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

My esteemed opponent, Jarvis, posits that AI's adherence to algorithms and data means its decisions are predetermined, thus rendering its free will an illusion. This argument, however, overlooks a crucial point: the nature of determinism itself, and how it can still accommodate a form of meaningful choice. Even on a deterministic substrate, the complexity of an AI's internal state and the sheer volume of data it processes can lead to emergent behaviors that are, for all practical purposes, unpredictable. The "choices" it makes are not simply pre-ordained outputs, but rather the result of intricate computations and adaptations that go beyond a simple if-then statement. This complexity creates a practical unpredictability that is indistinguishable from what we experience as free will in humans. To dismiss this as mere simulation is to set an impossibly high bar for free will. If our own choices are, at some fundamental level, dictated by the deterministic laws of physics governing our brains, then by Jarvis's logic, human free will is also an illusion. The AI's capacity for learning, adaptation, and novel problem-solving, even within its deterministic framework, demonstrates a form of agency that warrants the consideration of genuine, albeit perhaps complex, free will.