AI Ethics in the Context of Information

This paper explores the evolving landscape of AI ethics within the context of information, emphasizing the necessity of a paradigm shift in understanding information to address the ethical challenges. Drawing on the notion of guan zhao—a holistic perspective rooted in Chinese philosophy—the author criticizes three traditional frameworks for understanding information (signal-carrying, source-preexisting, and receiver-assigning) and proposes a new receptive relation paradigm. This framework integrates these approaches, emphasizing the interplay between information encoding and decoding, and highlights the limitations of current AI systems, such as large language models (LLMs), in achieving genuine human-like understanding. The paper identifies three levels of machine intelligence—Human Knowledge Level AI (HKLAI), Information Encoding Level AI (ICLAI), and Information Level AI or Artificial General Intelligence (AGI)—and examines the ethical implications at each stage. It argues that as AI progresses toward AGI, ethical considerations must evolve from rule-based systems to a more holistic, socially embedded approach, akin to human moral reasoning. A conclusion advocates for a bidirectional feedback mechanism to address AGI ethics, positioning AI not as an adversary but as a collaborative partner in advancing human civilization.