Foundations of Governed Intelligence Architectures

Governed intelligence architectures enable the embedding of governance controls in the design and operational behavior of intelligent systems. They balance the desire to capitalize on the benefits of artificial intelligence (AI) with the pursuit of broader governance, policy, and organizational objectives. By integrating policies from governance, enterprise risk management, internal control, and compliance functions, these architectures mitigate the risk of adverse impacts from intelligent systems. Deploying them presents verification, validation, and testing challenges that differ from those associated with traditional IT systems.

Governed intelligence architectures are complementary to traditional IT systems and data-intensive AI systems, such as those used for training large language models and generative image generation. They implement artificial intelligence services in a manner that embeds specific organizational policies within the design of the services and associated data systems. These services include the ability to engage with other intelligent systems, including those that are external to the organization, in a structured manner where both input and output follow a set of defined parameters.