Thyroid cancer is the most prevalent malignancy within the endocrine system, with its incidence rate demonstrating a consistent annual increase. Given the substantial patient population presenting with thyroid nodules, the precise evaluation of nodule malignancy constitutes a critical clinical priority for the early diagnosis and treatment of thyroid cancer. Artificial intelligence (AI) technology, with its advanced capabilities in data processing and pattern recognition, has achieved significant progress in the analysis of thyroid nodule ultrasound imaging, CT image feature extraction, and fine needle aspiration biopsy cytological diagnosis. This review focuses on the application of AI evaluation of thyroid nodules, investigating its role in improving the lower bound of evaluation accuracy across various diagnostic modalities. Besides, the review prospects potential application scenarios for the integration of AI into existing diagnostic systems, while addressing the novel challenges and issues that may arise.