In recent years, artificial intelligence (AI) has achieved significant advances in the management of diabetic retinopathy (DR). For screening, AI-based systems demonstrate a sensitivity of 95%-97% in detecting DR. In clinical assistance, deep learning models using preoperative fundus images can accurately predict visual acuity recovery at four weeks after surgery, while AI-assisted targeted photocoagulation has been shown to reduce risks of postoperative hemorrhage and corneal edema. In drug development, AI algorithms enable efficient screening of potential therapeutics and allow quantitative, objective assessment of treatment efficacy. However, clinical adoption still faces challenges such as limited model generalizability, insufficient interpretability, data privacy concerns, and the lack of robust regulatory frameworks. Moving forward, efforts should focus on deeper integration of data, algorithms, and clinical workflows, along with establishing standardized evaluation and regulatory systems, to foster reliable and sustainable implementation of AI in DR management.