A common view in current machine learning research is that machine learning itself can be used to improve the quality of AI dataset annotations – particularly image captions intended for use in vision ...