Early diagnosis and treatment of gastrointestinal cancer are crucial for improving patient prognosis, but traditional endoscopic techniques heavily rely on operator experience, resulting in high missed diagnosis rates and significant differences between observers. In recent years, artificial intelligence ( AI) has significantly improved the accuracy and efficiency of digestive endoscopy through deep learning and computer vision technology. Although AI has demonstrated superior performance compared to humans, there are still challenges in its generalization ability, data diversity, model transparency, and ethical responsibility delineation. In the future, it is necessary to promote the standardized application of AI in digestive endoscopy through multidisciplinary collaboration, large-scale clinical validation, and technological optimization, in order to achieve sustainable development of precision medicine.