Serious question: has our ability to detect deepfakes improved with our ability to create them?