For an IS auditor to effectively evaluate an AI system's algorithm, the most critical characteristic is its ability to provide transparent justification for its decisions. This concept, often referred to as explainability or interpretability, allows the auditor to understand the reasoning behind an output. It enables the assessment of the algorithm's logic, fairness, and adherence to organizational policies and regulations. Without transparency, the algorithm operates as a "black box," making it nearly impossible for an auditor to validate its internal workings, identify potential biases, or confirm its reliability beyond simple input-output testing.

A. The AI system algorithm uses training data to inform decision output. This is a fundamental definition of how most machine learning systems function, not a specific characteristic that aids in the evaluation of the algorithm's logic.

B. The AI system provides multiple options for model training. This feature relates to the flexibility during the development phase but does not help an auditor evaluate the logic of the final, deployed algorithm in operation.

D. The AI system uses archived transaction data to provide decisions. Similar to option A, this describes a common data source for an AI system but does not provide the insight needed to evaluate how the algorithm processes that data.

1. ISACA. (2021). Auditing Artificial Intelligence. "Explainability is the ability to explain the reasoning behind an AI model’s decision in an understandable way. This is a key requirement for building trust and confidence in AI systems and is a critical component of AI governance and audit." (Page 11).

2. National Institute of Standards and Technology (NIST). (2023). Artificial Intelligence Risk Management Framework (AI RMF 1.0). NIST AI 100-1. The "Explainable and Interpretable" characteristic is a core component of trustworthy AI

stating that systems should deliver "accompanying evidence

such as a confidence measure

or an explanation of how a decision was made." (Section 3.3

Page 13). https://doi.org/10.6028/NIST.AI.100-1

3. Jobin

A.

Ienca

M.

& Vayena

E. (2019). The global landscape of AI ethics guidelines. Nature Machine Intelligence

1(9)

389–399. The principle of "Transparency and Explainability" is identified as a primary convergent theme in global AI ethics guidelines

essential for accountability and audit. (Table 1

Page 391). https://doi.org/10.1038/s42256-019-0088-2