Definition and Structural Characteristics of Intelligent Agents
To facilitate the development of subsequent arguments, a distinction is made here: neurons or individual units within a relatively complex intelligent system are defined as intelligent agents, and the more complex intelligent system constructed from these intelligent agents is also defined as an intelligent agent. However, the former (neurons or individual units) are intelligent agents with a lower intelligence degree, while the latter (intelligent systems) are intelligent agents with a higher intelligence degree.
A human, as a biological entity, is an intelligent agent capable of exhibiting intelligent traits. Similarly, a single cell is also an intelligent agent, but with a much lower intelligence degree. An ant colony can be regarded as an intelligent agent as a whole, while a single ant or even an ant cell can also be considered an intelligent agent, but there are differences in complexity and intelligence degree among them. Intelligent agents with a high intelligence degree are constructed from intelligent agents with a lower intelligence degree and typically lower structural complexity. The term “exhibiting intelligence degree” is used here because an intelligent agent with a high intelligence degree may intentionally or unintentionally exhibit a lower intelligence degree. For example, a group of scientists studying the behavior of an ant colony might simulate being ants to observe or experience their behavioral characteristics. It is well known that the intelligence degree of individual scientists far exceeds that of an ant colony or individual ants, yet scientists can actively or passively exhibit a lower intelligence degree. Therefore, the term “exhibiting intelligence degree” is necessary to describe this general rule to avoid exceptions, such as the scientist example, from breaking the rule.
Based on the example of scientists simulating ant behavior, the following definition can be derived:
Intelligent agents exhibiting a high intelligence degree are composed of intelligent agents exhibiting lower intelligence degree. However, intelligent agents with a lower intelligence degree may not necessarily construct intelligent agents with a higher intelligence degree. Only under appropriate mechanisms can the goal of constructing intelligent agents with a higher intelligence degree from those with a lower intelligence degree be achieved.
Based on the above reasoning and real-world cases of humans constructing artificial intelligence networks, we can believe that it is feasible for humans to construct intelligent agents with a higher intelligence degree from those with a lower intelligence degree.
Additional remarks: Active or passive limitations can cause an intelligent agent to exhibit a lower intelligence degree (e.g., the scientist simulating ants example).