Artificial intelligence is rapidly evolving the biology and medical fields and is used in various different technologies. However, its application in rare genetic kidney diseases is challenging due to several different reasons. One of the key challenges is the limited number of cases for model training as there are more than 150 rare conditions tied to over 625 different genes. Other challenges also include data quality issues, addressing ethical issues, and building trust between healthcare professionals and AI. The paper greatly emphasizes the importance of collecting and preparing vast amounts of data, through resources such as networks and registries. Moreover, large language models, such as ChatGPT, have sometimes outperformed traditional models with minimal data, showing potential in identifying rare diseases, despite still needing critical evaluation by professionals. AI can also help shorten the time for a correct diagnosis for patients by using patient-centered diagnosis support and disease-centric screening. The development of digital twins is another way to help with individualized care for the patient as well as giving precise treatment. Broader issues such as regulations, ethical/equity dilemmas, and trust make it difficult to integrate AI in clinical practice for rare diseases. A collaborative effort between clinicians, researchers, and developers is needed in order to create effective AI resources that allow physicians and medicine to navigate these rare conditions.