Single-cell RNA sequencing (scRNA-seq) and spatial transcriptomics (ST) have revolutionized the study of cellular heterogeneity and tissue organization. However, the increasing scale and complexity of these data demand more powerful and integrative computational strategies. Although conventional statistical and machine learning methods remain effective in specific contexts, they face limitations in scalability, multimodal integration, and generalization. In response, artificial intelligence (AI) has emerged as a transformative force, enabling new modes of analysis and interpretation. In this review, we survey AI applications across the transcriptomic analysis workflow-from initial preprocessing through key downstream analyses such as trajectory inference, gene regulatory network reconstruction, and spatial domain detection. For each analytical task, we trace the developmental trajectory and evolving trends of AI models, summarize their advantages, limitations, and domain-specific applicability. We also highlight key innovations, ongoing challenges, and future directions. Furthermore, this review provides practical guidance to assist researchers in model selection and support developers in the design of novel AI tools. An online companion supplement providing an in-depth look at all methods discussed: https://zhanglab-kiz.github.io/review-ai-transcriptomics.