Abstract
A system and method for cross-instance prediction and auditable decision-provenance receipt generation for autonomous actors. The system comprises a local sensor suite and world model resident on an autonomous actor, an institutional prediction service maintaining a learned evidence model built from anonymous sensor submissions contributed by autonomous actors across multiple operators, and a decision receipt generator. In operation, the autonomous actor queries the institutional prediction service with its sensor observations and contextual state; the service returns an evidentiary prediction derived from the aggregated cross-instance evidence model operating on spatial and temporal axes; and the receipt generator cryptographically binds the submitted sensor inputs, the prediction, the contextual state, and the executed action into a signed, timestamped receipt. Receipt authenticity is verified through model reproducibility - re-submitting identical sensor inputs against a provenance-stamped historical model produces a matching output hash without requiring server-side transaction storage. The system addresses the problem of autonomous actors operating in isolation by providing a neutral, non-discriminating evidence stack that augments local predictions with cross-instance aggregations. Applications include autonomous vehicle operations, insurance adjudication, regulatory compliance, and post-hoc accountability.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Zavesky, Eric, "METHOD AND SYSTEM FOR CROSS-INSTANCE PREDICTION AND AUDITABLE DECISION-PROVENANCE RECEIPT GENERATION FOR AUTONOMOUS ACTORS", Technical Disclosure Commons, (October 20, 2026)
https://www.tdcommons.org/dpubs_series/12295