With the advent of the Fourth Industrial Revolution, Artificial Intelligence (AI) has become increasingly prevalent and is expected to replace a range of human tasks. In this context, this study seeks to identify occupations that are vulnerable to AI, and focuses on their occupational and spatial characteristics. Korean patent data within Google Patents and tasks of occupations based on ISCO-08 were collected and analyzed via dependency parsing to reveal the corresponding tasks and occupations with AI's technical characteristics. Our analysis highlights the vulnerability of tasks that have remained unchanged for a long time, while managerial professions and other occupations involving new and complex tasks based on know-how are less susceptible to the influence of AI. The study also shows significant regional disparities in labor distribution according to AI vulnerability in Korea, forming a strong spatial cluster. While previous waves of automation primarily affected manual labor involving unskilled work and low levels of education, AI is expected to replace various forms of middle and high skilled and educated human work. As such, this study recommends that local governments prepare for the polarization of the labor market by AI based on the understanding of their employment structure.