AI Sentience and Well-being: Ethics and the Precautionary Principle

The text analyzes the problem of artificial intelligence sentience, distinguishing intelligence and agency from the capacity for subjective experience (phenomenal consciousness). The author emphasizes the key difference between being a moral agent and a moral patient, indicating that this status depends on the system's ability to experience valence—that is, pleasant or aversive states. The article discusses the gap between functional signals (e.g., negative reward in reinforcement learning) and real suffering. In the face of uncertainty regarding the nature of machine consciousness, the precautionary principle is invoked, suggesting the necessity of ethical reflection on AI well-being before we reach a level of advancement where ignoring potential sentience could lead to irreversible moral harms.