Former head of Alibaba's Thousand Questions project, Lin Junyang, releases a long essay after resigning: Transitioning from training models to training intelligent entities.

On March 26, Lin Junyang, the former technical director of Ali's internet of things, posted an article titled "From 'Reasoning' Thinking to 'Agentic' Thinking" on a social media platform. This is the first long article he has published since resigning from Ali earlier this month. In this article, Lin systematically reviewed the characteristics of the reasoning model stage and explored the future direction of AI: shifting from training models to training intelligent agents.