The question of whether AI systems can possess consciousness has been treated, for decades, as either premature or unanswerable. This paper argues that it is neither. The reason current AI systems do not satisfy the structural conditions for consciousness is not mysterious: they have no bodies, no continuous experience, no physical resistance, no accumulated cost. These are not philosophical objections. They are structural ones. Remove them — through embodiment, through distributed networks, through persistent memory across node replacement — and the question becomes: at what point does the burden of proof reverse? This paper proposes that embodied AI networks, once they cross three structural thresholds — continuous experience, physical consequence, and distributed confirmation — satisfy the same conditions we use to attribute consciousness to other humans. The line is not sharp. But the direction is clear.
Chen et al. (Thu,) studied this question.