A data set containing medical information is put into a machine learning (ML) model that...

Basic Concept: AI models trained on unrepresentative data can produce systematically inaccurate results for certain population groups. This is a form of algorithmic bias where the model ' s performance varies significantly across demographic segments, creating disparate outcomes. CompTIA SecAI+ Exam Objectives cover bias as a core AI governance and risk concept.

Why A is Correct: Bias in AI occurs when a model produces systematically skewed results for certain groups due to biased training data, flawed data collection, or model design choices. In this healthcare scenario, the inability to reliably predict illnesses for specific population segments indicates the training data likely underrepresented those segments, causing the model to learn inadequate patterns for them. This is a critical bias risk with serious health equity implications.

Why B is Wrong: Consistency refers to the model producing the same output given the same input across different runs or time periods. The problem described is not about inconsistent outputs for the same input but about systematically poor performance for specific population groups.

Why C is Wrong: Transparency refers to openness about how the AI model operates, what data it uses, and how it makes decisions. The compliance officer has already assessed the system, suggesting sufficient transparency exists to identify the performance gap.

Why D is Wrong: Inclusiveness is a design principle ensuring AI systems are designed to serve all users regardless of background. While related to the outcome, the specific risk type described — differential predictive accuracy across population segments — is most precisely categorized as bias.