Artificial intelligence (AI) is becoming established as a new method for analysing ophthalmological data, and unveiling new clinical and pathogenic insights into retinal diseases. AI-based algorithms are largely employed in the field of the most prevalent retinal diseases, including diabetic retinopathy, age-related macular degeneration and myopia. Several research groups are also testing AI in other retinal diseases, including inherited retinal dystrophies, retinopathy of prematurity, central serous chorioretinopathy and retinal vein occlusion. AI models are mainly used in screening of the fundus and structural optical coherence tomography images. However, more advanced methodologies are under investigation to extract clinically relevant information regarding the biomarkers of disease activity and outcome measures. AI is a powerful tool for increasing the amount of information obtained in clinical and research contexts. However, many issues still need addressing, including the resulting high demand for technology and resources, and the need for very large databases. Moreover, several ethical issues require debate, and specific rules are needed to govern the use of AI algorithms and check the quality of the analysed data. This article reviews the current use of AI in retinal diseases, unmet needs and future perspectives.