Synthetic gestures are necessary for automating User Interface tests in applications using gestures, which often embed techniques from Artificial Intelligence and Machine Learning for gesture recognition. However, synthetic gestures must closely mimic the variances in human gestures. Therefore, algorithms capable of generating human-like fuzzy gestures depending on the configuration of user groups are necessary. In this paper, we analyze the differences between 2D and 3D gestures and correlate them with algorithms from Generative Artificial Intelligence designed for synthesizing gestures. Our findings support researchers and developers when developing components for gesture synthesis.