Aunt AI: Agent collaboration may be moving from "system capacity" to "model capacity"

According to Aunt AI, chain analyst, Agent collaboration is changing from "system capability" to "model capability". In this context, the division of labour and coordination between Agent may no longer require human design of complex systems, but rather the ability of models to learn on their own. If the strongest models require space-sharing, sufficient running time and computing to be able to divide themselves, share experience and adapt their organization, many of the complex multi-angels may be just a transition tool when the models are not mature. Scientific research is a direct application, and multiple Agents can simultaneously develop assumptions, carry out experiments, exchange results and continue to explore. What is truly irreplaceable may be a shared state, a real experimental environment and an objective scorer. AI works on a long-term basis, auto-scientific research, large-scale parallel operations and demonstrates the ability to collaborate across Agent. If these capabilities are closed, the way AI progresses may no longer be that individual models become smarter, but that multiple AIs work together to create the next generation of AI。