Define and analyze the Orthogonality Thesis which suggests that high intelligence does not imply any specific final goal. Discuss the implications of this thesis for AI safety, specifically focusing on the difficulty of aligning a superintelligent AI with human values without constraining its capabilities.