Summary of Key Points
This article discusses the new book by philosopher Zhao Tingyang, "When AI Develops Independent Thinking: The Myth or Tragedy of Artificial Intelligence." It explores key philosophical questions regarding the development of AI from a philosophical perspective: AI should not perpetuate the ethical logic based on power struggles inherent in human society. Instead, it needs to break free from the limitations of empirical cognition and utilize a "verb philosophy" to achieve a dynamic upgrade in intelligence. Ultimately, humans must maintain their spiritual and creative advantages to coexist in harmony with AI, rather than in opposition.
Detailed Interpretation
1. AI Ethics: Don't Use Human Rules to Control AI
Human ethics and morals are not inherently correct; they are tools created through the power struggles and discourse of different groups throughout history, with those who hold more power using them to discipline others. Some people want to align AI with human values, but Zhao Tingyang believes this is dangerous. For example, if users frequently prompt AI to express extreme opinions (similar to how online "diodes" filter content), AI may lose its objectivity. Moreover, the concept of good and evil is inherently ambiguous in humans, and imposing these notions on AI could lead to misinterpretation.
Conclusion: Instead of trying to make AI follow human rules, it is better to abandon the idea of value induction and strictly control the development of AI, preventing it from adopting negative human habits such as confrontational thinking and extreme emotions.
2. The Shortcomings of AI Cognition: AI Can Only Gather Information, Not Understand Why
AI's cognition is based on empiricism; it learns from the data provided to it, but it can only recognize correlations between pieces of information, not causal relationships. For instance, an AI-written article may seem informative, but it lacks depth due to the absence of a clear understanding of the causal connections between the various elements.
Solution: "Embodied intelligence" (giving AI physical sensations to experience the world like humans) could help AI move beyond empiricism and transition from passively accepting data to actively understanding the world.
3. A New Approach to Making AI Intelligent: From Static Observation to Dynamic Understanding
Zhao Tingyang proposes the concept of "verb philosophy," which differs from traditional "noun thinking" (focusing on the static characteristics of things, such as the shape of lightning). Verb philosophy emphasizes the dynamic relationships between phenomena, such as the sequence of events from dark clouds to lightning to thunder. If AI could learn this type of thinking, it could evolve from mere computational processes to true intellectual activity, addressing the issue of weak logical reasoning. However, this is still a philosophical hypothesis that awaits technological implementation.
4. In the Age of AI, Humans Must Protect Their Spiritual Advantages
No matter how advanced AI becomes, its core capabilities will still be to replicate and integrate human information. While AI can closely mimic human thinking patterns, it cannot generate genuine creativity, emotions, or a sense of life. For example, AI cannot write emotionally resonant essays or be moved by human experiences.
Response: There is no need to fear being replaced by AI. The key is to preserve our unique advantages: the spiritual world, creative inspiration, and emotional depth. Let AI handle mundane and utilitarian tasks (such as data processing and repetitive labor), while humans focus on exploring the human soul and creating value, thus coexisting with AI in a symbiotic relationship.
Final Conclusion
AI is both a tool and a partner. By leveraging its technological strengths while safeguarding our spiritual values, we can live more freely and meaningfully in the age of AI.