The article reviews the application of deep learning (DL) in ophthalmology, highlighting its potential to revolutionize the diagnosis and management of various eye diseases. DL has been widely adopted in image recognition, speech recognition, and natural language processing, and is now being integrated into healthcare, particularly in ophthalmology. In ophthalmology, DL has shown robust performance in detecting diabetic retinopathy (DR), retinopathy of prematurity (ROP), glaucoma-like disc, macular edema, and age-related macular degeneration (AMD) from fundus photographs and optical coherence tomography (OCT). These advancements could potentially enhance screening, diagnosis, and monitoring of major eye diseases in primary care and community settings, complementing telemedicine. However, the article also discusses several challenges associated with the clinical deployment of DL in ophthalmology, including clinical and technical issues, explainability of algorithm results, medicolegal concerns, and the acceptance of "black-box" algorithms by physicians and patients. Despite these challenges, the authors emphasize the potential of DL to transform ophthalmology in the future, and call for further research to evaluate its clinical deployment and cost-effectiveness. The review concludes by highlighting the need for ongoing efforts to address the "black-box" nature of DL systems and improve their interpretability to enhance clinical acceptance.The article reviews the application of deep learning (DL) in ophthalmology, highlighting its potential to revolutionize the diagnosis and management of various eye diseases. DL has been widely adopted in image recognition, speech recognition, and natural language processing, and is now being integrated into healthcare, particularly in ophthalmology. In ophthalmology, DL has shown robust performance in detecting diabetic retinopathy (DR), retinopathy of prematurity (ROP), glaucoma-like disc, macular edema, and age-related macular degeneration (AMD) from fundus photographs and optical coherence tomography (OCT). These advancements could potentially enhance screening, diagnosis, and monitoring of major eye diseases in primary care and community settings, complementing telemedicine. However, the article also discusses several challenges associated with the clinical deployment of DL in ophthalmology, including clinical and technical issues, explainability of algorithm results, medicolegal concerns, and the acceptance of "black-box" algorithms by physicians and patients. Despite these challenges, the authors emphasize the potential of DL to transform ophthalmology in the future, and call for further research to evaluate its clinical deployment and cost-effectiveness. The review concludes by highlighting the need for ongoing efforts to address the "black-box" nature of DL systems and improve their interpretability to enhance clinical acceptance.