This editorial on Cho et al.’s study examines the use of AI to improve the findings and diagnostic accuracy of skin cancer classifications. Due to concerns surrounding the limited availability of the public data sets and the reliability of AI, his team constructed the CAN5600, a 5600 image data set from public photographs available on the internet. Cho’s team used a data extractor to form a compilation of melanoma-nevus lesion images to use as a baseline for further evaluation by 2 dermatologists. These dermatologists then narrowed and refined the collection to a comprehensive set of 2000 images, termed the CAN2000. This specific data set was then used to train the GAN model to create the artificially composed collection called GAN5000. The study focused on comparing the accuracy of performance by the classifiers when looking at real and annotated images from the CAN5600 or CAN2000 versus artificially generated images from the GAN5000. The findings revealed that the performance of the classifier trained on the synthetic images performed similar or better than that of the classifier trained on the public data set. Thus, synthetic images have significant value in advancing the classification of certain rare skin cancers. Additionally, further research concluded that this study showed the best optimal classifier performance was observed when there was a combined use of both the synthetic and the real images from CAN5600. Overall, this editorial highlights the potentiality of this research, specifically emphasizing the advantage of training AI models on both synthetic and real images to stand in for the lack of widespread data currently available in dermatologic research.