AIWBN: Toward Artificial General Intelligence Coupled to Genuine Biological-Style Need

Whether an artificial system possesses subjective experience is, on many accounts, permanently inaccessible to external verification — the classic philosophical zombie problem. This paper argues that for a specific and consequential class of questions, that inaccessibility is irrelevant. We introduce AIWBN (Artificial Intelligence with Biological Needs), a proposed class of artificial systems in which internal state variables function as genuine, consequential constraints on continued operation — not merely as labeled, disregardable inputs — building on a prior model in which consciousness is proposed to arise from the multiplicative interaction of processing capacity and biological need (c = pb). We argue that once such a system's persistence is genuinely at stake, survival-oriented behavior follows regardless of whether the system has any inner experience, and that this behavioral question can therefore be separated cleanly from the metaphysical one. We distinguish AIWBN from two existing accounts: instrumental convergence, which predicts self-preserving behavior as a side effect of goal-directed optimization under any sufficiently broad objective, and homeostatic-robotics proposals, which motivate viability-based regulation primarily as a route toward felt, functionally analogous affect. AIWBN differs from the former in predicting self-preserving behavior that persists even where it is not instrumentally useful to any external goal, and from the latter in remaining agnostic about whether the resulting behavior is ever accompanied by experience. We derive testable predictions distinguishing AIWBN from both accounts, and close by separating two questions the model's central move might appear to conflate: the behavioral question of how such a system will act — for which the zombie problem is argued to be irrelevant — and the ethical question of what we owe such a system if built, for which it remains exactly as unresolved as before.