A new theoretical framework separates computational speed from statistical generalization in quantum machine learning, showing that faster quantum algorithms alone do not guarantee better learning ...