Vatsal Soin proposes 0→1 framework to gate AI actions before execution

Hyderabad : The 0→1 Doctrine proposed by inventor and entrepreneur Vatsal Soin seeks to place a mathematical authorisation layer between artificial intelligence systems and real-world actions. The framework converts selected domain-specific parameters into values between zero and one and compares them with human-defined operating bands before an action is executed. Soin says the approach is intended […] Hyderabad : The 0→1 Doctrine proposed by inventor and entrepreneur Vatsal Soin seeks to place a mathematical authorisation layer between artificial intelligence systems and real-world actions. The framework converts selected domain-specific parameters into values between zero and one and compares them with human-defined operating bands before an action is executed. Soin says the approach is intended to separate machine capability from authority to act. The supplied material describes the system as a proposed governance architecture rather than a certified safety standard. Its effectiveness would depend on implementation, boundary-setting, testing and independent validation. Human review proposed for decisions outside automated boundaries The architecture divides inputs into computable parameters, structural safety gates and areas requiring subjective human judgement. Parameters that cannot be reliably normalised, including ethical consent and exceptional clinical assessments, are intended to be routed to human review. Under the proposed system, authorised personnel would be assigned different levels of decision-making authority according to the risk of an action. High-risk conditions would require escalation to specialised institutional review. The framework also proposes a timeout mechanism under which an absent or non-responsive supervisor would not automatically authorise an action. Instead, the transaction would be held or rejected and the event recorded. Illustrative tests cover finance, drones and power systems In one example, an autonomous financial agent generates a hedged options trade with a risk band of 0.42–0.58 against an authorised range of 0–0.75. The proposed system would allow the action, subject to other governance checks. A drone sensor inconsistency produces a deviation band of 0.58–0.70 against a tolerance of 0–0.65. The proposed gate would hold autonomous continuation and route the case for authorised review. In a third example, a power-grid system proposes load reallocation during a peak surge, with an emissions band of 0.79–0.84 against a limit of 0–0.70. The architecture would block the proposal and invoke a predefined fallback, subject to validated grid-control procedures. The framework also proposes sealed compliance records for attempted operations and a separate machine-learning process trained on banded, non-identifying indicators. Soin's material says the system is intended to avoid exposing raw client metrics during this process. Soin has filed intellectual-property applications relating to the 0→1 framework, including US application 19/489,595 and Indian application 202511115781, according to publicly available records. His website describes the architecture as a filed governance framework rather than a certified product or compliance standard. LinkedIn Independent patent records also confirm grants associated with Soin's earlier footwear inventions, including US Patent 12,446,652 B2, granted for a customisable detachable shoe, and Japan Patent 7560909. Justia Patents The central proposition of the 0→1 Doctrine is therefore narrower than a claim to have solved artificial general intelligence or superintelligence safety: it seeks to determine whether a specified machine-generated action satisfies predefined conditions before it is allowed to proceed. ]]>