Use Tried and True Techniques for Auditing AI

As assurance professionals, auditors must understand the risk associated with AI/machine learning and know how to identify, assess and address it. The most effective way auditors can discover what they need to know is the same technique they use in all audits: asking questions. For example, some good basic questions auditors can ask AI system developers include: Does a backup system exist in case of performance degradation of the AI solution? How do you know if the AI system is working correctly? What mechanisms have been tested? Is a control in place and, if so, is it performing adequately? Does a clear and consistent change management process exist?

If questions like these generate only blank looks from the developers, there may be cause for concern.

ISACA has a number of articles and resources to help auditors prepare for the new wave of AI and machine learning applications. These include: