This paper specifies the Intelligence Sampling Infrastructure (ISI), a protocol for encoding, licensing, blending, attributing, and compensating human cognitive contributions within artificial intelligence systems. It advances the thesis that the absence of a structured attribution and compensation layer for human intelligence in AI training and inference is the root cause of emerging legal, economic, and ethical conflicts in generative AI. ISI addresses this gap through a system combining cryptographic identity, multi-vector cognitive encoding (semantic, stylistic, reasoning, value, and signature representations), ensemble-based probabilistic attribution, and automated micropayment settlement. The proposed architecture is explicitly opt-in, enabling individuals and organizations to register cognitive modules under defined licensing terms, while maintaining revocable control over usage. Attribution is computed through an ensemble methodology integrating attention-weighted similarity, counterfactual contribution analysis, and output-space similarity, further refined by a uniqueness adjustment mechanism to mitigate attribution inflation and gaming. The protocol produces auditable attribution records via cryptographically signed certificates and establishes a contractual framework in which probabilistic attribution is treated as a legally enforceable standard. ISI introduces seven primary contributions: (1) multi-vector encoding of cognitive patterns; (2) an ensemble attribution methodology; (3) uniqueness adjustment for attribution integrity; (4) a layered access and consent model; (5) verifiable attribution certification; (6) a forward-looking economic mechanism enabling capital formation against future cognitive output; and (7) a comprehensive implementation, governance, and dispute-resolution framework. Together, these elements define a new infrastructure layer for AI systems that transforms human intelligence into a traceable, composable, and compensable asset class.
Rashon Rahming (Tue,) studied this question.