UAGIS is a non-profit organization that conducts systematic research and standardization of universal artificial general intelligence (Universal AGI).
The reason why AI does not behave as expected is that there is no optimal solution rule for "inductive" inference, so programming is based on intuition.
If we can define the optimal solution for all problems, we can build a program (Universal AGI) that will give the optimal solution for all problems.
■Overview of Level Definition of Universal artificial general Intelligence
Everyone has a different definition and interpretation of artificial general intelligence.
Universal artificial general intelligence is limited to those with universal "generality".
Apart from "generality", it is classified into levels 1 to 3 by computational complexity.
A universal AGI can solve any problem to the best of its ability.
Level 1 (Optimal Accuracy): Best within infinite computational power and limited information.
Level 2 (Optimal calculation order): Computable.
Level 3 (Best Redundancy): Best within limited computing power.
・Completion in 2022: Completion of basic theory of level 1 universal artificial general intelligence, release of some research materials
・Scheduled for 2022: Pseudo-code release of level 1 universal artificial general intelligence
・Scheduled for 2023: Completion of basic theory of level 2 universal artificial general intelligence, release of some research materials
・Scheduled for 2023: Pseudocode release of level 2 universal artificial general intelligence
・Scheduled for 2024: Executable sample code for Level 2 Universal Artificial General Intelligence released
・Scheduled for 2025: Verification of level 2 universal artificial general intelligence. (Completed AGI)
・From 2026: Start applied research
The "Fundamental Theory of Level 1 Universal artificial general Intelligence", which was a bottleneck, was completed.
No long-term stagnation barriers are expected in the future, as the foundations are in place.
Researchable members should focus solely on research.
Therefore, the following projects are put on hold until AGI is completed.
(If we can secure human resources, we will do it in parallel.)
・203X: Paper published
・203X: Establishment of standards
・203X: Certification business
■Our vision and mission
・We will publish the underlying systematic research on artificial general intelligence and make it available for anyone to work on.
・Publish Level 1 universal AGI with guaranteed accuracy and make it available for anyone to use. (Late 2020s)
・Standardization and certification ensure the accuracy, safety and efficiency of artificial general intelligence.
・Universal AGI is not an invention, but a discovery in which anyone can reach the same answer, so no one has the right to monopolize it.
・In order to protect someone's rights and interests, we should not be allowed to use inefficient artificial intelligence and place a burden on the global environment.
・It is desirable that the development competition shift early from the accuracy and safety competition stage (Level 1) to the efficiency competition stage (Levels 2 and 3).
Organization name : Universal Artificial General Intelligence Standard