Predictive coding relies on a "handshake" where lower sensory layers signal when a top-down guess is sufficiently accurate. This "close enough" mechanism prevents systems from wasting resources on perfect optimization. Applying this logic to AI alignment suggests that agents should prioritize ballpark success over exact precision to avoid brittle behavior.