ASI 2030, a studio in the RISCON family, publishes synthetic records of an imagined path toward artificial superintelligence from late 2026 to 2030, in which a human auditor gives each judgment one of the four opinions of financial auditing. None reports a real event, company or model, and their developments toward advanced AI are the studio's scenarios, not forecasts made here. This paper reads the fifteen records on the axis the studio calls Concentration, through its shared past and both of the futures its story forks into, and asks: when a few providers' models make the judgments, what does the record set do with two judgments that agree?
Agreement is never accepted as a check by itself. From the shared past on, the studio's models cite low disagreement, similar errors, matched statistics or consistent scores to reassure, and the auditors answer that such likeness may mean a single source seen twice. After the fork, what separates the two futures is whether a second source is left. Where rivals remain, the question moves to who receives audit capacity and by which path. Where one provider is left, no one can ask whether another model would give the same answer, and reopening a judgment waits on a second provider existing. Relating this to the debates on algorithmic monoculture and on the concentration of AI's markets and infrastructure, the paper adds a third consequence of concentration: the loss of anything to compare a judgment with.