This page is translated by ChatGPT.
Other language: 简体中文
For a long time, people have been curious about what distinguishes certain systems from others, revealing some form of intelligence. For instance, brains, AI, and so forth, all composed of the fundamental matter that constitutes the world, yet significantly distinguished from non-intelligent systems such as a cup or a stone. Starting from these microstructures, what makes these simple entities exhibit intelligence amidst complex interactions? This process is so elusive that most refer to the emergence of intelligence as "emergence."
Firstly, we'll correct the viewpoint of intelligence emergence theory by including some simple systems such as light and slime molds in the discussion to more clearly pinpoint what distinguishes intelligence from non-intelligence. Through the commonality of intelligence, we'll discover that from the brain to artificial intelligence, from vast social organizations to simple beams of light, they all follow the same principles, exhibiting a seemingly "intelligent" aspect.
On the other hand, we'll examine the process of intelligence generation and development from a completely new perspective. Traditional scientific research follows Descartes' scientific paradigm, attempting to break down complex systems into understandable subsystems, using reduction to describe the operation of higher-order systems with low-degree-of-freedom equations, and then understanding the operating principles of complex systems layer by layer. When a system becomes complex enough to generate strong intelligence, it becomes too intricate to dissect, and research within the Cartesian paradigm can only attribute explanations to "emergence," thereby hindering further study of intelligence.
However, emergence theory also tells us that intelligence lies within complexity; a system that can be easily reduced obviously cannot be considered to possess any significant intelligence. The development of deep learning also indicates that complex systems often require a vast number of parameters to fit well; merely attempting to describe complex systems with simple models will inevitably lose some essential details. We will accept the complexity and non-reducibility of systems and study intelligent systems from a more macroscopic perspective as a whole.
To the best of my knowledge, this article represents the first attempt in the world to axiomatize intelligence. Some terms introduced in this article may coincide with those in other fields, but clear and precise definitions are provided within the text. Readers are advised to distinguish between them. The article primarily focuses on elucidating ideas in an illustrative manner. Although it lacks some logical reasoning processes, and certain parts may not be rigorously precise, the overall coherence of the theory is maintained. It is believed that readers will find it self-consistent. However, the most important aspect of this theory is its practical orientation and extensibility. For readers with a need for quantitative analysis, you are welcome to supplement the symbolic system and deductive processes on your own. You can submit additional content in the form of a pull request (PR).
-
Introduction
-
Towards to Intelligence
-
Building Intelligence
-
Media Theory
-
Epilogue and Miscellaneous Thoughts
- Epilogue