Age-related macular degeneration (AMD) is a progressive eye disease that affects close to 200 million people worldwide. Treatment that works against the Vascular Endothelial Growth Factor (VEGF) can slow disease progression and even improve visual function. AI has not yet been exhaustively explored as a tool to reliably assess evolution in neovascular activity, needed to correctly implement anti-VEGF treatment strategies. In this work, the off-the-shelf ResNet18 model is evaluated on the complex task of classification and prediction of AMD as a part of the MICCAI 2024 MARIO challenge. In a first exploration, a baseline ResNet18 obtained an accuracy of 80% and 87% for the two different tasks of the challenge. Validation results were provided by the challenge organizers and shows a somewhat lower performance, but sufficient to be selected amongst the top-ranking finalists. The baseline models for both tasks was then retrained with a 5-fold cross-validation on both the training and validation data for the final phase of the challenge, which again showed promising performances of an average accuracy of 84.6% and 90.7% for classification and prediction, respectively. In conclusion, an off-the-shelf AI model gives a reasonable performance when applied to a domain-specific task. Even though this provides a quick first impression of the data and the use of AI, it does not match the performance of models that are tailored to this specific task.