In recent years, research and development of new inspection technologies based on AI and image analysis has been actively conducted for various infrastructure assets that are deteriorating over time. Companies have started to provide services for new infrastructure inspection technologies, and some of them have already been used in the field. In Japan, the Ministry of Land, Infrastructure, Transport and Tourism is currently promoting the introduction of new inspection technologies. In many cases, the new inspection technologies are more cost-effective compared to conventional methods, which is a key advantage of these new technologies. On the other hand, the inspection accuracy of these new inspection technologies is not 100% and is often inferior to that of conventional inspection methods. Therefore, simply substituting new infrastructure inspection technologies for conventional inspection methods may increase the risk of deterioration due to incorrect decisions on maintenance management based on inaccurate inspection results. Under these circumstances, what is the level of accuracy and cost required for the new inspection technology to be introduced into practice? In this study, we propose a methodology for the optimal introduction of new infrastructure inspection technologies into practice. Specifically, we propose a methodology for optimizing inspection and repair policies based on the inspection results of new inspection technologies using a partially observable Markov decision process (POMDP) that takes into account the uncertainty of the inspection results of the new inspection technologies. The proposed methodology will be applied to a case study that simulates real-world infrastructure asset management, and its utility will be discussed. Additionally, the case study will demonstrate that the proposed methodology can determine the minimum required performance (in terms of accuracy and cost) of new inspection technologies for practical implementation.