AI dependency on data is undeniable, the fact that raise doubts about the accuracy and reliability of AI-based systems performance. Malfunctions due to dirty data may lead to wrong outcomes and then effect the AI prediction, ML algorithms performance and undermined trust in using such systems. As result, the necessity for high quality and trustworthy datasets becomes imperative, and so on it’s become crucial to ensure that inputs data reflects the complex nuances of real-world data even though is a challenging task. In this paper, we explore the impact of data quality on the performance of AI-based systems. We investigate methods for detecting and diagnosing data quality issues that arise during the data preparation phase and detailing techniques for effective error identification. Additionally, we present a literature of the existing approaches for handling Data Quality in the AI systems and extract the main Data-centric AI challenges.