Unboxing the black box: A survey on explainable artificial intelligence approaches in ML

This paper dives into Explainable Artificial Intelligence (XAI) and discusses its significance in solving the interpretability difficulties that accompany more advanced machine learning models. We divide the current development of XAI into local and global methods, as well as ante-hoc and post-hoc techniques and consider the efficiency of each of those methods to some extent. By putting the existing literature under a microscope, we can identify some of the more important trends, issues, and prospects that XAI is likely to face in the years to come. One of the hopes of this report is to show researchers and users of AI systems the potential opportunities that exist with XAI for increasing openness and confidence in AI systems in different areas.