The relationship between individual personality types and academic performance during online learning remains poorly understood. This study analysed extensive sampling data from 4,340 first-year high school students during semesters of face-to-face and online learning, respectively, to investigate the influence of students' Myers-Briggs Type Indicator (MBTI) types on their academic performance in Chinese, mathematics, English, and overall score during online learning. We observed that the online learning environment conferred disadvantages to extroverts, intuitives, thinkers, and perceivers. Specifically, the dimensions of extroversion-introversion and thinking-feeling significantly affected mathematics, while sensing-intuition and judging-perception significantly affected Chinese and English, respectively. Several significant interaction effects between different MBTI dimensions on online learning performance were also found. Girls outperformed boys in online learning performance. However, no sex difference in the effect of personality type on online learning performance was found. Student personality type identification could proactively identify students who may require additional support in the online learning environment and aid in designing effective tools for online education platforms to improve their learning performance. In particular, AI teaching assistants can be integrated into online classrooms as they could help these students address challenges posed by traditional online learning by offering personalised support. These supports include interactive dialogue for extroverts to foster engagement, creative discussion for intuitives to encourage exploration, performance feedback for thinkers to help them adjust their learning strategies, and time-management assistance for perceivers to ensure they stay organised.