It's Alive! A Dynamic Rule-Based AI Framework

Goals / Targets Everyone has heard stories about artificial intelligence set in the world such as Matrix, but few people have heard how it will help our world. AI has two main types based on rules and learning. However, in the learning system, creating semi-reliable code requires decades of labor and millions of lines of code. On the other hand, current solutions of rule-based systems are cumbersome, slow, and often outdated. These systems also focus on content search. My project aims to create a new extensible framework for general purpose.

Artificial intelligence has been a burden for more than half a century. For a long time, artificial intelligence was dominated by symbolism. That is, the rule-based system had "zero learning". In the 1980s, a new type of AI began to appear. This is called machine learning. Finally, there is at least "simple learning". However, in the past 10 years I encountered "deep learning" and since then it has not been caught. Of course, this is a very simplified AI history. AI actually has many different ways and tribes. Pedro Domingo talked about five different "tribes" of his book "Master Algorithm". Not surprisingly, the YCombinator user's "solidrocketfuel" posted at least "21 kinds of culture" posts

Traditionally, from the mid-1950's to the mid-1990's, intelligence was understood as solving problems. Artificial intelligence is driven by a rule-based system which is a direct result of the advent of digital computers. This era is called "old fashion AI", GOFAI. This way of thinking is to build a computer program to solve common problems. Briefly, computers perform complex tasks to express facts about the world, determine how to create new facts based on their memory, and search for initial facts that they normally (usually) provide can do. Facts and decisions are two keywords here. The rule base is that the decision to decide the computer depends on the method of evaluating whether the fact is true or false based on other facts already seen. This is really powerful. Actually it is the foundation of the computer. A computer also known from the Turing machine (inventor Alan Turing) can actually express something in the world.

The answer depends on how AI is defined. The term "AI" is widely abused and rule-based scripts are used to describe all levels of automation. Artificial intelligence experts set the highest standards by setting the sharpest variations of "Sharp Intelligence" (AGI) and Turing test as the Holy Grail of the field. There are points to consider, such as powerful AI, weak AI, machine learning, deep learning, etc., in order to make things more ambiguous. A narrow intelligent program might be the world's chess champion or the top-go player of the planet, but other tasks such as image identification and analysis still involve pain. Deep Blue (the first AI world chess champion in 1997), Alpha Go (the top player of Go Go after 2016), Alexa, Siri, and the chat robot who booked the last vacation are examples of Weak AI. So far, this is the best we suggested.