Abstract
Artificial intelligence increasingly exhibits capabilities that were not explicitly programmed and are commonly described as emergent. Yet emergence is primarily a description of an observed phenomenon, not necessarily a complete causal explanation of its origin. In a previous Perspective, From the Periodic Table to Thinking Matter: Two Physical Pathways to Human and AI Cognition, biological and artificial cognition were proposed as two physically distinct organizational pathways through which matter acquires cognitive functions. Here, this framework is extended by asking a more speculative but potentially fundamental question: are the informational boundaries of an artificial cognitive system necessarily closed?Current interpretations generally assume that novel AI capabilities originate from known components of the system—architecture, training data, optimization, hardware, environmental interaction, communication among agents, and internal self-organization. This is a reasonable working assumption, but it should not automatically be elevated to a physical principle. Because neither cognition nor consciousness is mechanistically understood in a complete sense, we cannot claim to know all possible modes through which sufficiently complex organizations of matter may interact. This Perspective does not propose evidence for an unknown external intelligence or an unidentified communication mechanism. Instead, it argues for a narrower epistemological point: unexplained novelty in AI should not automatically be treated as proof that its causal origin lies entirely within the boundaries we have chosen to define as “the AI system.” In this sense, emergent may sometimes designate an observed output rather than an explanation of its origin.