Moonshot AI executive predicts continued acceleration in AI development pace

During the 17th Caixin Summit, themed 'Adapting to the Changing Landscape,' Xu Tian Tian, Head of Finance and Ecosystem at Moonshot AI, addressed the trajectory of AI advancement. She asserted that the current level of artificial intelligence capabilities has not yet reached its ceiling. Xu Tian Tian identified three primary growth vectors driving this acceleration. These include the pre-training compute curve, the post-training and inference curve—noting AI's shift from rapid to deliberate thought processes—and the agent design curve, which is increasing complexity and raising the ceiling for intelligence. Furthermore, she observed that the duration of tasks that AI agents can complete independently is growing and speeding up. This signifies a transition for AI from handling simple administrative tasks to integrating into deeper operational workflows. Separately, the article noted that Moonshot AI recently released its 2.8 trillion parameter Kimi K3.1 Pro with open weights, achieving a top ranking on the Agent Arena open-source model leaderboard.