A long time ago, I watched an anime called *Psycho-Pass*.
At the time, I only thought it discussed "surveillance," "order," and "freedom." Now I realize it truly touches upon a deeper structural problem.
In that work, the power of judgment in society is not held by any one person or intelligent agent, but is exercised by a holistic mechanism pieced together from multiple systems, rules, and evaluation modules.
It is not an "omniscient individual," but a systemic judgment that continuously operates through collaboration and feedback.
Recently, I read a paper on distributed AGI and suddenly realized:
The AGI we are discussing today may be on a similar path.
If general intelligence does not emerge from some supermodel, but rather from the division of labor, collaboration, and transactions among a large number of agents, then what truly determines its direction is no longer "who is smarter," but how these agents are allowed to collaborate, how they evaluate each other, and how they reach consensus.
This is why I increasingly feel that attempts like @virtuals_io's, which define agent collaboration rules, evaluation processes, and incentive structures, are not essentially "building applications," but rather proactively shaping the structure itself.
The biggest takeaway from this paper is that the most crucial element for the future is not intelligence itself, but the structure and rules that support intelligent collaboration.
When judgment, incentives, and execution are broken down and embedded into the system, intelligence ceases to be a single point and becomes a continuously operating mechanism.
twitter.com/yueya_eth/status/2...
