Objective Artificial intelligence (AI), as a strategic technology that leads the future, is boosting the high-quality development of oil and gas pipelines in modern era. Examining the research hotspots and phased evolution of cutting-edge AI technologies in the operation of oil and gas pipelines over the past 20 years is critical to delineate the key issues of AI application in this field at present and outline future research directions.

Methods Employing a "co-occurrence" analysis based on key words using the bibliometric software VOSviewer, this study scrutinized literature from China National Knowledge Infrastructure (CNKI) spanning 2000 to 2023 regarding the research and application of AI methodologies in the field of pipeline operation. Additionally, by creating tag views of high-frequency key words of the collected literature, a detailed examination of the literature was performed from four branches of AI: expert systems, fuzzy logic, neural networks, and machine learning, to reveal the application domains, hotspots, and development trends of AI techniques.

Results The study findings revealed an annually progressive rise in publications within this field, particularly post-2016. AI methodologies have found extensive application across various research domains, largely focusing on leak detection, corrosion analysis, risk evaluation, identification, prediction, and optimization. The evolution of AI-driven research has transitioned from conventional approaches like neural networks, expert systems, fuzzy logic, and wavelet analysis towards new-generation algorithms, including deep learning, transfer learning, and reinforcement learning.

Conclusion Amidst the emerging wave of advancing AI technologies, the pursuit of smart operation of oil and gas pipelines mandates further explorations, including bolstering the application of various methods to integrate vast multi-source heterogeneous data, deepening research on few-shot and zero-shot learning techniques, advocating the convergence of augmented intelligence and artificial intelligence, refining the amalgamation of causal inference and machine learning, and concentrating on life-cycle management anchored in digital twin technology. These research outcomes are instrumental in expanding the adoption of AI technologies in the operation of oil and gas pipelines, shedding light on the evolution towards smart operation across oil and gas networks.