Artificial intelligence (AI) offers businesses a way to save expenses and a fundamental shift in innovation tools in the digital era. Whether AI can improve corporate innovation efficiency has been hotly discussed, but there is little empirical evidence. We use text mining to construct a firm-level AI application index innovatively. We investigate how AI affects corporate innovation efficiency using panel data from 3185 listed companies between 2008 and 2020. The results show that AI application significantly improves corporate innovation efficiency. Meanwhile, to avoid endogeneity problems, we use instrumental variables and propensity score matching (PSM) to test and obtain consistent conclusions. Further, we find that intensifying external market competition and flattening internal organizational structure, which are the important economic forms of innovation resource reallocation, play a moderating effect. Furthermore, the impact of AI on corporate innovation efficiency is enormous in companies with a more extensive size and less management power. In addition, the higher the level of AI development in the industry and region where the enterprise is located, the stronger the impact of AI on corporate innovation efficiency. This paper provides micro evidence for the innovation effects of AI.