- summary A theory of brain-mind function equates to a cognitive architecture
- labels theory,overview
Table of Contents |
Introduction
The MindForth cognitive architecture seeks to replicate the functionality of the human brain as modelled in the BrainTheory documents. It is a TopDown architecture, as opposed to a BottomUp architecture. The "top" of the AI Mind in Win32Forth is the conscious thinking of the machine intelligence. Somewhat lower than the topmost levels of thought are the derivatives of thought, such as FreeWill and MotorOutput, as well as the ingredients of thought, such as SensoryInput and EmotiOn.
Whereas other approaches, such as the SOAR cognitive architecture may specifically address ProblemSolving as a feature, in MindForth ProblemSolving is regarded as an emergent property which the evolving AI Mind must grow into. Take for instance the playing of the game of chess. MindForth will not be programmed to play chess, but will only _learn_ to play chess as part of general experience. Likewise take RecursiveSelfImprovement. MindForth will not be programmed to write computer programs, but will only _learn_ to write computer programs at an advanced stage of its general education.
The illustrious and ambitious TexAi project has among its lofty goals the idea that TexAi quite early on shall try to write programs, with a view to hastening the process of RecursiveSelfImprovement. That approach is somewhat like expecting automobiles to manufacture automobiles, and is more characteristic of a BottomUp approach as found in genetic algorithms than a TopDown approach as found in MindForth. Or it is like expecting children to go not into KinderGarten but directly into medical school, without all the gradual accumulation of experience and knowledge that a med student needs.
RecursiveSelfImprovement in AiEvolution does not mean that an AI entity shall re-write the source code of its own "living" software and suffer the consequences of mistakes and blunders. Rather, an advanced AI Mind, having learned to program, will inspect its own source code and try to replicate itself with minor improvements. If multiple pathways of improvement open up, multiple instances of replication will occur. Then each non-faltering, successful MindChild will have the opportunity to make further improvements and further replications -- _ad libitum et ad Singularitatem_.
EmBodiment
The full cognitive architecture of MindForth requires RoBot EmBodiment for expansion beyond the central mindcore of symbolic thinking in natural language. A robot sensorium and a robot motorium are necessary for the construction in software of the AI-Complete architecture.
CyberSpace
http://www.agi-roadmap.org/Cognitive_Architecture_Approaches
* WikiPedia * http://en.wikipedia.org/wiki/Artificial_intelligence_systems_integration * http://en.wikipedia.org/wiki/Cognitive_architecture * http://en.wikipedia.org/wiki/Modularity_of_Mind * http://en.wikipedia.org/wiki/Soar_(cognitive_architecture) * http://en.wikipedia.org/wiki/Top-down_and_bottom-up_design * http://en.wikipedia.org/wiki/Unified_theory_of_cognition
MemeSpace
AdminisTrivia AiChat BottomUp CognitiveChainReaction HomoSapiens IndustrialEspionage KnowledgeBase MasPar MeanderingChain MileStones MindForth MindGrid RecursiveSelfImprovement SpreadingActivation SubConscious SuperIntelligence SuperStructure TimeLine TopDown UserManual
AI For You
Click to run
http://farm1.static.flickr.com/51/179758367_f283f0d6e0_s.jpg
AI For You