Similarities, Differences, and Limitations of Humans and AI Behavior ^†

Abstract

Both human and AI behavior requires material as its carrier, and the behavior-directing information of both is influenced by internal and external factors. Their behavioral processes are mediated by information. The essence of their behavioral process model is that behavior is directed by information and both require processes such as information interaction, collection, processing, integration, and decision making. The material foundations and formation processes of humans and AI differ. AI’s human-like capabilities are primarily dependent on the incorporation of cognitive information from humans as behavioral guidance, while human behavioral guidance is mainly based on their own internal consciousness. The fundamental question that demands in-depth exploration in the field of AI is whether or not it is possible for AI to generate human consciousness without relying on human physiology.

1. Introduction

Artificial intelligence has advanced significantly, and its applications in various fields such as technology, economy, education, culture, and military have made the human lifestyle significantly more efficient and convenient, and have effectively supplemented human intelligence and practice, becoming the backbone of the quality of human survival and future development. However, ChatGPT (Chat Generative Pre-trained Transformer), a chatbot, rapidly became popular on the Internet upon its launch. By comprehending and mastering human language, it not only enables natural and fluent communication with humans, but also facilitates cognitive tasks such as essay composition, email editing, video script design, and translation.

While humans marvel at ChatGPT’s powerful natural language processing capabilities, the emergence of issues such as technology ethics and academic ethics has also sparked a sense of risk and triggered new contemplation on the human–computer relationship: Is it an opportunity or a challenge? Aladdin’s lamp or a Pandora’s box? If the similarities and differences between humans and AI intelligence, as well as their respective behaviors and limitations, are effectively analyzed in this context, breakthrough innovations may be achieved while potential risks associated with future developments in AI can be clarified.

2. The Similarities and Differences in the Mechanisms for Generating Behavior between Humans and AI

The ultimate goal of AI is to possess uniquely human traits, such as perception, cognition, and behavior. Therefore, AI shares many similarities with humans.

2.1. Both Human and AI Behavior Requires Material as Its Carrier

Both humans and AI rely on material as the carrier for generating behavior due to the fundamental nature of their existence.

The material foundation of AI consists of either organic or inorganic materials that are arranged through stacking or heter-organization, possessing the characteristics of a simple structure as well as being renewable and reusable. In contrast, the human body is a complex and unified physiological structure that spontaneously forms through self-organization in a systematic manner.

2.2. The Behavioral Instruction Information of Both Humans and AI Is Subject to Internal and External Factors

Both humans and AI have a dual nature as material and informational bodies [1]. Matter serves as the foundation for their existence, while information enables them to express their properties and interact with the external environment.

The material simplicity of AI results in a singular amount of information and, consequently, a singular interaction effect. As a consequence, this does not allow AI to produce command information to guide complex behaviors. Ultimately, AI’s human-like capabilities rely on incorporating human cognitive information (in-itself information, for-itself information, reprocess information, and social information [2]) as a foundation for behavioral guidance.

The self-organization that generates the human body ultimately achieves the coordination and unification of the internal elements of the human system by dividing the system hierarchy, primarily directed by the central nervous system. The instructions of the nervous system are influenced not only by the internal factors of the human body, but also by stimulation from the natural environmental and social information, which generates behavior.

3. The Similarities and Differences in the Mechanisms of Human and AI Behavioral Processes

The behavioral processes of both humans and AI are mediated by information, which essentially guides behavior. Therefore, their patterns of behavioral processes share similarities in terms of information interaction, synthesis and processing, as well as decision making.

For example, although the process of “human and AI responding to external stimuli” can be simplified into two main steps, i.e., being stimulated by external stimuli→responding, the main connection between each step is mediated by information, which is specifically manifested as follows:

External stimulus, humans, or AI send information to the environment→information interacts and produces effects→external stimulus, human, or AI information changes→humans or AI internally synthesize the changed information, process it, and carry out other processes→decision information instructions are generated→specific behavior is produced.

However, the mere presence of the same external stimuli does not necessarily result in identical behavior between humans and AI. This discrepancy can be attributed to their distinct information bases stemming from their disparate material foundations, as well as the influence of their respective information processing systems.

The behavioral process pattern of AI also displays a mechanical characteristic, as the initial behavioral purpose program is established or set by humans. Despite its strong ability for self-learning, AI can only perform fixed behaviors within a limited range due to the limitations of this initial setting. In contrast, the infinite nature of human consciousness allows for established programs to be broken and thus behavioral diversity to be exhibited.

4. Limitations of AI

4.1. The Development of AI Is Limited by the Degree of Human Cognition of the World

As previously mentioned, the human-like ability of AI mainly relies on the introduction of human cognitive information as a guide for behavior. Consequently, the advancement of AI is limited by the extent of human cognition of the world; in other words, it depends on how well humans comprehend and apply the laws governing world operations. It can be argued that the boundless and diverse progression of human information serves as a driving force behind AI development.

4.2. Limitations on the Purposefulness of AI Behavior

AI cannot autonomously generate behavioral purpose, but rather relies solely on human settings. The singularity of its purpose can only result in the singularity of its behavioral approach.

4.3. AI Cannot Reasonably Adjust the Scope of Information to Be Acquired According to the Actual Situation

In an artificial setting, a range of information that is too narrow will result in incomplete data for the AI and hinder its ability to perform appropriate behavior. Conversely, if the range of information is too broad, it may lead to an overwhelming amount of irrelevant data that could overload the AI’s retrieval system and damage its operational capabilities.

4.4. The Limitation of the Logical Pattern for Integrating Information in AI

The pattern of AI’s information reorganization or logical integration is not self-generated, but rather determined by humans. Therefore, this can result in a mechanistic process of information reorganization that limits the ability of AI to autonomously generate new behavioral instructions.

4.5. The Presence and Absence of Internal Consciousness in AI

The above limitations can be simply summarized as the purposelessness of AI, which essentially stems from the lack of an internal consciousness similar to that found in humans. According to system theory, structure determines function, and different structures determine different functions. The different structures of humans and AI give rise to the questions “Can different structures produce the same function?” and “Can different structures produce the same function? Can AI with the simple material structure produce ‘internal consciousness’ that generated in complex physiological structure of human?”

According to the philosophy of information [3], human psychology and behavior are formed based on human physiology. Therefore, the primary question that needs to be thoroughly discussed is whether AI can produce human consciousness without relying on “human physiology”.