Moonshot AI executive argues AI capabilities are not yet limited, citing three growth drivers

During the 17th Caixin Summit on November 12, Xu Tiantian, Head of Financial Industry and Ecosystem at Moonshot AI, asserted that the current intelligence ceiling of artificial intelligence is far from being reached, maintaining that the industry is undergoing rapid evolution. Xu Tiantian detailed that the rapid advancement in AI is fueled by the convergence of three distinct growth curves. The first is the traditional pretraining curve, which continues to advance through improvements in computing infrastructure, energy supply, and model parameter scale. The second involves advancements in post-training and inference, where techniques like Chain of Thought have enabled qualitative leaps in complex problem decomposition. The third driver is the iteration of agent system designs and their complex control frameworks, which raises the overall collaboration limit. Furthermore, the executive highlighted a significant industry hurdle: the pace of technological iteration is outpacing the speed of institutional and workflow redesign within enterprises. This suggests that for organizations to realize the full potential of AI, they must undertake deep structural and process adaptations, rather than merely implementing surface-level tool integrations.