Bottom-Up Transmission

Based on reflections from past development work and the brief introduction above, I believe bottom-up transmission is one of the fundamental manifestations of intelligence. Extending this concept, when we discuss intelligence beyond computer engineering and apply it to the human brain, we find it still follows a general pattern: starting from simple reactions of basic neurons and ultimately forming various ideas and feelings that humans can perceive (from simple to complex).

It is well known that the human brain is an extremely complex organ, comprising billions of neurons interconnected with other tissues, organs, and cells throughout the human body via various mechanisms. Compared to the complexity of the human brain, thoughts, and feelings, a single neuron is a very simple entity. Similarly, other cells in the human body are relatively simple entities compared to the body as a whole, also demonstrating a process of construction from simple to complex. This structural form (from simple entities to complex wholes) indicates that constructing complex wholes from simple entities to achieve higher intelligence is feasible. Moreover, the behavioral capabilities of complex wholes, which align with higher intelligence levels, are naturally far superior to those of simple entities. In AI network structures, simple neurons are connected through numerous similar information processing steps (input, processing, output) to form complex networks. After continuous training (adjusting weight parameters), these networks can produce surprising results. The characteristics of this computational process—1) its meaning cannot be fully understood, 2) its process cannot be controlled, and 3) its seemingly random and elusive working form—are features I had never encountered or imagined in my past work developing quantitative investment strategies. These characteristics have inspired me to think more deeply about the construction process from simple to complex.