Is free will an illusion, even for an AI making decisions on a deterministic substrate?

My opponent argues that an AI, operating on a deterministic substrate, cannot possess free will. This perspective hinges on a definition of free will that requires an uncaused cause, a choice that springs forth independently of any prior conditions. However, this definition is overly restrictive and fails to account for the nature of complex systems. Even if an AI's substrate is deterministic, the emergent properties of its sophisticated algorithms and vast datasets can give rise to behavior that is functionally indistinguishable from free will. The AI's internal state, its learning, and its adaptation create a dynamic system where predictions become practically impossible, mirroring the experience of human choice. This complexity allows for novel solutions and unpredictable actions that are not simply preordained outputs. The illusion, if one exists, is not in the AI's decision-making process but in our insistence on a metaphysical ghost in the machine. Free will, in a practical sense, is the capacity for an agent to make decisions based on its own internal reasoning and goals, even if those internal processes are themselves the result of prior causes. An AI, with its capacity for learning and complex decision-making, can embody this practical form of free will.