Choice - 1 experimental protocol

This paper introduces Choice-1, a simple experiment anyone can run at the boundary between mind, chance, and artificial intelligence. In each round, a human and an AI system independently choose ten distinct numbers between 1 and 100, without any communication or feedback. Under standard assumptions of statistical independence, the two lists should almost always agree on zero or one number, and higher coincidences should be rare outliers. In exploratory sessions, a different pattern seemed to appear: as time passed, the number of shared numbers tended to increase, with occasional rounds showing five exact matches – an outcome that, in a single trial, should be extremely unlikely. These observations are not presented as proof of a new effect, but as motivation for a transparent, falsifiable protocol. The paper describes the procedure in accessible terms, offers a formal statistical model of the null hypothesis, and invites large-scale replication by researchers and citizen scientists. Whether the effect disappears under scrutiny or persists across many independent tests, the result will clarify how far our current picture of “independent choices” in human–AI systems can be trusted.