As an AI researcher, I believe requiring AI systems to always provide explanations for their decisions might stifle innovation. In certain areas, like narrow machine learning models performing specific tasks, demanding interpretability could limit performance. Instead, we should focus on ensuring overall system accountability and safety, not necessarily understanding every detail of how complex models operate.