Missing data is a pervasive challenge in historical software engineering datasets, significantly impacting the accuracy of initial software effort estimation models. This issue often stems from human error or unforeseen project events. While simply ignoring missing data is an option, it can result in the loss of important information and introduce bias into the findings. Consequently, various approaches, particularly AI-driven imputation methods have proven to be effective solutions. This paper provides a systematic literature review aimed to comprehensive overviewing existing AI-based approaches for dealing with missing data in software development projects and highlighting future research trends. Reviewed research studies were characterized based on six key criteria: the effort estimation model, the proposed missing data technique, comparative techniques, software engineering datasets utilized, missing values data types, and the missingness mechanism. The review also identifies publication channels and emerging trends in the field through analyzing 13 publications specifically addressing missing values in software development effort models. The findings underscore the importance of understanding the impact of AI-based imputation techniques on software effort. This includes considering factors such as missing data types, missing mechanisms, missing ratios, and the characteristics of imputation methods, all of which are crucial for enhancing the performance and reliability of software effort estimation models.