, we mean, any object has a set of properties (including methods) . Operating on objects means defining programs, which are themselves objects. A system that has active target objects, is an agent with its will, and so would the individual that instantiates the F(X)=Y model with active Y, or a group of entities that collectively instantiate the model be.}