【1】 The essence of the AI production era is still supply side reform, which aims to reduce costs and increase efficiency through generalization ability
【2】 There is no iron throne in the world of computing power, and there is a division of labor between infrastructure and marketization
【3】 AI computing power operators make computing power truly “useful”
【4】 Make the big model more powerful, communicate more (prompt word engineering), read more (RAG), practice more (model fine-tuning)
【5】 Synthetic data ≠ high-quality data, and the model’s “self-improvement ability” is the future focus
【6】 Open source does not mean free, and closed source does not mean making money. Behind it is an economic account on the supply side
【7】 Going to the cloud does not mean being cheap, and going to the cloud does not mean being secure. It is an economic account for the demand side
【8】 AI+Enterprise Management, Starting from Building Super Intelligent Management Assistant
【9】 The core feature of AI native is end-to-end, where AI continuously approaches the “shortest path”
【10】 Changes in AI native and Internet underlying logic: the future is generated
