A new interdisciplinary framework reveals how machines and humans generalize knowledge in fundamentally different ways, a breakthrough that could shape the future of sustainable, human-aligned AI.