A Mathematical Framework for Legacy, Governance, and Decision Integrity in Enterprise AI

A new framework addresses the challenge of maintaining integrity in enterprise artificial intelligence (AI) decision systems amidst operational changes. It proposes a mathematical language to evaluate an institution's capacity to preserve sound judgment over time, focusing on the durability of beneficial, lawful, explainable, and adaptable outcomes even after original designers depart.

Why it matters

The increasing integration of AI into critical decision-making necessitates robust mechanisms to ensure long-term integrity and compliance. This framework offers a quantitative approach to assess and maintain the reliability and ethical performance of AI systems, mitigating risks associated with system evolution and personnel changes.

Key insights

• Enterprise AI decisions require ongoing lawfulness, explainability, adaptability, and accountability despite personnel turnover, model changes, regulatory shifts, and evolving organizational incentives.
• Current governance frameworks offer principles but lack a compact mathematical language for assessing sustained sound judgment in AI systems.
• The paper introduces a design-science framework for 'institutional legacy' in AI.
• The framework quantifies the durable capacity of a decision system to produce beneficial outcomes.
• A normalized Legacy Score is proposed as a key contribution for evaluating this capacity.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2609.00572