AI Autonomy or Human Dependency? Defining the Boundary in Responsible AI with the ${\alpha}$-Coefficient